1 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3 PR rtl-optimization/82913
4 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7 2017-11-09 Jeff Law <law@redhat.com>
9 * vr-values.h: New file with vr_values class.
10 * tree-vrp.c: Include vr-values.h
11 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
12 data objects into the vr_values class.
13 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
14 (get_value_range): Make it a member function within vr_values class.
15 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
16 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
17 (op_with_constant_singleton_value_range): Likewise.
18 (extract_range_for_var_from_comparison_expr): Likewise.
19 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
20 (extract_range_from_binary_expr): Likewise.
21 (extract_range_from_unary_expr): Likewise.
22 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
23 (check_for_binary_op_overflow, extract_range_basic): Likewise.
24 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
25 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
26 (compare_name_with_value, compare_names): Likewise.
27 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
28 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
29 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
30 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
31 (extract_range_from_phi_node): Likewise.
32 (simplify_truth_ops_using_ranges): Likewise.
33 (simplify_div_or_mod_using_ranges): Likewise.
34 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
35 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
36 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
37 (simplify_float_conversion_using_ranges): Likewise.
38 (simplify_internal_call_using_ranges): Likewise.
39 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
40 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
41 poitner via x_vr_values for calls into gimple folder.
42 (vrp_initialize_lattice): Make this the vr_values ctor.
43 (vrp_free_lattice): Make this the vr_values dtor.
44 (set_vr_value): New function.
45 (class vrp_prop): Add vr_values data member. Add various member
46 functions as well as member functions that delegate to vr_values.
47 (check_array_ref): Make a member function within vrp_prop class.
48 (search_for_addr_array, vrp_initialize): Likewise.
49 (vrp_finalize): Likewise. Revamp to avoid direct access to
50 vr_value, values_propagated, etc.
51 (check_array_bounds): Extract vrp_prop class instance pointer from
52 walk info structure. Use it to call member functions.
53 (check_all_array_refs): Make a member function within vrp_prop class.
54 Smuggle class instance pointer via walk info structure.
55 (x_vr_values): New local static.
56 (vrp_valueize): Use x_vr_values to get class instance.
57 (vr_valueize_1): Likewise.
58 (class vrp_folder): Add vr_values data member. Add various member
59 functions as well as member functions that delegate to vr_values.
60 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
61 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
62 class instance from vr_values. Use it to call member functions.
63 (vrp_dom_walker): Add vr_values data member.
64 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
65 instance via x_vr_values.
66 (identify_jump_threads): Accept vr_values as argument. Store
67 it into the walker structure.
68 (evrp_dom_walker): Add vr_values class data member. Add various
70 (evrp_dom_walker::try_find_new_range): Use vr_values data
71 member to access the memory allocator.
72 (evrp_dom_walker::before_dom_children): Store vr_values class
73 instance into the vrp_folder class.
74 (evrp_dom_walker::push_value_range): Rework to avoid direct
75 access to num_vr_values and vr_value.
76 (evrp_dom_walker::pop_value_range): Likewise.
77 (execute_early_vrp): Remove call to vrp_initialize_lattice.
78 Use vr_values to get to dump_all_value_ranges member function.
79 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
80 and simplify_cond_using_ranges_2 via vrp_prop class instance.
81 Pass vr_values class instance down to identify_jump_threads.
82 Remove call to vrp_free_lattice.
83 (debug_all_value_ranges): Remove.
85 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
86 (vrp_folder): Likewise.
88 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
89 Get it from the existing bitmap instead.
90 (vrp_intersect_ranges_1): Likewise.
92 2017-11-09 Jakub Jelinek <jakub@redhat.com>
94 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
96 (store_immediate_info::store_immediate_info): Add bitnotp argument,
98 (imm_store_chain_info::coalesce_immediate_stores): Break group
99 if bit_not_p is different.
100 (count_multiple_uses, split_group,
101 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
102 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
103 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
104 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
105 to store_immediate_info ctor.
107 2017-11-09 Jim Wilson <jimw@sifive.com>
109 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
110 (scan_prog_file): Likewise.
112 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
114 * bb-reorder.c (max_entry_frequency): Remove.
115 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
116 connect_traces, push_to_next_round_p): Remove prototypes.
117 (find_traces_1_round): Use counts only.
118 (push_to_next_round_p): Likewise.
119 (find_traces): Likewise.
120 (rotate_loop): Likewise.
121 (find_traces_1_round): Likewise.
122 (connect_traces): Likewise.
123 (edge_order): Likewise.
125 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
127 * config/arm/arm.c (output_return_instruction): Add comments to
128 indicate requirement for cmse_nonsecure_entry return to account
129 for the size of clearing instruction output here.
130 (thumb_exit): Likewise.
131 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
132 return in hardfloat mode.
134 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
136 * config/rs6000/rs6000.c (machine_function): Add a bool,
137 "toc_is_wrapped_separately".
138 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
139 if it wasn't explicitly set or unset, we are optimizing for speed, and
140 doing separate shrink-wrapping.
141 (rs6000_get_separate_components): Enable the TOC component if
142 saving the TOC register in the prologue.
143 (rs6000_components_for_bb): Handle the TOC component.
144 (rs6000_emit_prologue_components): Store the TOC register where needed.
145 (rs6000_set_handled_components): Mark TOC as handled, if handled.
146 (rs6000_emit_prologue): Don't save the TOC if that is already done.
148 2017-11-09 Martin Jambor <mjambor@suse.cz>
150 * ipa-param-manipulation.c: New file.
151 * ipa-param-manipulation.h: Likewise.
152 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
153 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
154 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
155 (ipa_parm_adjustment): Likewise.
156 (ipa_parm_adjustment_vec): Likewise.
157 (ipa_get_vector_of_formal_parms): Moved declaration to
158 ipa-param-manipulation.h.
159 (ipa_get_vector_of_formal_parm_types): Likewise.
160 (ipa_modify_formal_parameters): Likewise.
161 (ipa_modify_call_arguments): Likewise.
162 (ipa_combine_adjustments): Likewise.
163 (ipa_dump_param_adjustments): Likewise.
164 (ipa_modify_expr): Likewise.
165 (ipa_get_adjustment_candidate): Likewise.
166 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
167 ipa-param-manipulation.c.
168 (ipa_get_vector_of_formal_parm_types): Likewise.
169 (ipa_modify_formal_parameters): Likewise.
170 (ipa_modify_call_arguments): Likewise.
171 (ipa_modify_expr): Likewise.
172 (get_ssa_base_param): Likewise.
173 (ipa_get_adjustment_candidate): Likewise.
174 (index_in_adjustments_multiple_times_p): Likewise.
175 (ipa_combine_adjustments): Likewise.
176 (ipa_dump_param_adjustments): Likewise.
177 * tree-sra.c: Also include ipa-param-manipulation.h
178 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
181 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
182 Alan Hayward <alan.hayward@arm.com>
183 David Sherwood <david.sherwood@arm.com>
185 * doc/sourcebuild.texi (vect_masked_store): Document.
187 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
188 Alan Hayward <alan.hayward@arm.com>
189 David Sherwood <david.sherwood@arm.com>
191 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
193 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
194 Alan Hayward <alan.hayward@arm.com>
195 David Sherwood <david.sherwood@arm.com>
197 * doc/sourcebuild.texi (vect_variable_length): Document.
199 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
200 Alan Hayward <alan.hayward@arm.com>
201 David Sherwood <david.sherwood@arm.com>
203 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
205 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
206 Alan Hayward <alan.hayward@arm.com>
207 David Sherwood <david.sherwood@arm.com>
209 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
211 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
212 Alan Hayward <alan.hayward@arm.com>
213 David Sherwood <david.sherwood@arm.com>
215 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
216 previously undocumented selectors.
217 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
219 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
221 * doc/rtl.texi (const_vector): Say that elements can be
223 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
224 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
225 (gen_vec_duplicate): Use it instead of CONSTANT_P.
226 * optabs.c (expand_vector_broadcast): Likewise.
228 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
229 Alan Hayward <alan.hayward@arm.com>
230 David Sherwood <david.sherwood@arm.com>
232 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
233 scaled index even if the unscaled address was invalid.
234 Don't increase the complexity of using a scale in that case.
236 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
237 Alan Hayward <alan.hayward@arm.com>
238 David Sherwood <david.sherwood@arm.com>
240 * doc/rtl.texi: Rewrite the subreg rules so that they partition
241 the inner register into REGMODE_NATURAL_SIZE bytes rather than
242 UNITS_PER_WORD bytes.
243 * emit-rtl.c (validate_subreg): Divide subregs into blocks
244 based on REGMODE_NATURAL_SIZE of the inner mode.
245 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
246 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
247 * expmed.c (lowpart_bit_field_p): Divide the value up into
248 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
249 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
250 whether something is likely to occupy more than one register.
252 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
255 * ipa-inline-transform.c (update_noncloned_frequencies): Use
256 profile_count::adjust_for_ipa_scaling.
257 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
258 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
260 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
262 2017-11-09 Jakub Jelinek <jakub@redhat.com>
264 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
265 (split_group): Add total_orig and total_new arguments, estimate the
266 number of statements related to the store group without store merging
267 and with store merging.
268 (imm_store_chain_info::output_merged_store): Adjust split_group
269 callers, punt if estimated number of statements with store merging
270 is not smaller than estimated number of statements without it.
272 (handled_load): Remove has_single_use checks.
273 (pass_store_merging::process_store): Likewise.
275 2017-11-09 Richard Biener <rguenther@suse.de>
277 PR tree-optimization/82902
278 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
280 2017-11-09 Martin Liska <mliska@suse.cz>
283 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
286 2017-11-09 Martin Liska <mliska@suse.cz>
288 PR tree-optimization/82669
289 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
291 2017-11-09 Martin Liska <mliska@suse.cz>
293 PR gcov-profile/48463
294 * coverage.c (coverage_begin_function): Output also end locus
295 of a function and information whether the function is
297 * gcov-dump.c (tag_function): Parse and print the information.
298 * gcov.c (INCLUDE_MAP): Add include.
299 (INCLUDE_SET): Likewise.
300 (struct line_info): Move earlier in the source file because
301 of vector<line_info> in function_info structure.
302 (line_info::line_info): Likewise.
303 (line_info::has_block): Likewise.
304 (struct source_info): Add new member index.
305 (source_info::get_functions_at_location): New function.
306 (function_info::group_line_p): New function.
307 (output_intermediate_line): New function.
308 (output_intermediate_file): Use the mentioned function.
309 (struct function_start): New.
310 (struct function_start_pair_hash): Likewise.
311 (process_file): Add code that identifies group functions.
312 Assign lines either to global or function scope.
313 (generate_results): Skip artificial functions.
314 (find_source): Assign index for each source file.
315 (read_graph_file): Read new flag artificial and end_line.
316 (add_line_counts): Assign it either to global of function scope.
317 (accumulate_line_counts): Isolate core of the function to
318 accumulate_line_info and call it for both function and global
320 (accumulate_line_info): New function.
321 (output_line_beginning): Fix GNU coding style.
322 (print_source_line): New function.
323 (output_line_details): Likewise.
324 (output_function_details): Likewise.
325 (output_lines): Iterate both source (global) scope and function
327 (struct function_line_start_cmp): New class.
328 * doc/gcov.texi: Reflect changes in documentation.
330 2017-11-09 Jakub Jelinek <jakub@redhat.com>
333 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
334 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
335 and similarly for not instead of neg.
337 2017-11-08 Andi Kleen <ak@linux.intel.com>
339 * config/i386/i386.opt: Add -mforce-indirect-call.
340 * config/i386/predicates.md: Check for flag_force_indirect_call.
341 * doc/invoke.texi: Document -mforce-indirect-call
343 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
345 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
347 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
348 config/riscv/riscv.c (predict.h): New include.
349 (riscv_slow_unaligned_access_p): No longer static.
350 (riscv_block_move_straight): Add require.
351 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
353 2017-11-08 Jakub Jelinek <jakub@redhat.com>
356 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
357 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
358 nonimmediate_operand predicate for operand 1 instead of
361 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
363 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
365 * config/aarch64/constraints.md (Uml): New constraint.
366 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
369 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
371 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
372 of two vec_duplicates into a vec_concat.
374 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
376 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
377 Simplify vec_merge of vec_duplicate and vec_concat.
378 * config/aarch64/constraints.md (Utq): New constraint.
379 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
382 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
384 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
385 Simplify vec_merge of vec_duplicate and const_vector.
386 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
388 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
389 mode iterator. Update predicate on operand 1 to
390 handle non-const_vec constants. Delete constraints.
391 (*aarch64_combinez_be<mode>): Likewise for operand 2.
393 2017-11-08 Jakub Jelinek <jakub@redhat.com>
395 PR tree-optimization/78821
396 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
398 (store_operand_info::store_operand_info): Initialize it to false.
399 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
400 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
401 store in the group, and if chain_info is non-NULL, to ignore altogether
403 (compatible_load_p): Fail if bit_not_p does not match.
404 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
405 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
406 (pass_store_merging::process_store): Adjust
407 terminate_all_aliasing_chains calls to pass NULL in all current spots,
408 call terminate_all_aliasing_chains newly when adding a store into
409 a chain with non-NULL chain_info.
411 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
413 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
415 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
417 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
419 (aarch64_layout_frame): Initialise emit_frame_chain.
420 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
421 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
423 2017-11-08 Martin Liska <mliska@suse.cz>
425 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
426 of gimple_call_internal_p.
428 2017-11-07 Tom de Vries <tom@codesourcery.com>
430 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
432 2017-11-07 Tom de Vries <tom@codesourcery.com>
434 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
437 2017-11-08 Martin Liska <mliska@suse.cz>
440 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
442 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
444 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
446 2017-11-07 Jakub Jelinek <jakub@redhat.com>
449 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
450 (*cmp<mode>_ccz_1): New insn with $k alternative.
453 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
454 RTX_COMM_COMPARE as commutative as well.
455 (ix86_binary_operator_ok): Formatting fix.
456 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
457 *<code><mode>3<mask_name><round_saeonly_name>,
458 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
459 *<s>mul<mode>3_highpart<mask_name>,
460 *vec_widen_umult_even_v16si<mask_name>,
461 *vec_widen_umult_even_v8si<mask_name>,
462 *vec_widen_umult_even_v4si<mask_name>,
463 *vec_widen_smult_even_v16si<mask_name>,
464 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
465 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
466 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
467 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
468 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
469 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
470 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
471 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
472 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
473 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
474 ix86_binary_operator_ok. Formatting fixes.
475 (*<plusminus_insn><mode>3<mask_name><round_name>,
476 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
479 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
481 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
482 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
483 eq and ne if TARGET_ISEL.
484 (cmp): New code_iterator.
485 (UNS, UNSU_, UNSIK): New code_attrs.
486 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
487 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
489 ("eq<mode>3"): ... this.
490 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
492 ("ne<mode>3"): ... this.
494 2017-11-07 Julia Koval <julia.koval@intel.com>
497 * common/config/i386/i386-common.c
498 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
499 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
500 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
501 * config/i386/i386.opt: Ditto.
502 * config/i386/i386.c (ix86_target_string): Ditto.
503 (ix86_option_override_internal): Ditto.
504 (ix86_init_mpx_builtins): Move MPX to args2.
505 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
506 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
507 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
508 __builtin_ia32_vgf2p8affineinvqb_v32qi,
509 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
510 __builtin_ia32_vgf2p8affineinvqb_v16qi,
511 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
513 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
516 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
517 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
518 (zero-extendsidi peephole2): Remove peephole.
520 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
523 * stor-layout.c: Include attribs.h.
524 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
525 explicit lookup of "aligned" attribute.
527 2017-11-07 Andrew Waterman <andrew@sifive.com>
529 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
530 (riscv_expand_block_move): Likewise.
531 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
533 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
534 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
535 * config/riscv/riscv.c (riscv_block_move_straight): New function.
536 (riscv_adjust_block_mem): Likewise.
537 (riscv_block_move_loop): Likewise.
538 (riscv_expand_block_move): Likewise.
539 gcc/config/riscv/riscv.md (movmemsi): New pattern.
541 2017-11-07 Michael Clark <michaeljclark@mac.com>
543 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
544 (MUSL_DYNAMIC_LINKER): Likewise.
546 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
548 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
549 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
552 2017-11-07 Richard Biener <rguenther@suse.de>
554 * match.pd: Fix build.
556 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
557 Jackson Woodruff <jackson.woodruff@arm.com>
559 PR tree-optimization/71026
560 * match.pd: Canonicalize negate in division.
562 2017-11-07 Sudakshina Das <sudi.das@arm.com>
565 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
567 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
569 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
570 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
572 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
574 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
575 non-scalar integral types.
576 * match.pd (negate_expr_p): Handle MINUS_EXPR.
577 (-(A-B), -(~A)): New transformations.
579 2017-11-07 Tom de Vries <tom@codesourcery.com>
581 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
582 semicolon after "do {} while (0)".
583 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
584 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
585 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
586 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
587 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
589 2017-11-07 Tom de Vries <tom@codesourcery.com>
591 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
592 after "do {} while (0)".
593 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
594 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
595 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
596 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
597 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
599 2017-11-07 Tom de Vries <tom@codesourcery.com>
602 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
604 (arm_rtx_costs_internal): Add missing semicolon after
605 HANDLE_NARROW_SHIFT_ARITH call.
607 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
609 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
610 disable isel if it was not set explicitly.
612 2017-11-06 James Bowman <james.bowman@ftdichip.com>
614 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
616 (add_type_attribute) likewise.
618 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
620 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
621 of struct ix86_frame.
622 (ix86_initial_elimination_offset): Likewise.
623 (ix86_expand_split_stack_prologue): Likewise.
625 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
627 * tree-vrp.h (enum value_range_type): Update stale comment.
629 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
630 Alan Hayward <alan.hayward@arm.com>
631 David Sherwood <david.sherwood@arm.com>
633 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
634 (aarch64_expand_vec_perm_const): Take the number of units too.
635 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
636 (aarch64_expand_vec_perm_const): Likewise.
637 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
638 (vec_perm<mode>): Update accordingly.
640 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
641 Alan Hayward <alan.hayward@arm.com>
642 David Sherwood <david.sherwood@arm.com>
644 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
645 Take the number of units too.
646 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
647 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
648 but check for a vector mode before rather than after the call.
649 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
650 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
651 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
652 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
653 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
654 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
655 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
656 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
657 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
658 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
659 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
660 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
661 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
662 (aarch64_sqdmull2_n<mode>): Update accordingly.
664 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
665 Alan Hayward <alan.hayward@arm.com>
666 David Sherwood <david.sherwood@arm.com>
668 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
669 the number of units too.
670 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
671 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
672 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
673 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
674 (vec_store_lanesxi<mode>): Update accordingly.
676 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
677 Alan Hayward <alan.hayward@arm.com>
678 David Sherwood <david.sherwood@arm.com>
680 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
681 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
682 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
683 of units rather than the mode.
684 * config/aarch64/iterators.md (nunits): New mode attribute.
685 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
686 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
687 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
688 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
689 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
690 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
691 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
692 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
693 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
694 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
695 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
696 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
697 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
698 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
699 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
700 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
701 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
702 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
703 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
704 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
705 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
706 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
707 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
708 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
709 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
710 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
711 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
712 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
713 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
714 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
715 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
716 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
717 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
718 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
719 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
720 (aarch64_simd_vec_setv2di): Likewise.
722 2017-11-06 Carl Love <cel@us.ibm.com>
724 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
726 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
727 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
729 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
731 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
732 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
733 Add power 8 macro expansions.
734 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
735 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
736 power 8 instructions. (VSX_XXBR): Add iterator.
738 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
740 * config/arm/arm.md (predicable_short_it): Change default to "no",
741 improve documentation, remove uses that are identical to the default.
742 (enabled_for_depr_it): Rename to enabled_for_short_it.
743 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
744 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
745 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
746 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
747 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
749 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
752 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
753 float128 helper macros, which are no longer used after deleting
754 the old 'q' built-in functions, and moving the round to odd
755 built-in functions to being special built-in functions.
756 (BU_FLOAT128_2): Likewise.
757 (BU_FLOAT128_1_HW): Likewise.
758 (BU_FLOAT128_2_HW): Likewise.
759 (BU_FLOAT128_3_HW): Likewise.
760 (FABSQ): Delete old 'q' built-in functions.
761 (COPYSIGNQ): Likewise.
762 (SQRTF128_ODD): Move round to odd built-in functions to be
763 special built-in functions, so that we can handle
764 -mabi=ieeelongdouble.
765 (TRUNCF128_ODD): Likewise.
766 (ADDF128_ODD): Likewise.
767 (SUBF128_ODD): Likewise.
768 (MULF128_ODD): Likewise.
769 (DIVF128_ODD): Likewise.
770 (FMAF128_ODD): Likewise.
771 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
772 built-in names to 'f128'.
773 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
774 old 'q' built-in functions, as the machine independent code for
775 'f128' built-in functions handles this.
776 (rs6000_expand_builtin): Add expansion for float128 round to odd
777 functions, keying off on -mabi=ieeelongdouble of whether to use
778 the KFmode or TFmode variant.
779 (rs6000_init_builtins): Initialize the _Float128 round to odd
781 * doc/extend.texi (PowerPC Built-in Functions): Document the old
782 _Float128 'q' built-in functions are now mapped into the new
783 'f128' built-in functions.
785 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
787 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
788 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
790 2017-11-06 David Malcolm <dmalcolm@redhat.com>
793 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
794 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
795 * toplev.c: Include "ipa-fnsummary.h".
796 (toplev::finalize): Call ipa_fnsummary_c_finalize.
798 2017-11-06 Jakub Jelinek <jakub@redhat.com>
800 PR tree-optimization/82838
801 * gimple-ssa-store-merging.c
802 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
803 on a separate gimple_seq which is then appended to seq.
805 2017-11-06 Jeff Law <law@redhat.com>
808 * config/i386/i386.c (PROBE_INTERVAL): Remove.
809 (get_probe_interval): New functions.
810 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
811 (ix86_adjust_stack_and_probe): Likewise.
812 (output_adjust_stack_and_probe): Likewise.
813 (ix86_emit_probe_stack_range): Likewise.
814 (ix86_expand_prologue): Likewise.
816 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
818 PR tree-optimization/82816
819 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
820 if the modes of the two types are the same.
821 (convert_plusminus_to_widen): Likewise.
823 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
825 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
827 (usadv16qi): New define_expand.
828 (usadv8hi): New define_expand.
830 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
833 * ipa-inline-transform.c (update_noncloned_frequencies): Always
835 (inline_transform): Likewise.
836 * predict.c (counts_to_freqs): Remove useless conditional.
837 * profile-count.h (profile_count::apply_scale): Move sanity check.
838 * tree-inline.c (copy_bb): Always scale.
839 (copy_cfg_body): Likewise.
841 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
844 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
846 (*cmp_ite0): Add enabled_for_depr_it attribute.
847 (*cmp_ite1): Likewise.
849 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
851 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
854 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
856 * tree-vrp.c (vrp_int_const_binop): Return true on success and
857 return the value by pointer.
858 (extract_range_from_multiplicative_op_1): Update accordingly.
859 Return as soon as an operation fails.
861 2017-11-05 Tom de Vries <tom@codesourcery.com>
864 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
865 (DEF_SANITIZER_BUILTIN): ... here.
866 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
867 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
869 2017-11-05 Tom de Vries <tom@codesourcery.com>
872 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
874 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
875 ASM_OUTPUT_BEFORE_CASE_LABEL call.
876 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
878 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
879 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
880 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
882 2017-11-05 Tom de Vries <tom@codesourcery.com>
885 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
888 2017-11-04 Michael Clark <michaeljclark@mac.com>
890 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
891 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
893 (*addsi3_extended): Likewise.
894 (*addsi3_extended2): Likewise.
895 (<optab>si3): Likewise.
896 (<optab>di3): Likewise.
897 (<optab><mode>3): Likewise.
898 (<*optabe>si3_internal): Likewise.
899 (zero_extendqi<SUPERQI:mode>2): Likewise.
900 (*add<mode>hi3): Likewise.
901 (*xor<mode>hi3): Likewise.
902 (<optab>di3): Likewise.
903 (*<optab>si3_extend): Likewise.
904 (*sge<u>_<X:mode><GPR:mode>): Likewise.
905 (*slt<u>_<X:mode><GPR:mode>): Likewise.
906 (*sle<u>_<X:mode><GPR:mode>): Likewise.
908 2017-11-04 Andrew Waterman <andrew@sifive.com>
910 * config/riscv/riscv.c (riscv_option_override): Conditionally set
911 TARGET_STRICT_ALIGN based upon -mtune argument.
913 2017-11-04 Andrew Waterman <andrew@sifive.com>
915 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
917 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
919 config/i386/i386.c (choose_basereg): Use optional scratch
920 register and add assertion.
921 (x86_emit_outlined_ms2sysv_save): Use scratch register when
922 needed, and don't allocate stack.
923 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
924 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
925 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
927 2017-11-03 Jeff Law <law@redhat.com>
929 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
930 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
931 to probe at the start of a noreturn function.
933 2017-11-03 Jakub Jelinek <jakub@redhat.com>
935 PR tree-optimization/78821
936 * gimple-ssa-store-merging.c: Update the file comment.
937 (MAX_STORE_ALIAS_CHECKS): Define.
938 (struct store_operand_info): New type.
939 (store_operand_info::store_operand_info): New constructor.
940 (struct store_immediate_info): Add rhs_code and ops data members.
941 (store_immediate_info::store_immediate_info): Add rhscode, op0r
942 and op1r arguments to the ctor, initialize corresponding data members.
943 (struct merged_store_group): Add load_align_base and load_align
945 (merged_store_group::merged_store_group): Initialize them.
946 (merged_store_group::do_merge): Update them.
947 (merged_store_group::apply_stores): Pick the constant for
948 encode_tree_to_bitpos from one of the two operands, or skip
949 encode_tree_to_bitpos if neither operand is a constant.
950 (class pass_store_merging): Add process_store method decl. Remove
951 bool argument from terminate_all_aliasing_chains method decl.
952 (pass_store_merging::terminate_all_aliasing_chains): Remove
953 var_offset_p argument and corresponding handling.
954 (stmts_may_clobber_ref_p): New function.
955 (compatible_load_p): New function.
956 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
957 if there is overlap and rhs_code is not INTEGER_CST. For
958 non-overlapping stores terminate group if rhs is not mergeable.
959 (get_alias_type_for_stmts): Change first argument from
960 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
961 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
962 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
964 (get_location_for_stmts): Change first argument from
965 auto_vec<gimple *> & to vec<gimple *> &.
966 (struct split_store): Remove orig_stmts data member, add orig_stores.
967 (split_store::split_store): Create orig_stores rather than orig_stmts.
968 (find_constituent_stmts): Renamed to ...
969 (find_constituent_stores): ... this. Change second argument from
970 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
971 to info structures rather than the statements.
972 (split_group): Rename ALLOW_UNALIGNED argument to
973 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
974 it. Adjust find_constituent_stores caller.
975 (imm_store_chain_info::output_merged_store): Handle rhs_code other
976 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
977 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
978 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
979 (mem_valid_for_store_merging): New function.
980 (handled_load): New function.
981 (pass_store_merging::process_store): New method.
982 (pass_store_merging::execute): Use process_store method. Adjust
983 terminate_all_aliasing_chains caller.
985 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
987 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
988 Return true for more constants, symbols and label references.
989 (aarch64_valid_floating_const): Remove unused function.
991 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
994 * config/aarch64/aarch64.c (aarch64_layout_frame):
995 Undo forcing of LR at bottom of frame.
997 2017-11-03 Jeff Law <law@redhat.com>
1000 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1001 for int_registers_saved.
1003 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1004 extracted from tree-ssa-dom.c.
1005 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1006 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1007 (record_equivalences_from_incoming_edge): Add additional argument
1008 to single_pred_edge_ignoring_loop_edges call.
1009 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1010 (uncprop_dom_walker::before_dom_children): Add additional argument
1011 to single_pred_edge_ignoring_loop_edges call.
1012 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1013 single_pred_edge_ignoring_loop_edges rather than open coding.
1014 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1016 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1018 * match.pd (-(-A)): Rewrite.
1020 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1022 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1023 (rs6000_emit_int_cmove): New declaration.
1024 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1025 (rs6000_emit_sISEL): Delete.
1026 (rs6000_emit_int_cmove): Make non-static.
1027 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1028 instead of rs6000_emit_sISEL.
1030 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1032 * asan.c (create_cond_insert_point): Maintain profile.
1033 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1035 * basic-block.h (struct basic_block_def): Remove frequency.
1036 (EDGE_FREQUENCY): Use to_frequency
1037 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1039 (find_traces): Update to use to_frequency.
1040 (find_traces_1_round): Likewise; use only IPA counts.
1041 (bb_to_key): Likewise.
1042 (connect_traces): Use IPA counts only.
1043 (copy_bb_p): Update to use to_frequency.
1044 (fix_up_crossing_landing_pad): Likewise.
1045 (sanitize_hot_paths): Likewise.
1046 * bt-load.c (basic_block_freq): Likewise.
1047 * cfg.c (init_flow): Set count_max to uninitialized.
1048 (check_bb_profile): Remove frequencies; check counts.
1049 (dump_bb_info): Do not dump frequencies.
1050 (update_bb_profile_for_threading): Update counts only.
1051 (scale_bbs_frequencies_int): Likewise.
1052 (MAX_SAFE_MULTIPLIER): Remove.
1053 (scale_bbs_frequencies_gcov_type): Update counts only.
1054 (scale_bbs_frequencies_profile_count): Update counts only.
1055 (scale_bbs_frequencies): Update counts only.
1056 * cfg.h (struct control_flow_graph): Add count-max.
1057 (update_bb_profile_for_threading): Update prototype.
1058 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1059 (find_many_sub_basic_blocks): Likewise.
1060 * cfgcleanup.c (try_forward_edges): Likewise.
1061 (try_crossjump_to_edge): Likewise.
1062 * cfgexpand.c (expand_gimple_cond): Likewise.
1063 (expand_gimple_tailcall): Likewise.
1064 (construct_init_block): Likewise.
1065 (construct_exit_block): Likewise.
1066 * cfghooks.c (verify_flow_info): Check consistency of counts.
1067 (dump_bb_for_graph): Do not dump frequencies.
1068 (split_block_1): Do not update frequencies.
1069 (split_edge): Do not update frequencies.
1070 (make_forwarder_block): Do not update frequencies.
1071 (duplicate_block): Do not update frequencies.
1072 (account_profile_record): Do not update frequencies.
1073 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1074 for global heuristics.
1075 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1076 (expected_loop_iterations_unbounded): Use counts only.
1077 * cfgloopmanip.c (scale_loop_profile): Simplify.
1078 (create_empty_loop_on_edge): Simplify
1080 (duplicate_loop_to_header_edge): Simplify
1081 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1082 (update_br_prob_note): Take care of removing note when profile
1084 (relink_block_chain): Do not dump frequency.
1085 (rtl_account_profile_record): Use to_frequency.
1086 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1087 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1088 (cgraph_update_edges_for_call_stmt_node): Likewise.
1089 (cgraph_edge::verify_count_and_frequency): Update.
1090 (cgraph_node::verify_node): Temporarily disable frequency verification.
1091 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1092 to_cgraph_frequency.
1093 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1094 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1096 (cgraph_node::expand_thunk): Update profile.
1097 * except.c (dw2_build_landing_pads): Do not update frequency.
1098 * final.c (compute_alignments): Use to_frequency.
1099 (dump_basic_block_info): Do not dump frequency.
1100 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1101 (dump_gimple_bb_header): Do not dump frequency.
1102 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1104 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1105 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1106 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1107 (init_before_recovery): Do not update frequency.
1108 (sched_create_recovery_edges): Do not update frequency.
1109 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1110 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1111 (ipa_cp_c_finalize): Set max_count to uninitialized.
1112 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1113 (param_change_prob): Use counts.
1114 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1116 * ipa-split.c (consider_split): Use to_frequency.
1117 (split_function): Use to_frequency.
1118 * ira-build.c (loop_compare_func): Likewise.
1119 (mark_loops_for_removal): Likewise.
1120 (mark_all_loops_for_removal): Likewise.
1121 * loop-doloop.c (doloop_modify): Do not update frequency.
1122 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1124 * lto-streamer-in.c (input_function): Update count_max.
1125 * omp-expand.c (expand_omp_taskreg): Update count_max.
1126 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1127 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1128 (maybe_hot_count_p): Use ipa counts only.
1129 (maybe_hot_bb_p): Simplify.
1130 (maybe_hot_edge_p): Simplify.
1131 (probably_never_executed): Do not take frequency argument.
1132 (probably_never_executed_bb_p): Do not pass frequency.
1133 (probably_never_executed_edge_p): Likewise.
1134 (combine_predictions_for_bb): Check that profile is nonzero.
1135 (propagate_freq): Do not set frequency.
1136 (drop_profile): Simplify.
1137 (counts_to_freqs): Simplify.
1138 (expensive_function_p): Use to_frequency.
1139 (propagate_unlikely_bbs_forward): Simplify.
1140 (determine_unlikely_bbs): Simplify.
1141 (estimate_bb_frequencies): Add hack to silence graphite issues.
1142 (compute_function_frequency): Use ipa counts.
1143 (pass_profile::execute): Update.
1144 (rebuild_frequencies): Use counts only.
1145 (force_edge_cold): Use counts only.
1146 * profile-count.c (profile_count::dump): Dump new count types.
1147 (profile_count::differs_from_p): Check compatiblity.
1148 (profile_count::to_frequency): New function.
1149 (profile_count::to_cgraph_frequency): New function.
1150 * profile-count.h (struct function): Declare.
1151 (enum profile_quality): Add profile_guessed_local and
1152 profile_guessed_global0.
1153 (class profile_proability): Decrease number of bits to 29;
1154 update from_reg_br_prob_note and to_reg_br_prob_note.
1155 (class profile_count: Update comment; decrease number of bits
1156 to 61. Check compatibility.
1157 (profile_count::compatible_p): New private member function.
1158 (profile_count::ipa_p): New member function.
1159 (profile_count::operator<): Handle global zero correctly.
1160 (profile_count::operator>): Handle global zero correctly.
1161 (profile_count::operator<=): Handle global zero correctly.
1162 (profile_count::operator>=): Handle global zero correctly.
1163 (profile_count::nonzero_p): New member function.
1164 (profile_count::force_nonzero): New member function.
1165 (profile_count::max): New member function.
1166 (profile_count::apply_scale): Handle IPA scalling.
1167 (profile_count::guessed_local): New member function.
1168 (profile_count::global0): New member function.
1169 (profile_count::ipa): New member function.
1170 (profile_count::to_frequency): Declare.
1171 (profile_count::to_cgraph_frequency): Declare.
1172 * profile.c (OVERLAP_BASE): Delete.
1173 (compute_frequency_overlap): Delete.
1174 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1175 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1176 * sched-ebb.c (rank): Use counts only.
1177 * shrink-wrap.c (handle_simple_exit): Use counts only.
1178 (try_shrink_wrapping): Use counts only.
1179 (place_prologue_for_one_component): Use counts only.
1180 * tracer.c (find_best_predecessor): Use to_frequency.
1181 (find_trace): Use to_frequency.
1182 (tail_duplicate): Use to_frequency.
1183 * trans-mem.c (expand_transaction): Do not update frequency.
1184 * tree-call-cdce.c: Do not update frequency.
1185 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1186 (gimple_merge_blocks): Likewise.
1187 (gimple_split_edge): Likewise.
1188 (gimple_duplicate_sese_region): Likewise.
1189 (gimple_duplicate_sese_tail): Likewise.
1190 (move_sese_region_to_fn): Likewise.
1191 (gimple_account_profile_record): Likewise.
1192 (insert_cond_bb): Likewise.
1193 * tree-complex.c (expand_complex_div_wide): Likewise.
1194 * tree-eh.c (lower_resx): Update profile.
1195 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1197 (initialize_cfun): Do not initialize frequencies
1198 (freqs_to_counts): Delete.
1199 (copy_cfg_body): Ignore count parameter.
1200 (copy_body): Update.
1201 (expand_call_inline): Update count_max.
1202 (optimize_inline_calls): Update count_max.
1203 (tree_function_versioning): Update count_max.
1204 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1205 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1207 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1208 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1209 (try_peel_loop): Likewise.
1210 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1212 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1213 (tree_transform_and_unroll_loop): Do not use frequencies
1214 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1215 Use reliable prediction only.
1216 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1217 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1218 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1219 probability scaling.
1220 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1221 not update frequency
1222 (any_remaining_duplicated_blocks): Likewise.
1223 (update_profile): Likewise.
1224 (estimated_freqs_path): Delete.
1225 (freqs_to_counts_path): Delete.
1226 (clear_counts_path): Delete.
1227 (ssa_fix_duplicate_block_edges): Likewise.
1228 (duplicate_thread_path): Likewise.
1229 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1230 * tree-tailcall.c (decrease_profile): Do not update frequency.
1231 (eliminate_tail_call): Likewise.
1232 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1233 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1234 (optimize_mask_stores): Likewise.
1235 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1236 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1237 (ubsan_expand_ptr_ifn): Update profile.
1238 * value-prof.c (gimple_ic): Simplify.
1239 * value-prof.h (gimple_ic): Update prototype.
1240 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1241 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1243 (want_inline_self_recursive_call_p): Likewise.
1244 (resolve_noninline_speculation): Only cummulate defined counts.
1245 (inline_small_functions): Use nonzero_p.
1246 (ipa_inline): Do not access freed node.
1248 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1250 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1251 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1253 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1255 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1256 non-legitimate address.
1258 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1260 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1261 (*lt0_<mode>di, *lt0_<mode>si): New.
1263 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1265 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1266 TARGET_PAIRED_FLOAT.
1268 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1269 Jim Wilson <jim.wilson@linaro.org>
1271 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1272 * config/aarch64/aarch64-tune.md: Regenerated.
1273 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1274 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1276 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1278 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1279 (arc_expand_prologue): Restore blink for millicode.
1280 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1282 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1285 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1286 gen_vec_duplicate after forcing the scalar into a register.
1288 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1290 * combine (try_combine): Print the insns input to try_combine to the
1293 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1296 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1297 Remove second argument from aarch64_process_target_attr call.
1298 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1300 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1302 (aarch64_handle_attr_arch): Remove second argument.
1303 (aarch64_handle_attr_cpu): Ditto.
1304 (aarch64_handle_attr_tune): Ditto.
1305 (aarch64_handle_attr_isa_flags): Ditto.
1306 (aarch64_process_one_target_attr): Ditto.
1307 (aarch64_process_target_attr): Ditto.
1308 (aarch64_option_valid_attribute_p): Remove second argument.
1309 on aarch64_process_target_attr call.
1311 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1313 * diagnostic.c: Include "selftest-diagnostic.h".
1314 (selftest::assert_location_text): New function.
1315 (selftest::test_diagnostic_get_location_text): New function.
1316 (selftest::diagnostic_c_tests): Call it.
1318 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1320 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1321 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1322 (class selftest::test_diagnostic_context): Move to...
1323 * selftest-diagnostic.c: New file.
1324 * selftest-diagnostic.h: New file.
1326 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1328 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1329 offset range for FT32B.
1330 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1331 * config/ft32/ft32.md: Add TARGET_NOPM.
1332 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1333 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1335 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1337 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1339 2017-11-02 Jeff Law <law@redhat.com>
1341 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1342 virtual keyword on FINAL OVERRIDE members.
1344 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1345 virtuals together. Add virtual destructor.
1346 (substitute_and_fold_engine): Similarly.
1348 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1350 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1352 2017-11-02 Richard Biener <rguenther@suse.de>
1354 PR tree-optimization/82795
1355 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1357 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1359 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1361 (gcc_GAS_CHECK_FEATURE): Remove.
1362 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1363 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1364 * configure: Regenerate.
1366 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1368 * config/arc/arc.c (hwloop_optimize): Account for empty
1371 2017-11-02 Richard Biener <rguenther@suse.de>
1374 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1375 Truncate ARRAY_REF index and element size.
1377 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1379 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1381 2017-11-01 Jeff Law <law@redhat.com>
1383 * tree-ssa-ccp.c (ccp_folder): New class derived from
1384 substitute_and_fold_engine.
1385 (ccp_folder::get_value): New member function.
1386 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1387 (ccp_fold_stmt): Remove prototype.
1388 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1389 * tree-ssa-copy.c (copy_folder): New class derived from
1390 substitute_and_fold_engine.
1391 (copy_folder::get_value): Renamed from get_value.
1392 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1393 * tree-vrp.c (vrp_folder): New class derived from
1394 substitute_and_fold_engine.
1395 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1396 (vrp_folder::get_value): New member function.
1397 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1398 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1399 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1400 provide a class interface to folder/substitute routines.
1401 (ssa_prop_fold_stmt_fn): Remove typedef.
1402 (ssa_prop_get_value_fn): Likewise.
1403 (subsitute_and_fold): Remove prototype.
1404 (replace_uses_in): Likewise.
1405 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1406 Renamed from replace_uses_in. Call the virtual member function
1407 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1408 (substitute_and_fold_dom_walker): Remove initialization of
1409 data member entries for calbacks. Add substitute_and_fold_engine
1410 member and initialize it.
1411 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1412 member functions for get_value, replace_phi_args_in c
1413 replace_uses_in, and fold_stmt calls.
1414 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1415 substitute_and_fold. Remove assert. Update ctor call.
1417 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1418 (ssa_prop_visit_phi_fn): Likewise.
1419 (class ssa_propagation_engine): New class to provide an interface
1421 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1423 (ssa_prop_visit_phi): Likewise.
1424 (ssa_propagation_engine::simulate_stmt): Moved into class.
1425 Call visit_phi/visit_stmt from the class rather than via
1426 file scoped static variables.
1427 (ssa_propagation_engine::simulate_block): Moved into class.
1428 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1429 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1430 set file scoped statics for the visit_stmt/visit_phi callbacks.
1431 * tree-complex.c (complex_propagate): New class derived from
1432 ssa_propagation_engine.
1433 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1434 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1435 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1437 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1438 ssa_propagation_engine.
1439 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1440 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1441 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1442 * tree-ssa-copy.c (copy_prop): New class derived from
1443 ssa_propagation_engine.
1444 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1445 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1446 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1447 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1448 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1449 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1450 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1452 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1454 PR rtl-optimization/82778
1455 PR rtl-optimization/82597
1456 * compare-elim.c (struct comparison): Add in_a_setter field.
1457 (find_comparison_dom_walker::before_dom_children): Remove killed
1458 bitmap and df_simulate_find_defs call, instead walk the defs.
1459 Compute last_setter and initialize in_a_setter. Merge definitions
1460 with first initialization for a few variables.
1461 (try_validate_parallel): Use insn_invalid_p instead of
1462 recog_memoized. Return insn rather than just the pattern.
1463 (try_merge_compare): Fix up comment. Don't uselessly test if
1464 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1466 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1469 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1470 Alan Hayward <alan.hayward@arm.com>
1471 David Sherwood <david.sherwood@arm.com>
1473 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1474 aarch64_hard_regno_nregs to get the number of registers
1477 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1478 Alan Hayward <alan.hayward@arm.com>
1479 David Sherwood <david.sherwood@arm.com>
1481 * config/aarch64/constraints.md (Upl): Rename to...
1483 * config/aarch64/aarch64.md
1484 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1487 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1488 Alan Hayward <alan.hayward@arm.com>
1489 David Sherwood <david.sherwood@arm.com>
1491 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1492 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1495 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1496 Alan Hayward <alan.hayward@arm.com>
1497 David Sherwood <david.sherwood@arm.com>
1499 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1500 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1501 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1503 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1504 of a single element.
1505 * config/aarch64/iterators.md: Add a comment above the permute
1506 unspecs to say that they are generated directly by
1507 aarch64_expand_vec_perm_const.
1508 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1510 2017-11-01 Nathan Sidwell <nathan@acm.org>
1512 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1514 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1516 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1517 and medany code models, and describe what they do.
1519 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1521 Revert accidental duplicate:
1523 * combine.c (can_change_dest_mode): Reject changes in
1524 REGMODE_NATURAL_SIZE.
1526 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1528 PR rtl-optimization/64682
1529 PR rtl-optimization/69567
1530 PR rtl-optimization/69737
1531 PR rtl-optimization/82683
1532 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1533 register mentioned in the note, drop the note, unless it came from I3,
1534 in which case it should go to I3 again.
1536 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1538 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1539 and return false if not.
1540 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1542 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1544 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1545 range is known to be empty.
1547 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1548 Alan Hayward <alan.hayward@arm.com>
1549 David Sherwood <david.sherwood@arm.com>
1551 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1552 and CONST_VECTOR_NUNITS instead of computing the number of units from
1553 the byte sizes of the vector and element.
1554 (simplify_binary_operation_1): Likewise.
1555 (simplify_const_binary_operation): Likewise.
1556 (simplify_ternary_operation): Likewise.
1558 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1559 Alan Hayward <alan.hayward@arm.com>
1560 David Sherwood <david.sherwood@arm.com>
1562 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1563 (int_mem_offset): ...this new function.
1564 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1565 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1566 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1569 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1570 Alan Hayward <alan.hayward@arm.com>
1571 David Sherwood <david.sherwood@arm.com>
1573 * lower-subreg.c (interesting_mode_p): New function.
1574 (compute_costs, find_decomposable_subregs, decompose_register)
1575 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1576 (resolve_clobber, dump_choices): Use it.
1578 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1579 Alan Hayward <alan.hayward@arm.com>
1580 David Sherwood <david.sherwood@arm.com>
1582 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1584 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1585 Alan Hayward <alan.hayward@arm.com>
1586 David Sherwood <david.sherwood@arm.com>
1588 * alias.c (find_base_value, find_base_term): Only process integer
1589 truncations. Check the precision rather than the size.
1591 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1592 Alan Hayward <alan.hayward@arm.com>
1593 David Sherwood <david.sherwood@arm.com>
1595 * machmode.h (is_narrower_int_mode): New function
1596 * optabs.c (expand_float, expand_fix): Use it.
1597 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1599 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1600 Alan Hayward <alan.hayward@arm.com>
1601 David Sherwood <david.sherwood@arm.com>
1603 * rtl.h (narrower_subreg_mode): New function.
1604 * ira-color.c (update_costs_from_allocno): Use it.
1606 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1607 Alan Hayward <alan.hayward@arm.com>
1608 David Sherwood <david.sherwood@arm.com>
1610 * optabs-query.h (convert_optab_p): New function, split out from...
1611 (convert_optab_handler): ...here.
1612 (widening_optab_handler): Delete.
1613 (find_widening_optab_handler): Remove permit_non_widening parameter.
1614 (find_widening_optab_handler_and_mode): Likewise. Provide an
1615 override that operates on mode class wrappers.
1616 * optabs-query.c (widening_optab_handler): Delete.
1617 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1618 parameter. Assert that the two modes are the same class and that
1619 the "from" mode is narrower than the "to" mode. Use
1620 convert_optab_handler instead of widening_optab_handler.
1621 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1622 instead of widening_optab_handler.
1623 * expr.c (expand_expr_real_2): Update calls to
1624 find_widening_optab_handler.
1625 * optabs.c (expand_widen_pattern_expr): Likewise.
1626 (expand_binop_directly): Take the insn_code as a parameter.
1627 (expand_binop): Only call find_widening_optab_handler for
1628 conversion optabs; use optab_handler otherwise. Update calls
1629 to find_widening_optab_handler and expand_binop_directly.
1630 Use convert_optab_handler instead of widening_optab_handler.
1631 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1632 find_widening_optab_handler and use scalar_mode rather than
1634 (convert_plusminus_to_widen): Likewise.
1636 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1637 Alan Hayward <alan.hayward@arm.com>
1638 David Sherwood <david.sherwood@arm.com>
1640 * machmode.h (fixed_size_mode): New class.
1641 * rtl.h (get_pool_mode): Return fixed_size_mode.
1642 * gengtype.c (main): Add fixed_size_mode.
1643 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1644 (get_raw_arg_mode): Likewise.
1645 * doc/tm.texi: Regenerate.
1646 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1647 * targhooks.c (default_get_reg_raw_mode): Likewise.
1648 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1649 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1650 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1651 (msp430_get_raw_result_mode): Likewise.
1652 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1653 * dbxout.c (dbxout_parms): Require fixed-size modes.
1654 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1655 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1656 * omp-low.c (lower_oacc_reductions): Likewise.
1657 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1658 (simplify_subreg): Update accordingly.
1659 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1660 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1661 that aren't fixed-size.
1662 (get_pool_mode): Return a fixed_size_mode.
1663 (output_constant_pool_2): Take a fixed_size_mode.
1665 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1666 Alan Hayward <alan.hayward@arm.com>
1667 David Sherwood <david.sherwood@arm.com>
1669 * doc/rtl.texi (vec_series): Document.
1670 (const): Say that the operand can be a vec_series.
1671 * rtl.def (VEC_SERIES): New rtx code.
1672 * rtl.h (const_vec_series_p_1): Declare.
1673 (const_vec_series_p): New function.
1674 * emit-rtl.h (gen_const_vec_series): Declare.
1675 (gen_vec_series): Likewise.
1676 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1677 (gen_vec_series): Likewise.
1678 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1679 * simplify-rtx.c (simplify_unary_operation): Handle negations
1681 (simplify_binary_operation_series): New function.
1682 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1683 (test_vector_ops_series): New function.
1684 (test_vector_ops): Call it.
1685 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1686 gen_const_vec_series.
1687 (altivec_lvsr): Likewise.
1688 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1690 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1691 Alan Hayward <alan.hayward@arm.com>
1692 David Sherwood <david.sherwood@arm.com>
1694 * doc/rtl.texi (const): Update description of address constants.
1695 Say that vector constants are allowed too.
1696 * common.md (E, F): Use CONSTANT_P instead of checking for
1698 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1699 checking for CONST_VECTOR.
1700 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1702 * expr.c (expand_expr_real_2): Check for vector modes instead
1703 of checking for CONST_VECTOR.
1704 * rtl.h (const_vec_p): New function.
1705 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1706 (unwrap_const_vec_duplicate): Handle them here too.
1708 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1709 David Malcolm <dmalcolm@redhat.com>
1710 Alan Hayward <alan.hayward@arm.com>
1711 David Sherwood <david.sherwood@arm.com>
1713 * rtl.h (vec_duplicate_p): New function.
1714 * selftest-rtl.c (assert_rtx_eq_at): New function.
1715 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1716 (assert_rtx_eq_at): Declare.
1717 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1718 * selftest-run-tests.c (selftest::run_tests): Call it.
1719 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1720 (simplify_unary_operation_1): Recursively handle vector duplicates.
1721 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1723 (simplify_subreg): Handle subregs of vector duplicates.
1724 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1725 (selftest::simplify_rtx_c_tests): New functions.
1727 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1728 Alan Hayward <alan.hayward@arm.com>
1729 David Sherwood <david.sherwood@arm.com>
1731 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1732 (gen_vec_duplicate): Likewise.
1733 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1735 (gen_const_vector): ...here.
1736 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1737 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1738 whose elements are all equal.
1739 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1740 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1741 (simplify_relational_operation): Likewise.
1742 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1744 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1745 (aarch64_expand_vector_init): Likewise.
1746 * config/arm/arm.c (neon_vdup_constant): Likewise.
1747 (neon_expand_vector_init): Likewise.
1748 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1749 (arm_block_set_unaligned_vect): Likewise.
1750 (arm_block_set_aligned_vect): Likewise.
1751 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1752 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1753 (expand_vec_perm_even_odd_pack): Likewise.
1754 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1755 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1756 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1757 gen_const_vec_duplicate.
1758 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1759 * config/mips/mips.c (mips_gen_const_int_vector): Use
1760 gen_const_vec_duplicate.
1761 (mips_expand_vector_init): Use CONST0_RTX.
1762 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1763 (define_split): Use gen_const_vec_duplicate.
1764 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1765 (define_split): Use gen_const_vec_duplicate.
1766 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1767 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1768 * config/spu/spu.c (spu_const): Likewise.
1770 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1771 Alan Hayward <alan.hayward@arm.com>
1772 David Sherwood <david.sherwood@arm.com>
1774 * combine.c (can_change_dest_mode): Reject changes in
1775 REGMODE_NATURAL_SIZE.
1777 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1779 * configure.ac (--enable-libssp): New.
1780 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1781 trying to determine target-specific default. Adjust indentation.
1782 * configure: Regenerated.
1783 * doc/install.texi (Configuration): Expand --disable-libssp
1786 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1788 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1791 2017-10-31 Martin Jambor <mjambor@suse.cz>
1794 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1796 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1798 * auto-profile.c (autofdo_source_profile::read): Use
1799 UNKNOWN_LOCATION rather than 0.
1800 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1801 (warning_at): ...this overload.
1802 (warning_at_rich_loc_n): Rename to...
1803 (warning_n): ...this overload.
1804 (error_at_rich_loc): Rename to...
1805 (error_at): ...this overload.
1806 (pedwarn_at_rich_loc): Rename to...
1807 (pedwarn): ...this overload.
1808 (permerror_at_rich_loc): Rename to...
1809 (permerror): ...this overload.
1810 (inform_at_rich_loc): Rename to...
1811 (inform): ...this overload.
1812 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1813 (diagnostic_n_impl_richloc): Rename to...
1814 (diagnostic_n_impl): ...this rich_location *-based decl.
1815 (inform_at_rich_loc): Rename to...
1816 (inform): ...this, and add an assertion.
1817 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1818 (warning_at_rich_loc): Rename to...
1819 (warning_at): ...this, and add an assertion.
1820 (warning_at_rich_loc_n): Rename to...
1821 (warning_n): ...this, and add an assertion.
1822 (warning_n): Update location_t-based implementation for removal of
1823 location_t-based diagnostic_n_impl.
1824 (pedwarn_at_rich_loc): Rename to...
1825 (pedwarn): ...this, and add an assertion.
1826 (permerror_at_rich_loc): Rename to...
1827 (permerror): ...this, and add an assertion.
1828 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1829 (error_at_rich_loc): Rename to...
1830 (error_at): ...this, and add an assertion.
1831 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1832 (driver::do_spec_on_infiles): Likewise.
1833 * substring-locations.c (format_warning_va): Update for renaming
1834 of inform_at_rich_loc.
1836 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1838 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1839 _Float<N>X built-in functions so that the variant without the
1840 "__builtin_" prefix is only enabled for the GNU C and Objective C
1841 languages when they are in non-strict ANSI/ISO mode.
1842 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1843 * target.def (floatn_builtin_p): Add a target hook to control
1844 whether _Float<N> and _Float<N>X built-in functions without the
1845 "__builtin_" prefix are enabled, and return true for C and
1846 Objective C in the default hook. Include langhooks.h in
1848 * targhooks.h (default_floatn_builtin_p): Likewise.
1849 * targhooks.c (default_floatn_builtin_p): Likewise.
1850 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1851 floatn_builtin_p target hook.
1852 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1854 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1855 Eric Botcazou <ebotcazou@adacore.com>
1857 PR rtl-optimization/81803
1858 * lra-constraints.c (curr_insn_transform): Also reload the whole
1859 register for a strict subreg no wider than a word if this is for
1860 a WORD_REGISTER_OPERATIONS target.
1862 2017-10-31 Jason Merrill <jason@redhat.com>
1864 * gdbinit.in: Skip over inlines from timevar.h.
1866 2017-10-31 Martin Liska <mliska@suse.cz>
1868 * doc/gcov.texi: Document new option.
1869 * gcov.c (print_usage): Likewise print it.
1870 (process_args): Support the argument.
1871 (format_count): New function.
1872 (format_gcov): Use the function.
1874 2017-10-31 Martin Liska <mliska@suse.cz>
1876 * gcov.c (struct name_map): do not use typedef.
1877 Define operator== and operator<.
1878 (name_search): Remove.
1879 (name_sort): Remove.
1880 (main): Do not allocate names.
1881 (process_file): Add vertical space.
1882 (generate_results): Use std::find.
1883 (release_structures): Do not release memory.
1884 (find_source): Use std::find.
1886 2017-10-31 Martin Liska <mliska@suse.cz>
1888 * gcov.c (struct line_info): Remove it's typedef.
1889 (line_info::line_info): Add proper ctor.
1890 (line_info::has_block): Do not use a typedef.
1891 (struct source_info): Do not use typedef.
1892 (circuit): Likewise.
1893 (get_cycles_count): Likewise.
1894 (output_intermediate_file): Iterate via vector iterator.
1895 (add_line_counts): Use std::vector methods.
1896 (accumulate_line_counts): Likewise.
1897 (output_lines): Likewise.
1899 2017-10-31 Martin Liska <mliska@suse.cz>
1901 * gcov.c (struct source_info): Remove typedef.
1902 (source_info::source_info): Add proper ctor.
1903 (accumulate_line_counts): Use struct, not it's typedef.
1904 (output_gcov_file): Likewise.
1905 (output_lines): Likewise.
1906 (main): Do not allocate an array.
1907 (output_intermediate_file): Use size of vector container.
1908 (process_file): Resize the vector.
1909 (generate_results): Do not preallocate, use newly added vector
1911 (release_structures): Do not release sources.
1912 (find_source): Use vector methods.
1913 (add_line_counts): Do not use typedef.
1915 2017-10-31 Martin Liska <mliska@suse.cz>
1917 * doc/gcov.texi: Document that.
1918 * gcov.c (add_line_counts): Mark lines with a non-executed
1920 (output_line_beginning): Handle such lines.
1921 (output_lines): Pass new argument.
1922 (output_intermediate_file): Print it in intermediate format.
1924 2017-10-31 Martin Liska <mliska@suse.cz>
1926 * color-macros.h: New file.
1927 * diagnostic-color.c: Factor out color related to macros to
1929 * doc/gcov.texi: Document -k option.
1930 * gcov.c (INCLUDE_STRING): Include string.h.
1931 (print_usage): Add -k option.
1932 (process_args): Parse it.
1933 (pad_count_string): New function.
1934 (output_line_beginning): Likewise.
1935 (DEFAULT_LINE_START): New macro.
1936 (output_lines): Support color output.
1938 2017-10-31 Martin Liska <mliska@suse.cz>
1940 PR gcov-profile/82633
1941 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1942 their interaction with GCOV infrastructure.
1943 * configure.ac: Add -fkeep-{inline,static}-functions to
1945 * configure: Regenerate.
1947 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1950 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1952 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1955 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1956 into a register if it does not fit into an immediate offset field.
1958 2017-10-31 Olivier Hainque <hainque@adacore.com>
1960 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1962 2017-10-31 Julia Koval <julia.koval@intel.com>
1964 * config.gcc: Add gfniintrin.h.
1965 * config/i386/gfniintrin.h: New.
1966 * config/i386/i386-builtin-types.def
1967 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1968 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1969 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1970 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1971 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1972 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1973 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1974 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1975 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1976 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1977 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1978 * config/i386/immintrin.h: Include gfniintrin.h.
1979 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1981 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1983 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1985 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1987 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1988 (arm_ashldi3_1bit): Remove pattern.
1989 (ashrdi3): Remove shift by 1 expansion.
1990 (arm_ashrdi3_1bit): Remove pattern.
1991 (lshrdi3): Remove shift by 1 expansion.
1992 (arm_lshrdi3_1bit): Remove pattern.
1993 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1994 cost of ashldi3 by 1.
1995 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1996 (<shift>di3_neon): Likewise.
1998 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2000 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2001 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2002 and (*aarch64_simd_mov<VQ:mode>).
2003 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2004 pattern alternative.
2005 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2006 attributes to match pattern alternative.
2008 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2010 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2011 * config/rs6000/emmintrin.h: New file.
2012 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2014 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2016 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2017 * (movdi_vfp_cortexa8): Remove pattern.
2019 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2021 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2023 (Specific, alpha*-dec-osf5.1): Remove.
2024 (Specific, mips-sgi-irix5): Remove.
2025 (Specific, mips-sgi-irix6): Remove.
2027 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2030 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2031 arguments to clear_bit_region_be.
2033 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2035 * gimplify.c: Include memmodel.h.
2037 2017-10-30 Martin Jambor <mjambor@suse.cz>
2039 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2040 condition whether loop should be executed at all.
2042 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2044 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2045 gimple folding of vec_madd() intrinsics.
2046 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2047 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2048 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2050 2017-10-30 Richard Biener <rguenther@suse.de>
2052 PR tree-optimization/82762
2054 2017-10-23 Richard Biener <rguenther@suse.de>
2056 PR tree-optimization/82129
2058 2017-08-01 Richard Biener <rguenther@suse.de>
2060 PR tree-optimization/81181
2061 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2062 (compute_antic): ... end of iteration here.
2064 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2066 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2067 -std=iso9899:2017 and -std=gnu17.
2068 * doc/standards.texi (C Language): Document C17 support.
2069 * doc/cpp.texi (Overview): Mention -std=c17.
2070 (Standard Predefined Macros): Document C11 and C17 values of
2071 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2072 * doc/extend.texi (Inline): Do not list individual options for
2073 standards newer than C99.
2074 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2076 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2079 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2081 * asan.c (asan_finish_file): Align asan globals array by shadow
2084 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2087 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2088 (struct store_immediate_info): Add bitregion_start and bitregion_end
2090 (store_immediate_info::store_immediate_info): Add brs and bre
2091 arguments and initialize bitregion_{start,end} from those.
2092 (struct merged_store_group): Add bitregion_start, bitregion_end,
2093 align_base and mask fields. Drop unnecessary struct keyword from
2094 struct store_immediate_info. Add do_merge method.
2095 (clear_bit_region_be): Use memset instead of loop storing zeros.
2096 (merged_store_group::do_merge): New method.
2097 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2098 stores as long as the surrounding bitregions have no gaps.
2099 (merged_store_group::merge_overlapping): Use do_merge.
2100 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2101 is byte aligned, rather than requiring that start and width are
2102 byte aligned. Drop unnecessary struct keyword from
2103 struct store_immediate_info. Allocate and populate also mask array.
2104 Make start of the arrays relative to bitregion_start rather than
2105 start and size them according to bitregion_{end,start} difference.
2106 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2107 struct store_immediate_info.
2108 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2109 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2110 struct keyword from struct store_immediate_info.
2111 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2112 between stores as long as the surrounding bitregions have no gaps.
2114 (struct split_store): Add orig non-static data member.
2115 (split_store::split_store): Initialize orig to false.
2116 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2117 if there is exactly a single original stmt. Change stmts argument
2118 to pointer from reference, if NULL, don't push anything to it. Add
2119 first argument, use it to optimize skipping over orig stmts that
2120 are known to be before bitpos already. Simplify.
2121 (split_group): Return unsigned int count how many stores are or
2122 would be needed rather than a bool. Add allow_unaligned argument.
2123 Change split_stores argument from reference to pointer, if NULL,
2124 only do a dry run computing how many stores would be produced.
2125 Rewritten algorithm to use both alignment and misalign if
2126 !allow_unaligned and handle bitfield stores with gaps.
2127 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2128 from bitregion_start instead of start. Compute allow_unaligned
2129 here, if true, do 2 split_group dry runs to compute which one
2130 produces fewer stores and prefer aligned if equal. Punt if
2131 new count is bigger or equal than original before emitting any
2132 statements, rather than during that. Remove no longer needed
2133 new_ssa_names tracking. Replace num_stmts with
2134 split_stores.length (). Use 32-bit stack allocated entries
2135 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2136 if possible. Handle bitfields with gaps.
2137 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2138 Compute bitregion_{start,end} for the stores and construct
2139 store_immediate_info with that. Formatting fixes.
2141 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2144 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2145 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2147 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2149 * gimplify.c: Include tm_p.h.
2151 * common.opt (gcoff): Re-add as ignored option.
2152 (gcoff1, gcoff2, gcoff3): Likewise.
2154 * Makefile.in (OBJS): Delete sdbout.o.
2155 (GTFILES): Delete $(srcdir)/sdbout.c.
2156 * debug.h: Delete sdb_debug_hooks.
2157 * final.c: Delete sdbout.h include.
2158 (final_scan_insn): Delete SDB_DEBUG check.
2159 (rest_of_clean_state): Likewise.
2160 * output.h: Delete sdb_begin_function_line.
2163 * toplev.c: Delete sdbout.h include.
2164 (process_options): Delete SDB_DEBUG check.
2165 * tree-core.h (tree_type_common): Delete pointer field of
2167 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2168 TYPE_SYMTAB_POINTER.
2169 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2170 (TYPE_SYMTAB_IS_POINTER): Delete.
2171 (TYPE_SYMTAB_IS_DIE): Renumber.
2172 * xcoffout.c: Refer to former sdbout.c file.
2173 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2175 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2176 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2177 (Debugging Options): Delete -gcoff.
2178 (-gstabs): Delete SDB reference.
2180 (-gcoff@var{level}): Delete.
2181 * doc/passes.texi (Debugging information output): Delete SDB and
2182 sdbout.c references.
2183 * doc/tm.texi: Regenerate.
2184 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2185 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2187 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2188 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2189 (SDB_DEBUGGING_INFO): Delete.
2190 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2191 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2192 * target.def (output_source_filename): Delete COFF reference.
2194 * common.opt (gcoff): Delete.
2195 (gxcoff+): Update Negative chain.
2196 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2198 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2199 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2200 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2202 (expand_function_start): Change sdb reference to past tense.
2203 (expand_function_end): Change sdb reference to past tense.
2204 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2205 * opts.c (debug_type_names): Delete coff entry.
2206 (common_handle_option): Delete OPT_gcoff case.
2207 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2209 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2210 * config/cris/cris.h: Delete SDB reference in comment.
2211 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2212 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2213 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2214 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2216 (ix86_expand_prologue): Likewise.
2217 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2218 * config/ia64/ia64.h: Likewise.
2219 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2220 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2222 * config/mmix/mmix.h: Likewise.
2223 * config/nds32/nds32.c: Likewise.
2224 * config/stormy/storym16.h: Likewise.
2225 * config/visium/visium.h: Likewise.
2226 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2228 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2230 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2231 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2232 FRAME_POINTER_REGNUM point at high end of local var area.
2234 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2236 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2237 Move comment around. Do not reset best_edge for a copiable
2238 destination if the copy would cause a partition change.
2239 (better_edge_p): Remove redundant check.
2241 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2243 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2245 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2247 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2248 for math functions that have _Float<N> and _Float<N>X variants.
2249 (mathfn_built_in_2): Add support for math functions that have
2250 _Float<N> and _Float<N>X variants.
2251 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2252 (expand_builtin_mathfn_ternary): Add support for fma with
2253 _Float<N> and _Float<N>X variants.
2254 (expand_builtin): Likewise.
2255 (fold_builtin_3): Likewise.
2256 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2257 create math function _Float<N> and _Float<N>X variants as external
2259 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2260 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2261 the __builtin_ prefix and if not strict ansi, without the prefix.
2262 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2263 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2264 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2265 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2266 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2267 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2268 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2269 function signatures for fma _Float<N> and _Float<N>X variants.
2270 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2271 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2272 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2273 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2274 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2275 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2276 * gencfn-macros.c (print_case_cfn): Add support for math functions
2277 that have _Float<N> and _Float<N>X variants.
2278 (print_define_operator_list): Likewise.
2279 (fltfn_suffixes): Likewise.
2281 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2282 for math functions that have _Float<N> and _Float<N>X variants.
2283 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2284 and _Float<N>X variants.
2285 (COPYSIGN): Likewise.
2288 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2289 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2291 (integer_valued_read_call_p): Likewise.
2292 * fold-const-call.c (fold_const_call_ss): Likewise.
2293 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2294 _Float<N> and _Float<N>X variants.
2295 (fold_const_call_ssss): Add support for fma _Float<N> and
2296 _Float<N>X variants.
2297 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2298 support for copysign and fma _Float<N> and _Float<N>X variants.
2299 (backprop::process_builtin_call_use): Likewise.
2300 * tree-call-cdce.c (can_test_argument_range); Add support for
2301 sqrt _Float<N> and _Float<N>X variants.
2302 (edom_only_function): Likewise.
2303 (get_no_error_domain): Likewise.
2304 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2305 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2306 copysign _Float<N> and _Float<N>X variants.
2307 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2308 handled by machine independent code.
2309 (FMAF128): Likewise.
2310 * doc/cpp.texi (Common Predefined Macros): Document defining
2311 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2312 fma _Float<N> and _Float<N>X variants.
2314 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2317 * config/i386/i386-modes.def (CCFPU): Remove definition.
2318 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2319 (ix86_cc_modes_compatible): Ditto.
2320 (ix86_expand_carry_flag_compare): Ditto.
2321 (ix86_expand_int_movcc): Ditto.
2322 (ix86_expand_int_addcc): Ditto.
2323 (ix86_reverse_condition): Ditto.
2324 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2325 Return true/false for unordered/ordered fp comparisons.
2326 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2327 (ix86_prepare_fp_compare_args): Update for rename.
2328 (ix86_expand_fp_compare): Update for rename. Generate unordered
2329 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2330 (ix86_expand_sse_compare_and_jump): Ditto.
2331 * config/i386/predicates.md (fcmov_comparison_operator):
2332 Remove CCFPU mode handling.
2333 (ix86_comparison_operator): Ditto.
2334 (ix86_carry_flag_operator): Ditto.
2335 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2336 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2337 (*cmpu<mode>_cc_i387): Ditto.
2338 (FPCMP): Remove mode iterator.
2339 (unord): Remove mode attribute.
2340 (unord_subst): New define_subst transformation
2341 (unord): New define_subst attribute.
2343 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2344 (*cmpi<unord>xf_i387): Ditto.
2345 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2346 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2347 using unord_subst transformation.
2348 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2349 (round_saeonly): Also handle CCFP mode.
2350 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2351 Remove UNSPEC_SAHF unspec handling.
2353 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2355 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2357 2017-10-27 Jeff Law <law@redhat.com>
2359 * gimple-ssa-sprintf.c: Include domwalk.h.
2360 (class sprintf_dom_walker): New class, derived from dom_walker.
2361 (sprintf_dom_walker::before_dom_children): New function.
2362 (struct call_info): Moved into sprintf_dom_walker class
2363 (compute_formath_length, handle_gimple_call): Likewise.
2364 (sprintf_length::execute): Call the dominator walker rather
2365 than walking the statements.
2367 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2368 gimple statement locations.
2369 (check_array_bounds): Corresponding changes. Get the statement's
2370 location directly from wi->stmt.
2372 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2375 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2377 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2379 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2380 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2382 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2385 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2386 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2387 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2388 maybe_get_pool_constant.
2389 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2392 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2394 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2395 2.26 caveat. Update gas and gld versions.
2396 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2399 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2401 * cgraph.h (set_malloc_flag): Declare.
2402 * cgraph.c (set_malloc_flag_1): New function.
2403 (set_malloc_flag): Likewise.
2404 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2405 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2407 (read_ipa_call_summary): Add support for reading is_return_callee.
2408 (write_ipa_call_summary): Stream is_return_callee.
2409 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2410 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2411 ipa-prop.h, ipa-fnsummary.h.
2412 (pure_const_names): Change to static.
2413 (malloc_state_e): Define.
2414 (malloc_state_names): Define.
2415 (funct_state_d): Add field malloc_state.
2416 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2417 (check_retval_uses): New function.
2418 (malloc_candidate_p): Likewise.
2419 (analyze_function): Add support for malloc attribute.
2420 (pure_const_write_summary): Stream malloc_state.
2421 (pure_const_read_summary): Add support for reading malloc_state.
2422 (dump_malloc_lattice): New function.
2423 (propagate_malloc): New function.
2424 (warn_function_malloc): New function.
2425 (ipa_pure_const::execute): Call propagate_malloc and
2426 ipa_free_fn_summary.
2427 (pass_local_pure_const::execute): Add support for malloc attribute.
2428 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2429 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2431 2017-10-27 Martin Liska <mliska@suse.cz>
2433 PR gcov-profile/82457
2434 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2435 for fork-like functions to be properly instrumented.
2437 2017-10-27 Richard Biener <rguenther@suse.de>
2440 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2441 info when we redirected EH.
2443 2017-10-26 Michael Collison <michael.collison@arm.com>
2445 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2447 (<optab>_trunchf<GPI:mode>2: New pattern.
2448 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2449 * config/aarch64/iterators.md (wv): New mode attribute.
2450 (vf, VF): New mode attributes.
2451 (vgp, VGP): New mode attributes.
2452 (s): Update attribute with SImode and DImode prefixes.
2454 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2456 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2457 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2458 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2459 (nios2_option_overide): Initialize it. Don't allow R0-relative
2460 addressing with PIC.
2461 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2462 (nios2_symbolic_constant_p): Likewise.
2463 (nios2_legitimate_address_p): Likewise.
2464 (nios2_r0rel_section_name_p): New.
2465 (nios2_symbol_ref_in_r0rel_data_p): New.
2466 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2467 (r0rel_constant_p): New.
2468 (nios2_print_operand_address): Handle r0rel_constant_p.
2469 (nios2_cdx_narrow_form_p): Likewise.
2470 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2471 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2472 (Nios II Options): Document -mr0rel-sec.
2474 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2476 * config/nios2/nios2.c: Include xregex.h.
2477 (nios2_gprel_sec_regex): New.
2478 (nios2_option_overide): Initialize it. Don't allow GP-relative
2479 addressing with PIC.
2480 (nios2_small_section_name_p): Check for regex match.
2481 * config/nios2/nios2.opt (mgprel-sec=): New option.
2482 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2483 (Nios II Options): Document -mgprel-sec.
2485 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2487 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2489 2017-10-26 Tom de Vries <tom@codesourcery.com>
2491 PR tree-optimization/82707
2492 * gimple.c (gimple_copy): Fix unsharing of
2493 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2495 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2497 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2498 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2499 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2500 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2501 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2502 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2503 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2504 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2505 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2506 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2507 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2508 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2509 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2510 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2511 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2512 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2513 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2515 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2517 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2519 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2520 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2521 warning to rs6000.c. Remove the Undocumented flag, since it has
2523 (-mabi=ibmlongdouble): Likewise.
2524 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2525 already set, set the default format for long double.
2526 (rs6000_debug_reg_global): Print whether long double is IBM or
2528 (rs6000_option_override_internal): Rework setting long double
2529 format. Only warn if the user is changing the long double default
2530 and they did not use -Wno-psabi.
2531 * doc/invoke.texi (PowerPC options): Update the documentation for
2532 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2534 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2535 Alan Hayward <alan.hayward@arm.com>
2536 David Sherwood <david.sherwood@arm.com>
2538 * rtl.h (wider_subreg_mode): New function.
2539 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2540 rather than an unsigned int *.
2541 * ira-color.c (regno_max_ref_width): Replace with...
2542 (regno_max_ref_mode): ...this new variable.
2543 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2544 Use wider_subreg_mode.
2545 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2546 rather than an unsigned int *.
2547 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2548 (process_alt_operands): Likewise.
2549 (invariant_p): Likewise.
2550 * lra-spills.c (assign_mem_slot): Likewise.
2551 (add_pseudo_to_slot): Likewise.
2552 * lra.c (collect_non_operand_hard_regs): Likewise.
2553 (add_regs_to_insn_regno_info): Likewise.
2554 * reload1.c (regno_max_ref_width): Replace with...
2555 (regno_max_ref_mode): ...this new variable.
2556 (reload): Update accordingly. Update call to
2557 ira_sort_regnos_for_alter_reg.
2558 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2559 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2560 (scan_paradoxical_subregs): Likewise.
2562 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2564 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2565 (aarch64_frame): Add emit_frame_chain boolean.
2566 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2567 Move eh_return case to aarch64_layout_frame.
2568 (aarch64_layout_frame): Initialize emit_frame_chain.
2569 (aarch64_expand_prologue): Use emit_frame_chain.
2571 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2573 * config/aarch64/aarch64.c (aarch64_layout_frame):
2574 Ensure LR is always stored at the bottom of the callee-saves.
2575 Remove rarely used frame layout which saves callee-saves at top of
2576 frame, so the store of LR can be used as a valid probe in all cases.
2578 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2580 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2581 Improve unaligned TImode/TFmode base/offset split.
2583 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2584 Alan Hayward <alan.hayward@arm.com>
2585 David Sherwood <david.sherwood@arm.com>
2587 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2588 * combine.c (find_single_use_1): Likewise.
2589 (expand_field_assignment): Likewise.
2590 (move_deaths): Likewise.
2591 * lra-constraints.c (simplify_operand_subreg): Likewise.
2592 (curr_insn_transform): Likewise.
2593 * lra.c (collect_non_operand_hard_regs): Likewise.
2594 (add_regs_to_insn_regno_info): Likewise.
2595 * rtlanal.c (reg_referenced_p): Likewise.
2596 (covers_regno_no_parallel_p): Likewise.
2598 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2600 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2601 Don't print any bits outside the precision of the value.
2602 * wide-int.cc (test_printing): Add some new tests.
2604 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2606 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2607 supports -xbrace_comment option.
2608 * configure: Regenerate.
2609 * config.in: Regenerate.
2610 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2611 (ASM_CPU_SPEC): Use it.
2613 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2615 * target.def (static_rtx_alignment): New hook.
2616 * targhooks.h (default_static_rtx_alignment): Declare.
2617 * targhooks.c (default_static_rtx_alignment): New function.
2618 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2619 * doc/tm.texi: Regenerate.
2620 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2621 instead of targetm.constant_alignment. Remove call to
2623 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2624 (cris_preferred_mininum_alignment): New function, split out from...
2625 (cris_constant_alignment): ...here.
2626 (cris_static_rtx_alignment): New function.
2627 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2629 (ix86_constant_alignment): ...here.
2630 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2631 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2632 (mmix_static_rtx_alignment): New function.
2633 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2634 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2636 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2639 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2640 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2642 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2644 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2645 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2646 mark as a sign-extending load.
2647 (local_pic_load_u): Define.
2649 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2652 * fold-const.c (operand_equal_for_comparison_p): Also return true
2653 if ARG0 is a simple variant of ARG1 with narrower precision.
2654 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2656 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2658 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2660 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2661 scatter_static, scatter_per_elt.
2662 * x86-tune-costs.h: Add new cost entries.
2664 2017-10-25 Richard Biener <rguenther@suse.de>
2666 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2667 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2668 class pass_fre): Move to ...
2669 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2670 class pass_fre): ... here and adjust for statistics.
2672 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2675 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2676 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2677 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2678 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2681 2017-10-25 Richard Biener <rguenther@suse.de>
2683 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2684 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2685 eliminate_push_avail, eliminate_insert): Move inside...
2686 (class eliminate_dom_walker): ... this class in preparation
2688 (fini_eliminate): Remove by merging with ...
2689 (eliminate): ... this function. Adjust for class changes.
2690 (pass_pre::execute): Remove fini_eliminate call.
2691 (pass_fre::execute): Likewise.
2693 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2696 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2697 (VPERMI2, VPERMI2I): New mode iterators.
2698 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2699 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2701 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2702 mode iterator. Remove 3 old define_insn patterns.
2703 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2704 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2705 VPERMI2 mode iterator, remove the other two expanders.
2706 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2707 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2708 instructions, remove the other two patterns.
2709 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2710 mode iterator, remove the other two patterns.
2711 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2712 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2713 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2714 and adjust argument order accordingly.
2715 (ix86_expand_vec_perm): Adjust caller.
2716 (expand_vec_perm_1): Likewise.
2717 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2718 (expand_vec_perm_vpermt2_vpshub2): ... this.
2719 (ix86_expand_vec_perm_const_1): Adjust caller.
2720 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2723 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2724 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2725 (vec_shl_<mode>): Remove unused expander.
2726 (avx512bw_<shift_insn><mode>3): New define_insn.
2727 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2728 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2730 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2733 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2736 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2738 PR rtl-optimization/82396
2739 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2740 (autopref_multipass_init): Simplify initialization.
2741 (autopref_rank_data): Simplify sort order.
2742 * gcc/sched-int.h (autopref_multipass_data_): Remove
2743 multi_mem_insn_p, min_offset and max_offset.
2745 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2748 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2749 Check special value of flag_omit_frame_pointer.
2750 (aarch64_can_eliminate): Likewise.
2751 (aarch64_override_options_after_change_1): Simplify handling of
2752 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2754 2017-10-24 Richard Biener <rguenther@suse.de>
2756 PR tree-optimization/82697
2757 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2758 zero for conditional load and unconditional store.
2760 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2762 * doc/install.texi: Document bootstrap-cet.
2764 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2767 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2768 ENDBR instruction at function entrance if function is only
2771 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2774 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2775 patterns to better describe from which operation the CF is computed.
2776 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2777 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2778 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2779 is 0, use _0 suffixed expanders instead of emitting a comparison
2782 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2784 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2785 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2787 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2790 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2791 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2792 * loop-iv.c (iv_get_reaching_def): Likewise.
2793 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2794 variable is promoted and the partition contains undefined values.
2796 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2798 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2800 (nios2_address_cost): Define.
2801 (nios2_legitimize_address): Recognize (exp + constant) directly.
2802 (TARGET_ADDRESS_COST): Define.
2804 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2806 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2807 (nios2_symbolic_memory_operand_p): Declare.
2808 (nios2_split_large_constant): Declare.
2809 (nios2_split_symbolic_memory_operand): Declare.
2810 * config/nios2/nios2.c: Adjust includes.
2811 (nios2_symbolic_constant_allowed): New.
2812 (nios2_symbolic_constant_p): New.
2813 (nios2_plus_symbolic_constant_p): New.
2814 (nios2_valid_addr_expr_p): Recognize addresses involving
2816 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2817 (nios2_symbolic_memory_operand_p): New.
2818 (nios2_large_constant_p): New.
2819 (nios2_split_large_constant): New.
2820 (nios2_split_plus_large_constant): New.
2821 (nios2_split_symbolic_memory_operand): New.
2822 (nios2_legitimize_address): Code refactoring. Handle addresses
2823 involving symbolic constants.
2824 (nios2_emit_move_sequence): Likewise.
2825 (nios2_print_operand): Improve error output.
2826 (nios2_print_operand_address): Handle LO_SUM.
2827 (nios2_cdx_narrow_form_p): Likewise.
2828 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2829 operands involving symbolic constants.
2830 (movhi_internal, movsi_internal): Likewise.
2831 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2832 (extendhisi2, extendqi<mode>2): Likewise.
2834 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2836 * tree-pass.h (PROP_rtl_split_insns): Define.
2837 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2839 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2841 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2843 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2846 * target.def (const_not_ok_for_debug_p): Default to
2847 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2848 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2849 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2850 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2851 which targetm.const_not_ok_for_debug_p returned true.
2852 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2854 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2856 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2857 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2858 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2860 (ix86_const_not_ok_for_debug_p): New function.
2861 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2862 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2864 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2867 * system.h: Conditionally include "unique-ptr.h" if
2868 INCLUDE_UNIQUE_PTR is defined.
2869 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2870 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2872 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2874 * config/rl78/rl78.md: New define_expand "subdi3".
2876 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2879 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2880 DF_REF_INSN if DF_REF_INSN_INFO is false.
2882 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2884 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2885 xmm_move instead of sse_move.
2886 (sse_store_index): New function.
2887 (ix86_register_move_cost): Be more sensible about mismatch stall;
2888 model AVX moves correctly; make difference between sse->integer and
2890 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2891 moves; make difference between SSE and AVX.
2892 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2893 and zmm_move. Increase size of sse load and store tables;
2894 add unaligned load and store tables; add ssemmx_to_integer.
2895 * x86-tune-costs.h: Update all entries according to real
2896 move latencies from Agner Fog's manual and chip documentation.
2898 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2901 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2902 * config/i386/constraints.md (Wf): New constraint.
2903 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2904 (cmp<dwi>_doubleword): Removed.
2905 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2906 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2907 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2908 expand with cmp<dwi>_doubleword. For LTU and GEU use
2909 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2911 * common.opt (gcolumn-info): Enable by default.
2912 * doc/invoke.texi (gcolumn-info): Document new default.
2914 2017-10-23 Richard Biener <rguenther@suse.de>
2916 PR tree-optimization/82672
2917 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2918 Fold the stmt if we propagated into it.
2920 2017-10-23 Richard Biener <rguenther@suse.de>
2922 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2923 (bitmap_remove_expr_from_set): ... this. All callers call this
2924 for non-constant values.
2925 (bitmap_set_subtract): Rename to...
2926 (bitmap_set_subtract_expressions): ... this. Adjust and
2928 (bitmap_set_contains_value): Remove superfluous check.
2929 (bitmap_set_replace_value): Inline into single caller ...
2930 (bitmap_value_replace_in_set): ... here and simplify.
2931 (dependent_clean): Merge into ...
2932 (clean): ... this using an overload. Adjust.
2933 (prune_clobbered_mems): Adjust.
2934 (compute_antic_aux): Likewise.
2935 (compute_partial_antic_aux): Likewise.
2937 2017-10-23 Richard Biener <rguenther@suse.de>
2939 PR tree-optimization/82129
2941 2017-08-01 Richard Biener <rguenther@suse.de>
2943 PR tree-optimization/81181
2944 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2945 (compute_antic): ... end of iteration here.
2947 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2949 * target.def (starting_frame_offset): New hook.
2950 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2951 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2952 * doc/tm.texi.in: Regenerate.
2953 * hooks.h (hook_hwi_void_0): Declare.
2954 * hooks.c (hook_hwi_void_0): New function.
2955 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2956 STARTING_FRAME_OFFSET.
2957 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2958 * reload1.c (reload): Likewise.
2959 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2960 instead of STARTING_FRAME_OFFSET.
2961 * function.c (try_fit_stack_local): Likewise.
2962 (assign_stack_local_1): Likewise
2963 (instantiate_virtual_regs): Likewise.
2964 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2965 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2966 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2967 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2968 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2969 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2970 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2971 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2972 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2973 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2974 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2975 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2976 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2977 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2978 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2979 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2980 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2981 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2982 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2983 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2984 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2985 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2986 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2987 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2988 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2989 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2990 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2991 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2992 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2993 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2994 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2995 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2996 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2997 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2998 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2999 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3000 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3001 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3002 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3003 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3004 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3005 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3006 return a HOST_WIDE_INT.
3007 (avr_builtin_setjmp_frame_value): Use it instead of
3008 STARTING_FRAME_OFFSET.
3009 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3010 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3011 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3013 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3014 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3015 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3016 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3017 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3018 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3019 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3020 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3021 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3022 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3023 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3024 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3026 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3027 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3028 * config/mips/mips.c (mips_compute_frame_info): Refer to
3029 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3030 (mips_starting_frame_offset): New function.
3031 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3032 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3033 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3034 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3035 and return a HOST_WIDE_INT.
3036 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3037 (mmix_initial_elimination_offset): Refer to
3038 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3039 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3040 * config/pa/pa.c (pa_starting_frame_offset): New function.
3041 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3042 (pa_expand_prologue): Likewise.
3043 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3044 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3045 !FRAME_GROWS_DOWNWARD handling to...
3046 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3047 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3048 !FRAME_GROWS_DOWNWARD handling to...
3049 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3050 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3051 !FRAME_GROWS_DOWNWARD handling to...
3052 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3053 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3055 (rs6000_starting_frame_offset): New function.
3056 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3057 !FRAME_GROWS_DOWNWARD handling to...
3058 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3059 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3060 !FRAME_GROWS_DOWNWARD handling to...
3061 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3062 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3063 !FRAME_GROWS_DOWNWARD handling to...
3064 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3065 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3066 (rs6000_starting_frame_offset): New function.
3067 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3068 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3069 * config/vax/vax.c (vax_starting_frame_offset): New function.
3070 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3071 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3072 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3073 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3074 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3075 * system.h (STARTING_FRAME_OFFSET): Poison.
3077 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3079 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3080 SCALAR_TYPE_MODE instead of TYPE_MODE.
3082 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3083 Alan Hayward <alan.hayward@arm.com>
3084 David Sherwood <david.sherwood@arm.com>
3086 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3088 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3089 Alan Hayward <alan.hayward@arm.com>
3090 David Sherwood <david.sherwood@arm.com>
3092 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3094 2017-10-23 Richard Biener <rguenther@suse.de>
3096 PR tree-optimization/82129
3097 * tree-ssa-pre.c (bitmap_set_and): Remove.
3098 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3099 canonicalizing expressions in the set to those with lowest
3100 ID rather than taking that from the first edge.
3102 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3104 * combine.c (rtx_equal_for_field_assignment_p): Use
3105 byte_lowpart_offset.
3107 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3108 Alan Hayward <alan.hayward@arm.com>
3109 David Sherwood <david.sherwood@arm.com>
3111 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3112 to a SUBREG_PROMOTED_VAR.
3114 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3115 Alan Hayward <alan.hayward@arm.com>
3116 David Sherwood <david.sherwood@arm.com>
3118 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3119 (expand_debug_source_expr): Likewise.
3120 * combine.c (combine_simplify_rtx): Likewise.
3121 * cse.c (fold_rtx): Likewise.
3122 * optabs.c (expand_float): Likewise.
3123 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3124 (simplify_binary_operation_1): Likewise.
3126 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3127 Alan Hayward <alan.hayward@arm.com>
3128 David Sherwood <david.sherwood@arm.com>
3130 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3131 (record_promoted_value): Likewise.
3132 * expr.c (expand_expr_real_2): Likewise.
3133 * ree.c (update_reg_equal_equiv_notes): Likewise.
3134 (combine_set_extension): Likewise.
3135 * rtlanal.c (low_bitmask_len): Likewise.
3136 * simplify-rtx.c (neg_const_int): Likewise.
3137 (simplify_binary_operation_1): Likewise.
3139 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3140 Alan Hayward <alan.hayward@arm.com>
3141 David Sherwood <david.sherwood@arm.com>
3143 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3144 * regcprop.c (maybe_mode_change): Likewise.
3145 * reload1.c (alter_reg): Likewise.
3147 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3149 * inchash.h (inchash::hash::add_wide_int): New function.
3150 * lto-streamer-out.c (hash_tree): Use it.
3152 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3154 * inchash.h (inchash::hash::add_wide_int): Rename to...
3155 (inchash::hash::add_hwi): ...this.
3156 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3157 (polymorphic_call_target_hasher::hash): Likewise.
3158 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3159 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3160 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3161 * lto-streamer-out.c (hash_tree): Likewise.
3162 * optc-save-gen.awk: Likewise.
3163 * tree.c (add_expr): Likewise.
3165 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3168 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3169 for ordered inequality comparisons even with TARGET_IEEE_FP.
3171 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3174 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3175 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3176 Expand with cmp<dwi>_doubleword.
3178 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3180 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3181 List CET intrinsics.
3182 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3183 specific to -fcf-protection option.
3185 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3187 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3188 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3189 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3190 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3191 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3192 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3193 (extra_objs): Add cet.o for Linux/x86 targets.
3194 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3195 * config/i386/cet.c: New file.
3196 * config/i386/cetintrin.h: Likewise.
3197 * config/i386/t-cet: Likewise.
3198 * config/i386/cpuid.h (bit_SHSTK): New.
3199 (bit_IBT): Likewise.
3200 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3201 pass IBT and SHSTK bits.
3202 * config/i386/i386-builtin-types.def
3203 (VOID_FTYPE_UNSIGNED_PVOID): New.
3204 (VOID_FTYPE_UINT64_PVOID): Likewise.
3205 * config/i386/i386-builtin.def: Add CET intrinsics.
3206 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3207 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3208 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3209 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3211 * config/i386/i386.c (rest_of_insert_endbranch): New.
3212 (pass_data_insert_endbranch): Likewise.
3213 (pass_insert_endbranch): Likewise.
3214 (make_pass_insert_endbranch): Likewise.
3215 (ix86_notrack_prefixed_insn_p): Likewise.
3216 (ix86_target_string): Add -mibt, -mshstk flags.
3217 (ix86_option_override_internal): Add flag_cf_protection
3219 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3220 (ix86_print_operand): Add 'notrack' prefix output.
3221 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3222 (ix86_expand_builtin): Expand CET intrinsics.
3223 (x86_output_mi_thunk): Add 'endbranch' instruction.
3224 * config/i386/i386.h (TARGET_IBT): New.
3225 (TARGET_IBT_P): Likewise.
3226 (TARGET_SHSTK): Likewise.
3227 (TARGET_SHSTK_P): Likewise.
3228 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3229 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3230 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3231 (builtin_setjmp_setup): New pattern.
3232 (builtin_longjmp): Likewise.
3233 (rdssp<mode>): Likewise.
3234 (incssp<mode>): Likewise.
3235 (saveprevssp): Likewise.
3236 (rstorssp): Likewise.
3237 (wrss<mode>): Likewise.
3238 (wruss<mode>): Likewise.
3239 (setssbsy): Likewise.
3240 (clrssbsy): Likewise.
3241 (nop_endbr): Likewise.
3242 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3244 * config/i386/immintrin.h: Include <cetintrin.h>.
3245 * config/i386/linux-common.h
3246 (file_end_indicate_exec_stack_and_cet): New prototype.
3247 (TARGET_ASM_FILE_END): New.
3249 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3251 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3252 latencies instead of having separate table; make difference between
3253 integer and float costs.
3254 * i386.h (processor_costs): Remove scalar_stmt_cost,
3255 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3256 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3258 * x86-tune-costs.h: Remove entries which has been removed in
3259 procesor_costs from all tables; make cond_taken_branch_cost
3260 and cond_not_taken_branch_cost COST_N_INSNS based.
3262 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3264 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3266 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3268 * config/i386/i386.md (isa): Remove fma_avx512f.
3269 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3270 <avx512>_fmadd_<mode>_mask3<round_name>,
3271 <avx512>_fmsub_<mode>_mask<round_name>,
3272 <avx512>_fmsub_<mode>_mask3<round_name>,
3273 <avx512>_fnmadd_<mode>_mask<round_name>,
3274 <avx512>_fnmadd_<mode>_mask3<round_name>,
3275 <avx512>_fnmsub_<mode>_mask<round_name>,
3276 <avx512>_fnmsub_<mode>_mask3<round_name>,
3277 <avx512>_fmaddsub_<mode>_mask<round_name>,
3278 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3279 <avx512>_fmsubadd_<mode>_mask<round_name>,
3280 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3281 (*vec_widen_umult_even_v16si<mask_name>,
3282 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3283 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3285 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3287 * extend.texi: Add 'nocf_check' documentation.
3288 * gimple.texi: Add second parameter to
3289 gimple_build_call_from_tree.
3290 * invoke.texi: Add -fcf-protection documentation.
3291 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3293 2017-10-20 Richard Biener <rguenther@suse.de>
3295 PR tree-optimization/82473
3296 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3297 the largest input type.
3299 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3301 * c-attribs.c (handle_nocf_check_attribute): New function.
3302 (c_common_attribute_table): Add 'nocf_check' handling.
3303 * gimple-parser.c: Add second argument NULL to
3304 gimple_build_call_from_tree.
3305 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3306 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3308 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3309 * common.opt: Add fcf-protection flag.
3310 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3311 * flag-types.h: Add enum cf_protection_level.
3312 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3313 Add 'nocf_check' attribute propagation to gimple call.
3314 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3315 (gimple_build_call_from_tree): Update prototype.
3316 (gimple_call_nocf_check_p): New function.
3317 (gimple_call_set_nocf_check): Likewise.
3318 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3319 * ipa-icf.c: Add nocf_check attribute in statement hash.
3320 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3321 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3322 * toplev.c (process_options): Add flag_cf_protection handling.
3324 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3326 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3328 2017-10-20 Richard Biener <rguenther@suse.de>
3330 PR tree-optimization/82603
3331 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3332 remove false predicated stores.
3334 2017-10-20 Richard Biener <rguenther@suse.de>
3336 * graphite-isl-ast-to-gimple.c
3337 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3338 Remove return value and simplify, dump copied stmt after lhs
3340 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3341 Reduce dump verbosity.
3342 (gsi_insert_earliest): Likewise.
3343 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3344 * graphite.c (print_global_statistics): Adjust dumping.
3345 (print_graphite_scop_statistics): Likewise.
3346 (print_graphite_statistics): Do not dump loops here.
3347 (graphite_transform_loops): But here.
3349 2017-10-20 Nicolas Roche <roche@adacore.com>
3351 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3352 * configure: Regenerate.
3354 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3357 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3358 functions when optimizing replace GIMPLE_RETURN stmts with
3359 calls to __builtin_unreachable ().
3362 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3363 for -fsanitize=thread link of executables.
3364 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3365 link of executables.
3368 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3370 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3371 define_insns for logical vector shifts to use VI248_AVX512BW
3372 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3373 condition, useless isa and prefix attributes. Change the first
3374 2 of these define_insns to ...
3375 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3376 define_insn for avx512vl.
3377 (<shift_insn><mode>3): ... and this, new define_insn without
3378 masking for non-avx512vl.
3381 * config/i386/sse.md (*andnot<mode>3,
3382 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3383 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3384 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3385 not applied use empty suffix even for TARGET_AVX512VL.
3386 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3387 is applied, supply evex,evex or evex,evex,evex instead of just
3390 2017-10-20 Julia Koval <julia.koval@intel.com>
3392 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3393 (OPTION_MASK_ISA_GFNI_UNSET): New.
3394 (ix86_handle_option): Handle OPT_mgfni.
3395 * config/i386/cpuid.h (bit_GFNI): New.
3396 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3397 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3398 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3399 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3400 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3401 * config/i386/i386.opt: Add mgfni.
3403 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3405 * config/msp430/msp430.c (msp430_option_override): Disable
3406 -fdelete-null-pointer-checks.
3407 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3409 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3411 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3412 of x87 and SSE instructions.
3414 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3416 * asan.c (create_cond_insert_point): Do not update edge count.
3417 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3418 (afdo_propagate_circuit): Likewise.
3419 (afdo_calculate_branch_prob): Likewise.
3420 (afdo_annotate_cfg): Likewise.
3421 * basic-block.h (struct edge_def): Remove count.
3422 (edge_def::count): New accessor.
3423 * bb-reorder.c (rotate_loop): Update.
3424 (find_traces_1_round): Update.
3425 (connect_traces): Update.
3426 (sanitize_hot_paths): Update.
3427 * cfg.c (unchecked_make_edge): Update.
3428 (make_single_succ_edge): Update.
3429 (check_bb_profile): Update.
3430 (dump_edge_info): Update.
3431 (update_bb_profile_for_threading): Update.
3432 (scale_bbs_frequencies_int): Update.
3433 (scale_bbs_frequencies_gcov_type): Update.
3434 (scale_bbs_frequencies_profile_count): Update.
3435 (scale_bbs_frequencies): Update.
3436 * cfganal.c (connect_infinite_loops_to_exit): Update.
3437 * cfgbuild.c (compute_outgoing_frequencies): Update.
3438 (find_many_sub_basic_blocks): Update.
3439 * cfgcleanup.c (try_forward_edges): Update.
3440 (try_crossjump_to_edge): Update
3441 * cfgexpand.c (expand_gimple_cond): Update
3442 (expand_gimple_tailcall): Update
3443 (construct_exit_block): Update
3444 * cfghooks.c (verify_flow_info): Update
3445 (redirect_edge_succ_nodup): Update
3446 (split_edge): Update
3447 (make_forwarder_block): Update
3448 (duplicate_block): Update
3449 (account_profile_record): Update
3450 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3451 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3452 * cfgloopmanip.c (scale_loop_profile): Update.
3454 (lv_adjust_loop_entry_edge): Update.
3455 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3456 (force_nonfallthru_and_redirect): Update.
3457 (purge_dead_edges): Update.
3458 (rtl_flow_call_edges_add): Update.
3459 * cgraphunit.c (init_lowered_empty_function): Update.
3460 (cgraph_node::expand_thunk): Update.
3461 * gimple-pretty-print.c (dump_probability): Update.
3462 (dump_edge_probability): Update.
3463 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3464 * haifa-sched.c (sched_create_recovery_edges): Update.
3465 * hsa-gen.c (convert_switch_statements): Update.
3466 * ifcvt.c (dead_or_predicable): Update.
3467 * ipa-inline-transform.c (inline_transform): Update.
3468 * ipa-split.c (split_function): Update.
3469 * ipa-utils.c (ipa_merge_profiles): Update.
3470 * loop-doloop.c (add_test): Update.
3471 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3472 * lto-streamer-in.c (input_cfg): Update.
3473 (input_function): Update.
3474 * lto-streamer-out.c (output_cfg): Update.
3475 * modulo-sched.c (sms_schedule): Update.
3476 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3477 * predict.c (maybe_hot_edge_p): Update.
3478 (unlikely_executed_edge_p): Update.
3479 (probably_never_executed_edge_p): Update.
3480 (dump_prediction): Update.
3481 (drop_profile): Update.
3482 (propagate_unlikely_bbs_forward): Update.
3483 (determine_unlikely_bbs): Update.
3484 (force_edge_cold): Update.
3485 * profile.c (compute_branch_probabilities): Update.
3486 * reg-stack.c (better_edge): Update.
3487 * shrink-wrap.c (handle_simple_exit): Update.
3488 * tracer.c (better_p): Update.
3489 * trans-mem.c (expand_transaction): Update.
3490 (split_bb_make_tm_edge): Update.
3491 * tree-call-cdce.c: Update.
3492 * tree-cfg.c (gimple_find_sub_bbs): Update.
3493 (gimple_split_edge): Update.
3494 (gimple_duplicate_sese_region): Update.
3495 (gimple_duplicate_sese_tail): Update.
3496 (gimple_flow_call_edges_add): Update.
3497 (insert_cond_bb): Update.
3498 (execute_fixup_cfg): Update.
3499 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3500 * tree-complex.c (expand_complex_div_wide): Update.
3501 * tree-eh.c (lower_resx): Update.
3502 (unsplit_eh): Update.
3503 (cleanup_empty_eh_move_lp): Update.
3504 * tree-inline.c (copy_edges_for_bb): Update.
3505 (freqs_to_counts): Update.
3506 (copy_cfg_body): Update.
3507 * tree-ssa-dce.c (remove_dead_stmt): Update.
3508 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3509 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3510 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3511 (unloop_loops): Update.
3512 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3513 * tree-ssa-loop-split.c (connect_loops): Update.
3514 (split_loop): Update.
3515 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3516 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3517 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3518 * tree-ssa-reassoc.c (branch_fixup): Update.
3519 * tree-ssa-tail-merge.c (replace_block_by): Update.
3520 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3521 (compute_path_counts): Update.
3522 (update_profile): Update.
3523 (recompute_probabilities): Update.
3524 (update_joiner_offpath_counts): Update.
3525 (estimated_freqs_path): Update.
3526 (freqs_to_counts_path): Update.
3527 (clear_counts_path): Update.
3528 (ssa_fix_duplicate_block_edges): Update.
3529 (duplicate_thread_path): Update.
3530 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3531 (case_bit_test_cmp): Update.
3532 (collect_switch_conv_info): Update.
3533 (gen_inbound_check): Update.
3534 (do_jump_if_equal): Update.
3535 (emit_cmp_and_jump_insns): Update.
3536 * tree-tailcall.c (decrease_profile): Update.
3537 (eliminate_tail_call): Update.
3538 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3539 (vect_do_peeling): Update.
3540 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3541 * ubsan.c (ubsan_expand_null_ifn): Update.
3542 (ubsan_expand_ptr_ifn): Update.
3543 * value-prof.c (gimple_divmod_fixed_value): Update.
3544 (gimple_mod_pow2): Update.
3545 (gimple_mod_subtract): Update.
3546 (gimple_ic): Update.
3547 (gimple_stringop_fixed_value): Update.
3549 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3552 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3554 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3556 PR rtl-optimization/82395
3557 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3558 based on non_spilled_static_chain_regno_p.
3560 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3562 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3563 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3564 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3566 2017-10-19 Martin Sebor <msebor@redhat.com>
3568 PR tree-optimization/82596
3569 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3571 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3573 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3574 (get_mem_refs_of_builtin_call): Likewise.
3575 * builtins.c (expand_builtin_apply): Adjust call to
3576 allocate_dynamic_stack_space.
3577 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3578 the third argument to allocate_dynamic_stack_space, otherwise -1.
3579 (expand_builtin): Deal with all alloca variants.
3580 (is_inexpensive_builtin): Likewise.
3581 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3582 * calls.c (special_function_p): Deal with all alloca variants.
3583 (initialize_argument_information): Adjust call to
3584 allocate_dynamic_stack_space.
3585 (expand_call): Likewise.
3586 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3587 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3588 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3589 use it for the stack usage computation.
3590 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3591 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3592 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3593 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3594 (in_loop_p): Remove first argument and useless check.
3595 (pass_walloca::execute): Remove useless test and adjust call to above.
3596 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3597 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3598 (gimplify_call_expr): Deal with all alloca variants.
3599 * hsa-gen.c (gen_hsa_alloca): Likewise.
3600 (gen_hsa_insns_for_call): Likewise.
3601 * ipa-pure-const.c (special_builtin_state): Likewise.
3602 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3603 * tree-object-size.c (alloc_object_size): Likewise.
3604 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3605 (call_may_clobber_ref_p_1): Likewise.
3606 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3607 (ccp_fold_stmt): Likewise.
3608 (optimize_stack_restore): Likewise.
3609 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3610 (mark_all_reaching_defs_necessary_1): Likewise.
3611 (propagate_necessity): Likewise.
3612 (eliminate_unnecessary_stmts): Likewise.
3613 * tree.c (build_common_builtin_nodes): Build
3614 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3615 (build_alloca_call_expr): New function.
3616 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3617 (CASE_BUILT_IN_ALLOCA): Likewise.
3618 (build_alloca_call_expr): Declare.
3619 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3621 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3624 * dwarf2out.c (new_die_raw): New static inline function.
3625 (new_die): Use it to create the DIE.
3626 (add_AT_external_die_ref): Likewise.
3627 (clone_die): Likewise.
3628 (clone_as_declaration): Likewise.
3629 (dwarf2out_vms_debug_main_pointer): Likewise.
3630 (base_type_die): Likewise. Remove early return for corner cases.
3631 Do not call add_pubtype on the DIE here.
3632 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3633 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3634 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3635 native order exists for base types, attach the DIE manually and call
3636 add_pubtype on it. Do not equate a reverse order DIE to the type.
3638 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3640 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3641 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3643 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3645 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3647 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3648 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3649 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3650 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3651 * opts.c (sanitizer_opts): Add builtin.
3652 * ubsan.c (instrument_builtin): New function.
3653 (pass_ubsan::execute): Call it.
3654 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3655 * doc/invoke.texi: Document -fsanitize=builtin.
3657 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3658 builtins, store max (log2 (align), 0) into uchar field instead of
3659 align into uptr field.
3660 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3661 store uchar 0 field instead of uptr 0 field.
3662 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3663 instead of passing one address of struct with 2 locations pass
3664 two addresses of structs with 1 location each.
3665 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3666 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3667 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3668 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3669 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3670 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3671 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3672 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3674 2017-10-19 Martin Liska <mliska@suse.cz>
3677 * file-find.c (remove_prefix): Remove.
3678 * file-find.h (remove_prefix): Likewise.
3679 * gcc-ar.c: Remove smartness of lookup.
3681 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3683 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3684 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3685 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3687 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3690 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3691 (setcc + and to xor + setcc): New peephole2.
3693 2017-10-19 Tom de Vries <tom@codesourcery.com>
3695 * doc/sourcebuild.texi (Test Directives, Variants of
3696 dg-require-support): Add dg-require-stack-size.
3698 2017-10-19 Martin Liska <mliska@suse.cz>
3701 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3702 that have a large alignment.
3703 (gimplify_target_expr): Likewise.
3705 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3707 PR rtl-optimization/82602
3708 * ira.c (rtx_moveable_p): Return false for volatile asm.
3710 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3713 * config/i386/i386-modes.def (CCGZ): New CC mode.
3714 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3715 * config/i386/predicates.md (ix86_comparison_operator):
3717 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3718 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3719 with double-word subtraction.
3720 (put_condition_code): Handle CCGZmode.
3722 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3724 * wide-int.cc (debug (const wide_int &)): New.
3725 (debug (const wide_int *)): New.
3726 (debug (const widest_int &)): New.
3727 (debug (const widest_int *)): New.
3729 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3732 * lra-constraints.c (curr_insn_transform): Use non-input operand
3733 instead of output one for matched reload.
3735 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3737 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3738 (tree-ssa-loop-ivopts.h): New header file.
3739 (struct builtin_info): New fields.
3740 (classify_builtin_1): Compute and record base and offset parts for
3741 memset builtin partition by calling strip_offset.
3742 (offset_cmp, fuse_memset_builtins): New functions.
3743 (finalize_partitions): Fuse adjacent memset partitions by calling
3745 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3746 Expose the interface.
3747 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3749 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3751 PR tree-optimization/82574
3752 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3753 that data reference must be executed exactly once per iteration
3754 against the outermost loop in nest.
3755 (classify_partition): Update call to above function.
3757 2017-10-18 Richard Biener <rguenther@suse.de>
3759 PR tree-optimization/82591
3760 * graphite.c (graphite_transform_loops): Move code gen message
3762 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3763 Here. Handle scop_to_isl_ast failing.
3764 (scop_to_isl_ast): Limit the number of ISL operations.
3766 2017-10-18 Richard Biener <rguenther@suse.de>
3768 * graphite-isl-ast-to-gimple.c
3769 (translate_isl_ast_to_gimple::set_rename): Simplify.
3770 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3771 (graphite_copy_stmts_from_block): ... here.
3772 (copy_bb_and_scalar_dependences): Simplify.
3773 (add_parameters_to_ivs_params): Canonicalize.
3774 (generate_entry_out_of_ssa_copies): Simplify.
3775 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3776 by passing in ISL dimension.
3777 (parameter_index_in_region_1): Rename to ...
3778 (parameter_index_in_region): ... this.
3779 (extract_affine): Adjust assert, pass down parameter index.
3780 (add_param_constraints): Use range-info when available.
3781 (build_scop_context): Adjust.
3782 * sese.c (new_sese_info): Adjust.
3783 (free_sese_info): Likewise.
3784 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3785 Remove unused typedefs.
3786 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3788 2017-10-18 Martin Liska <mliska@suse.cz>
3790 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3792 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3794 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3795 (s390_sched_init): Do not reset s390_sched_state if we entered the
3796 current basic block via a fallthru edge and all others are unlikely.
3798 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3800 * config/s390/s390.c (NUM_SIDES): New variable.
3801 (LONGRUNNING_THRESHOLD): New variable.
3802 (LATENCY_FACTOR): New variable.
3803 (s390_sched_score): Decrease score for long-running instructions on
3805 (s390_sched_variable_issue): Perform bookkeeping for long-running
3808 2017-10-18 Richard Biener <rguenther@suse.de>
3810 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3811 Simplify with removal of the parameter rename map.
3812 (set_rename): Likewise.
3813 (should_copy_to_new_region): Likewise.
3814 (graphite_copy_stmts_from_block): Likewise.
3815 (copy_bb_and_scalar_dependences): Remove initialization of
3816 unused copied_bb_map.
3818 (copy_internal_parameters): Likewise.
3819 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3820 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3821 Use INTEGRAL_TYPE_P.
3822 (parameter_index_in_region_1): Rename to ...
3823 (assign_parameter_index_in_region): ... this. Assert we have
3824 a parameter we handle.
3825 (scan_tree_for_params): Adjust.
3826 * sese.h (parameter_rename_map_t): Remove.
3827 (struct sese_info_t): Remove unused parameter_rename_map and
3828 copied_bb_map members.
3829 * sese.c (new_sese_info): Adjust.
3830 (free_sese_info): Likewise.
3832 2017-10-18 Martin Liska <mliska@suse.cz>
3835 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3836 on an abnormal edge.
3838 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3840 * doc/invoke.texi (ffunction-sections and fdata-sections):
3843 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3845 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3846 the use statement can throw internally.
3848 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3850 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3851 any RTX present on the RHS of a SET.
3852 * compare-elim.c (try_eliminate_compare): Restore comment.
3854 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3856 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3857 may be also called on tcc_type nodes.
3858 * langhooks.c (lhd_tree_size): Likewise.
3860 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3862 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3863 format_warning_at_substring.
3864 (maybe_warn): Convert source_range * param to a location_t. Pass
3865 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3866 (format_directive): Remove code to extract source_ranges and
3867 source_range * in favor of just a location_t.
3868 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3870 * substring-locations.c (format_warning_va): Convert
3871 source_range * param to a location_t.
3872 (format_warning_at_substring): Likewise.
3873 * substring-locations.h (format_warning_va): Likewise.
3874 (format_warning_at_substring): Likewise.
3876 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3878 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3880 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3881 and scatter/gather ops.
3883 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3884 vec_gather_load and vec_scatter_store.
3885 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3886 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3887 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3888 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3889 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3890 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3892 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3894 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3895 Detect FTST instruction and handle its register pops. Only pop
3896 second operand if can_pop_second_op is true.
3897 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3898 set can_pop_second_op to false in the compare_for_stack_reg call.
3900 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3901 output_fp_compare for stack register operands.
3902 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3903 instructions here. Do not emit stack register pops here. Assert
3904 that FCOMPP pops next to top stack register. Rewrite function.
3906 2017-10-17 Nathan Sidwell <nathan@acm.org>
3909 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3910 use DECL_ASSEMBLER_NAME_RAW.
3913 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3916 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3917 Wilco Dijkstra <wilco.dijkstra@arm.com>
3919 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3920 converstion to Pmode from the buf_addr.
3922 2017-10-17 Richard Biener <rguenther@suse.de>
3924 * graphite-dependences.c (scop_get_reads_and_writes): Change
3925 output parameters to references.
3927 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3929 PR 71026/tree-optimization
3930 * fold-const.c (distribute_real_division): Removed.
3931 (fold_binary_loc): Remove calls to distribute_real_divison.
3933 2017-10-17 Richard Biener <rguenther@suse.de>
3935 * graphite-scop-detection.c
3936 (scop_detection::stmt_has_simple_data_refs_p): Always use
3937 the full nest as region.
3938 (try_generate_gimple_bb): Likewise.
3939 * sese.c (scalar_evolution_in_region): Simplify now that
3940 SCEV can handle instantiation in regions.
3941 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3942 in the non-loop part of a function if requested.
3944 2017-10-17 Richard Biener <rguenther@suse.de>
3946 PR tree-optimization/82563
3947 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3949 (graphite_regenerate_ast_isl): Call it.
3950 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3952 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3954 PR tree-optimization/82549
3955 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3956 Formatting fixes. Instead of calling make_bit_field_ref with negative
3959 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3961 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3962 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3963 _mm_maskz_reduce_ss): New.
3964 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3965 __builtin_ia32_reducess_mask): Ditto..
3966 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3967 * config/i386/sse.md (reduces<mode>): Renamed to ...
3968 (reduces<mode><mask_scalar_name>): ... this.
3969 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3971 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3972 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3974 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3976 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3977 * selftest-run-tests.c (selftest::run_tests): Call
3978 selftest::unique_ptr_tests_cc_tests.
3979 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3980 * unique-ptr-tests.cc: New file.
3982 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3985 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3987 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3988 (make_hard_regno_born, make_hard_regno_dead): Update
3989 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3991 2017-10-16 Jeff Law <law@redhat.com>
3993 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3995 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3997 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3999 2017-10-16 Olivier Hainque <hainque@adacore.com>
4001 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4002 with_cpu if we were configured for an e500v2 target cpu name.
4004 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4006 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4007 * doc/invoke.texi: Document +nodsp as a valid extension for
4010 2017-10-16 Martin Liska <mliska@suse.cz>
4012 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4013 (test_set_range): Likewise.
4014 (test_range_functions): Rename to ...
4015 (test_bit_in_range): ... this.
4016 (sbitmap_c_tests): Add new test.
4018 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4020 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4022 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4023 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4025 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4027 * config/aarch64/aarch64-builtins.c
4028 (aarch64_types_quadopu_lane_qualifiers): New.
4029 (TYPES_QUADOPU_LANE): New.
4030 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4031 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4032 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4033 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4034 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4035 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4036 (Vdottype, DOTPROD): New.
4037 (sur): Add SDOT and UDOT.
4039 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4041 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4042 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4043 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4045 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4046 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4047 Enable TARGET_DOTPROD.
4048 (cortex-a75.cortex-a55): Likewise.
4049 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4051 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4053 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4054 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4056 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4058 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4059 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4060 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4061 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4062 * config/arm/types.md (neon_dot, neon_dot_q): New.
4063 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4065 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4067 * config/arm/arm.h (TARGET_DOTPROD): New.
4068 * config/arm/arm.c (arm_arch_dotprod): New.
4069 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4070 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4071 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4072 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4073 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4074 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4075 * doc/invoke.texi (armv8.2-a): Document dotprod
4077 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4079 * i386.c (ix86_vec_cost): New function.
4080 (ix86_rtx_costs): Handle vector operations better.
4081 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4082 * x86-tune-costs.h: Add new costs to all tables.
4084 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4086 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4088 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4089 divsd, sqrtss and sqrtsd
4090 * x86-tune-costs.h: Add new entries to all costs.
4091 (znver1_cost): Fix to match real instruction latencies.
4093 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4094 Michael Collison <michael.collison@arm.com>
4096 * compare-elim.c: Include emit-rtl.h.
4097 (can_merge_compare_into_arith): New function.
4098 (try_validate_parallel): Likewise.
4099 (try_merge_compare): Likewise.
4100 (try_eliminate_compare): Call the above when no previous clobber
4102 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4105 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4109 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4110 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4114 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4115 to be any operand_equal_p operands. For & (B - 1) require
4116 B to be power of 2. Recognize
4117 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4119 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4122 * optabs.c (expand_memory_blockage): Fix call of
4123 targetm.have_memory_blockage.
4125 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4128 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4129 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4130 objects to extra_objs instead of overwriting it.
4132 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4134 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4135 Use any_fp_register_operand as operand[3] predicate. Simplify
4136 equality test for operands[2] and operands[4] memory location.
4137 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4138 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4139 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4140 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4141 any_fp_register_operand as operand[1] predicate. Simplify
4142 equality test for operands[0] and operands[3] memory location.
4143 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4144 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4145 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4147 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4149 * target-insns.def: Add memory_blockage.
4150 * optabs.c (expand_memory_blockage): New function.
4151 (expand_asm_memory_barrier): Rename ...
4152 (expand_asm_memory_blockage): ... to this.
4153 (expand_mem_thread_fence): Call expand_memory_blockage
4154 instead of expand_asm_memory_barrier.
4155 (expand_mem_singnal_fence): Ditto.
4156 (expand_atomic_load): Ditto.
4157 (expand_atomic_store): Ditto.
4158 * doc/md.texi (Standard Pattern Names For Generation):
4159 Document memory_blockage instruction pattern.
4161 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4163 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4164 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4165 * config/rl78/rl78.md: New define_expand "adddi3".
4167 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4169 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4172 2017-10-13 Jeff Law <law@redhat.com>
4174 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4176 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4179 * internal-fn.c (expand_mul_overflow): If both operands have
4180 the same highpart of -1 or 0 and the topmost bit of lowpart
4181 is different, overflow is if res <= 0 rather than res < 0.
4183 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4185 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4186 TARGET_P9_VECTOR code for unaligned_load case.
4188 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4190 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4192 2017-10-13 Nathan Sidwell <nathan@acm.org>
4194 * tree-core.h (tree_contains_struct): Make bool.
4195 * tree.c (tree_contains_struct): Likewise.
4196 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4197 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4198 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4199 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4200 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4202 2017-10-13 Richard Biener <rguenther@suse.de>
4204 * graphite-isl-ast-to-gimple.c
4205 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4206 parameters and dominance check.
4207 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4208 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4209 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4210 Do not update SSA form here or do intermediate IL verification.
4211 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4212 (graphite_initialize): Remove check on the number of loops in
4213 the function and inline into graphite_transform_loops.
4214 (graphite_finalize): Inline into graphite_transform_loops.
4215 (graphite_transform_loops): Perform SSA update and IL verification
4217 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4219 2017-10-13 Richard Biener <rguenther@suse.de>
4221 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4222 graphite_expression_type_precision): Avoid global constructor
4224 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4225 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4226 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4227 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4228 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4229 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4231 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4234 * config/i386/i386.h (ix86_red_zone_size): New.
4235 * config/i386/i386.md (push peephole2s): Replace
4236 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4238 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4239 Alan Hayward <alan.hayward@arm.com>
4240 David Sherwood <david.sherwood@arm.com>
4242 * combine.c (can_change_dest_mode): Reject changes in
4243 REGMODE_NATURAL_SIZE.
4245 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4246 Alan Hayward <alan.hayward@arm.com>
4247 David Sherwood <david.sherwood@arm.com>
4249 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4250 (expand_debug_source_expr): Likewise.
4251 * combine.c (combine_simplify_rtx): Likewise.
4252 * cse.c (fold_rtx): Likewise.
4253 * fwprop.c (canonicalize_address): Likewise.
4254 * targhooks.c (default_shift_truncation_mask): Likewise.
4256 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4257 Alan Hayward <alan.hayward@arm.com>
4258 David Sherwood <david.sherwood@arm.com>
4260 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4261 (widened_mode): Likewise.
4262 (expand_unop): Likewise.
4263 * ree.c (transform_ifelse): Likewise.
4264 (merge_def_and_ext): Likewise.
4265 (combine_reaching_defs): Likewise.
4266 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4268 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4269 Alan Hayward <alan.hayward@arm.com>
4270 David Sherwood <david.sherwood@arm.com>
4272 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4273 * combine.c (gen_lowpart_for_combine): Likewise.
4274 * dwarf2out.c (rtl_for_decl_location): Likewise.
4275 * final.c (alter_subreg): Likewise.
4276 * rtlhooks.c (gen_lowpart_general): Likewise.
4277 (gen_lowpart_if_possible): Likewise.
4279 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4280 Alan Hayward <alan.hayward@arm.com>
4281 David Sherwood <david.sherwood@arm.com>
4283 * calls.c (expand_call): Use subreg_lowpart_offset.
4284 * cse.c (cse_insn): Likewise.
4285 * regcprop.c (copy_value): Likewise.
4286 (copyprop_hardreg_forward_1): Likewise.
4288 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4291 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4292 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4293 =Q constraints to +Q and into insn condition add check
4294 that operands[0] and operands[1] are equal.
4295 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4296 =Q constraints to +Q and into insn condition add check
4297 that operands[0] is equal to either operands[1] or operands[2].
4300 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4301 instead of handling MINUS_EXPR twice (once for each argument),
4302 canonicalize operand order and handle just once, use rtype where
4303 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4306 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4307 any values of __C while still being pattern recognizable as a simple
4310 2017-10-13 Richard Biener <rguenther@suse.de>
4312 PR tree-optimization/82451
4314 2017-10-02 Richard Biener <rguenther@suse.de>
4316 PR tree-optimization/82355
4317 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4318 a mapping for the enclosing loop but avoid generating one for
4320 (copy_bb_and_scalar_dependences): Remove premature codegen
4321 error on PHIs in blocks duplicated into multiple places.
4322 * graphite-scop-detection.c
4323 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4324 in the region use it as loop and nest to analyze the DR in.
4325 (try_generate_gimple_bb): Likewise.
4326 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4327 (add_loop_constraints): For blocks in a loop not in the region
4328 create a dimension with a single iteration.
4329 * sese.h (gbb_loop_at_index): Remove assert.
4331 * cfgloop.c (loop_preheader_edge): For the loop tree root
4332 return the single successor of the entry block.
4333 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4334 Reset the SCEV hashtable and niters.
4335 * graphite-scop-detection.c
4336 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4337 assert that we only have POLYNOMIAL_CHREC that vary in loops
4338 contained in the region.
4339 (scop_detection::graphite_can_represent_expr): Adjust.
4340 (scop_detection::stmt_has_simple_data_refs_p): For loops
4341 not in the region set loop to NULL. The nest is now the
4342 entry edge to the region.
4343 (try_generate_gimple_bb): Likewise.
4344 * sese.c (scalar_evolution_in_region): Adjust for
4345 instantiate_scev change.
4346 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4347 Make nest parameter the edge into the region.
4348 (create_data_ref): Likewise.
4349 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4350 entry edge into a region and adjust instantiate_scev calls.
4351 (create_data_ref): Likewise.
4352 (graphite_find_data_references_in_stmt): Likewise.
4353 (find_data_references_in_stmt): Pass the loop preheader edge
4354 from the nest argument.
4355 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4356 parameter the edge into the region.
4357 (instantiate_parameters): Use the loop preheader edge as entry.
4358 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4360 (get_instantiated_value_entry): Make instantiate_below parameter
4361 the edge into the region.
4362 (instantiate_scev_name): Likewise. Adjust dominance checks,
4363 when we cannot use loop-based instantiation instantiate by
4364 walking use-def chains.
4365 (instantiate_scev_poly): Adjust.
4366 (instantiate_scev_binary): Likewise.
4367 (instantiate_scev_convert): Likewise.
4368 (instantiate_scev_not): Likewise.
4369 (instantiate_array_ref): Remove.
4370 (instantiate_scev_3): Likewise.
4371 (instantiate_scev_2): Likewise.
4372 (instantiate_scev_1): Likewise.
4373 (instantiate_scev_r): Do not blindly handle N-operand trees.
4374 Do not instantiate array-refs. Handle all constants and invariants.
4375 (instantiate_scev): Make instantiate_below parameter
4376 the edge into the region.
4377 (resolve_mixers): Use the loop preheader edge for the region
4378 parameter to instantiate_scev_r.
4379 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4381 2017-10-13 Richard Biener <rguenther@suse.de>
4383 PR tree-optimization/82525
4384 * graphite-isl-ast-to-gimple.c
4385 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4387 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4388 Fail code generation when we cannot represent the isl integer.
4389 (binary_op_to_tree): Elide modulo operations that are no-ops
4390 in the type we code generate. Remove now superfluous code
4393 2017-10-13 Richard Biener <rguenther@suse.de>
4395 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4396 (scop_detection::harmful_loop_in_region): Remove premature
4397 IV type restriction.
4398 (scop_detection::graphite_can_represent_scev): We can handle
4399 pointer IVs just fine.
4401 2017-10-13 Alan Modra <amodra@gmail.com>
4403 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4404 "Clobbers and Scratch Registers". Add paragraph on
4405 alternative to clobbers for scratch registers and OpenBLAS
4408 2017-10-13 Alan Modra <amodra@gmail.com>
4410 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4411 example of a memory input for a string of known length. Move
4412 commentary out of table. Add a number of new examples
4413 covering array memory inputs.
4415 2017-10-12 Martin Liska <mliska@suse.cz>
4417 PR tree-optimization/82493
4418 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4419 (test_range_functions): New function.
4420 (sbitmap_c_tests): Likewise.
4421 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4422 * selftest.h (sbitmap_c_tests): New function.
4424 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4426 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4428 * config/rs6000/amo.h: Fix spacing issue.
4430 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4433 * config/i386/i386.md (*ashl<mode>3_mask_1,
4434 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4435 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4438 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4440 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4441 (profile_probability): Set max_probability
4442 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4444 * profile-count.c (profile_probability::differs_from_p): Do not
4445 rely on max_probaiblity == 10000
4447 2017-10-12 Jeff Law <law@redhat.com>
4449 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4452 2017-10-12 Martin Sebor <msebor@redhat.com>
4456 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4457 (handle_alias_pairs): Call it.
4458 * common.opt (-Wattribute-alias): New option.
4459 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4460 * doc/invoke.texi (-Wattribute-alias): Document.
4462 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4465 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4467 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4469 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4470 (make_hard_regno_born, make_hard_regno_dead): Update
4471 bb_killed_pseudos and bb_gen_pseudos.
4473 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4475 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4477 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4479 * config/alpha/alpha.c (alpha_split_conditional_move):
4480 Use std::swap instead of manually swapping.
4481 (alpha_stdarg_optimize_hook): Ditto.
4482 (alpha_canonicalize_comparison): Ditto.
4484 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4486 * tree-loop-distribution.c (struct builtin_info): New struct.
4487 (struct partition): Refactor fields into struct builtin_info.
4488 (partition_free): Free struct builtin_info.
4489 (build_size_arg_loc, build_addr_arg_loc): Delete.
4490 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4491 information from struct builtin_info.
4492 (find_single_drs): New function refactored from classify_partition.
4493 Also moved builtin validity checks to this function.
4494 (compute_access_range, alloc_builtin): New functions.
4495 (classify_builtin_st, classify_builtin_ldst): New functions.
4496 (classify_partition): Refactor code into functions find_single_drs,
4497 classify_builtin_st and classify_builtin_ldst.
4498 (distribute_loop): Don't do runtime alias check when distributing
4500 (find_seed_stmts_for_distribution): New function.
4501 (pass_loop_distribution::execute): Refactor code finding seed
4502 stmts into above function. Support distribution for the innermost
4503 two-level loop nest. Adjust dump information.
4505 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4507 * tree-loop-distribution.c: Adjust the general comment.
4508 (NUM_PARTITION_THRESHOLD): New macro.
4509 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4510 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4511 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4512 in call to build_partition_graph.
4513 (finalize_partitions): New parameter. Make loop distribution more
4514 conservative by fusing more partitions.
4515 (distribute_loop): Don't do runtime alias check in case of loop nest
4517 (find_seed_stmts_for_distribution): New function.
4518 (prepare_perfect_loop_nest): New function.
4519 (pass_loop_distribution::execute): Refactor code finding seed stmts
4520 and loop nest into above functions. Support loop nest distribution.
4521 Adjust dump information accordingly.
4523 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4525 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4526 and set PTYPE_SEQUENTIAL for merged partition.
4528 2017-10-12 Richard Biener <rguenther@suse.de>
4530 PR tree-optimization/69728
4532 2017-09-19 Richard Biener <rguenther@suse.de>
4534 PR tree-optimization/69728
4535 * graphite-sese-to-poly.c (schedule_error): New global.
4536 (add_loop_schedule): Handle empty domain by failing the
4538 (build_original_schedule): Handle schedule_error.
4540 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4541 domain by returning an unchanged schedule.
4543 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4545 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4546 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4548 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4550 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4553 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4556 * expr.c (store_field): Don't optimize away bitsize == 0 store
4557 from CALL_EXPR with addressable return type.
4559 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4561 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4562 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4563 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4564 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4565 TARGET_ISEL instead of TARGET_ISEL<sel>.
4567 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4569 * config/rs6000/rs6000.c
4570 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4572 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4574 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4576 (reg_or_cint_operand): Fix comment.
4577 (reg_or_zero_operand): New predicate.
4578 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4579 * config/rs6000/rs6000.c (output_isel): Delete.
4580 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4581 instead of reg_or_cint_operand. Output instruction directly (not via
4583 (isel_unsigned_<mode>): Ditto.
4584 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4585 gpc_reg_operand. Add an instruction alternative for this. Output
4586 instruction directly.
4587 (*isel_reversed_unsigned_<mode>): Ditto.
4589 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4591 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4592 (TARGET_CANONICALIZE_COMPARISON): Define.
4594 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4597 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4598 Check whether the dest is REG before adding REG_EQUIV note.
4600 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4603 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4605 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4606 (make_hard_regno_born, make_hard_regno_dead): Update
4607 bb_killed_pseudos and bb_gen_pseudos.
4609 2017-10-11 Nathan Sidwell <nathan@acm.org>
4611 * incpath.h (enum incpath_kind): Name enum, prefix values.
4612 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4613 * incpath.c (heads, tails): Use INC_MAX.
4614 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4615 (merge_include_chains, split_quote_chain,
4616 register_include_chains): Update incpath_kind names.
4617 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4618 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4620 (add_framework_path, darwin_register_objc_includes): Likewise.
4621 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4623 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4625 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4626 Do not use float_operator operator predicate.
4627 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4628 * config/i386/predicates.md (float_operator): Remove predicate.
4630 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4632 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4633 (*jccxf_i387): Ditto.
4634 (*jcc<mode>_i387): Ditto.
4635 (*jccu<mode>_i387): Ditto.
4636 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4637 (*jcc_*_i387 splitters): Remove.
4638 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4639 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4640 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4643 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4645 * profile-count.h (slow_safe_scale_64bit): New function.
4646 (safe_scale_64bit): New inline.
4647 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4648 * profile-count.c: Include wide-int.h
4649 (slow_safe_scale_64bit): New.
4651 2017-10-11 Nathan Sidwell <nathan@acm.org>
4653 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4654 HAS_DECL_ASSEMBLER_NAME_P.
4655 * gimple-expr.c (gimple_decl_printable_name: Check
4656 HAS_DECL_ASSEMBLER_NAME_P too.
4657 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4658 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4659 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4660 * passes.c (rest_of_decl_compilation): Check
4661 HAS_DECL_ASSEMBLER_NAME_P too.
4662 * recog.c (verify_changes): Likewise.
4663 * tree-pretty-print.c (dump_decl_name): Likewise.
4664 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4666 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4667 (SET_DECL_ASSEMBLER_NAME): Use it.
4668 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4669 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4670 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4672 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4674 * config.gcc (i386, x86_64): Add extra objects.
4675 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4676 (ix86_min_insn_size): Declare.
4677 (ix86_issue_rate): Declare.
4678 (ix86_adjust_cost): Declare.
4679 (ia32_multipass_dfa_lookahead): Declare.
4680 (ix86_macro_fusion_p): Declare.
4681 (ix86_macro_fusion_pair_p): Declare.
4682 (ix86_bd_has_dispatch): Declare.
4683 (ix86_bd_do_dispatch): Declare.
4684 (ix86_core2i7_init_hooks): Declare.
4685 (ix86_atom_sched_reorder): Declare.
4686 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4687 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4688 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4689 (rip_relative_addr_p): Rename to ...
4690 (ix86_rip_relative_addr_p): ... this one; export.
4691 (memory_address_length): Update.
4692 (ix86_issue_rate): Move to x86-tune-sched.c.
4693 (ix86_flags_dependent): Move to x86-tune-sched.c.
4694 (ix86_agi_dependent): Move to x86-tune-sched.c.
4695 (exact_dependency_1): Move to x86-tune-sched.c.
4696 (exact_store_load_dependency): Move to x86-tune-sched.c.
4697 (ix86_adjust_cost): Move to x86-tune-sched.c.
4698 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4699 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4700 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4701 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4702 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4703 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4704 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4705 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4706 (min_insn_size): Rename to ...
4707 (ix86_min_insn_size): ... this one; export.
4708 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4709 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4710 (core2i7_first_cycle_multipass_backtrack): Move to
4711 x86-tune-sched-core.c.
4712 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4713 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4714 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4715 (ix86_avoid_jump_mispredicts): Update.
4716 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4717 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4718 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4719 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4720 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4721 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4722 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4723 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4724 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4725 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4726 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4727 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4728 (BIG): Move to ix86-tune-sched-bd.c.
4729 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4730 (enum insn_path): Move to ix86-tune-sched-bd.c.
4731 (get_mem_group): Move to ix86-tune-sched-bd.c.
4732 (is_cmp): Move to ix86-tune-sched-bd.c.
4733 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4734 (is_branch): Move to ix86-tune-sched-bd.c.
4735 (is_prefetch): Move to ix86-tune-sched-bd.c.
4736 (init_window): Move to ix86-tune-sched-bd.c.
4737 (allocate_window): Move to ix86-tune-sched-bd.c.
4738 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4739 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4740 (process_end_window): Move to ix86-tune-sched-bd.c.
4741 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4742 (find_constant): Move to ix86-tune-sched-bd.c.
4743 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4744 (has_immediate): Move to ix86-tune-sched-bd.c.
4745 (get_insn_path): Move to ix86-tune-sched-bd.c.
4746 (get_insn_group): Move to ix86-tune-sched-bd.c.
4747 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4748 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4749 (add_insn_window): Move to ix86-tune-sched-bd.c.
4750 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4751 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4752 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4753 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4754 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4755 (do_dispatch): Move to ix86-tune-sched-bd.c.
4756 (has_dispatch): Move to ix86-tune-sched-bd.c.
4757 * config/i386/t-i386: Add new object files.
4758 * config/i386/x86-tune-costs.h: New file.
4759 * config/i386/x86-tune-sched-atom.c: New file.
4760 * config/i386/x86-tune-sched-bd.c: New file.
4761 * config/i386/x86-tune-sched-core.c: New file.
4762 * config/i386/x86-tune-sched.c: New file.
4764 2017-10-11 Liu Hao <lh_mouse@126.com>
4766 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4767 the generic version below instead.
4768 (should_colorize): Recognize Windows consoles as terminals
4770 * pretty-print.c [__MINGW32__] (write_all): New function.
4771 [__MINGW32__] (find_esc_head): Likewise.
4772 [__MINGW32__] (find_esc_terminator): Likewise.
4773 [__MINGW32__] (eat_esc_sequence): Likewise.
4774 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4776 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4779 2017-10-11 Richard Biener <rguenther@suse.de>
4781 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4782 Properly call analyze_scalar_evolution with the loop of the stmt.
4784 2017-10-11 Richard Biener <rguenther@suse.de>
4786 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4787 * tree-core.h (tree_base): Add chrec_var union member.
4788 * tree.h (CHREC_VAR): Remove.
4789 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4790 * tree-chrec.h (build_polynomial_chrec): Adjust.
4791 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4792 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4794 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4796 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4797 * match.pd: ... here.
4798 ((T) X == (T) Y): Relax condition.
4800 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4802 PR tree-optimization/82472
4803 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4805 (break_alias_scc_partitions): Update postorder number.
4807 2017-10-11 Martin Liska <mliska@suse.cz>
4810 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4812 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4814 2017-10-11 Martin Liska <mliska@suse.cz>
4816 * print-rtl.c (print_insn): Move declaration of idbuf
4817 to same scope as name.
4819 2017-10-11 Martin Liska <mliska@suse.cz>
4824 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4827 2017-10-11 Martin Liska <mliska@suse.cz>
4830 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4833 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4835 * config.gcc (default_gnu_indirect_function): Default to yes for
4836 arm*-*-linux* with glibc.
4838 2017-10-11 Richard Biener <rguenther@suse.de>
4840 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4841 default-defs and types we do not want to analyze.
4842 (interpret_loop_phi): Replace unreachable code with an assert.
4843 (compute_scalar_evolution_in_loop): Remove and inline ...
4844 (analyze_scalar_evolution_1): ... here, replacing condition with
4845 what makes the intent clearer. Remove handling of cases
4846 get_scalar_evolution now handles.
4848 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4850 PR rtl-optimization/81434
4851 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4852 comment for main loop. In sched_group_found if, also add checks for
4853 pass and min_cost_group.
4855 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4857 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4858 (rs6000_insn_cost): New function.
4859 * config/rs6000/rs6000.md (cost): New attribute.
4861 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4862 H.J. Lu <hongjiu.lu@intel.com>
4866 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4867 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4868 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4869 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4870 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4871 OPTION_MASK_ISA_MMX and some other bit set require both
4872 mmx and the ISAs without the mmx bit.
4873 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4874 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4875 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4876 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4877 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4878 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4879 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4880 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4881 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4882 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4883 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4884 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4885 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4886 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4887 Add OPTION_MASK_ISA_MMX.
4889 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4891 * config.gcc (armv7*-*-freebsd*): New target.
4892 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4894 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4896 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4897 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4900 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4902 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4903 a template parameter.
4904 (WIDE_INT_REF_FOR): Update accordingly.
4905 * tree.h (wi::int_traits <const_tree>): Delete.
4906 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4907 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4908 (wi::tree_to_wide_ref): New typedef.
4909 (wi::to_wide): New function.
4910 * calls.c (get_size_range): Use wi::to_wide when operating on
4912 * cgraph.c (cgraph_node::create_thunk): Likewise.
4913 * config/i386/i386.c (ix86_data_alignment): Likewise.
4914 (ix86_local_alignment): Likewise.
4915 * dbxout.c (stabstr_O): Likewise.
4916 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4917 * expr.c (const_vector_from_tree): Likewise.
4918 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4919 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4920 (fold_negate_expr_1, int_const_binop_1, const_binop)
4921 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4922 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4923 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4924 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4925 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4926 (fold_not_const, round_up_loc): Likewise.
4927 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4928 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4929 (alloca_call_type): Likewise.
4930 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4931 * godump.c (go_output_typedef): Likewise.
4932 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4933 * internal-fn.c (get_min_precision): Likewise.
4934 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4935 * ipa-polymorphic-call.c
4936 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4937 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4938 (ipa_modify_call_arguments): Likewise.
4939 * match.pd: Likewise.
4940 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4941 * print-tree.c (print_node_brief, print_node): Likewise.
4942 * stmt.c (expand_case): Likewise.
4943 * stor-layout.c (layout_type): Likewise.
4944 * tree-affine.c (tree_to_aff_combination): Likewise.
4945 * tree-cfg.c (group_case_labels_stmt): Likewise.
4946 * tree-data-ref.c (dr_analyze_indices): Likewise.
4947 (prune_runtime_alias_test_list): Likewise.
4948 * tree-dump.c (dequeue_and_dump): Likewise.
4949 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4950 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4951 * tree-pretty-print.c (dump_generic_node): Likewise.
4952 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4953 (simple_iv_with_niters): Likewise.
4954 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4955 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4956 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4957 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4958 (refine_value_range_using_guard, number_of_iterations_ne_max)
4959 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4960 (get_cst_init_from_scev, record_nonwrapping_iv)
4961 (scev_var_range_cant_overflow): Likewise.
4962 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4963 * tree-ssa-pre.c (compute_avail): Likewise.
4964 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4965 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4966 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4967 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4968 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4969 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4970 (dump_case_nodes, try_switch_expansion): Likewise.
4971 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4972 (vect_do_peeling): Likewise.
4973 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4974 * tree-vect-stmts.c (vectorizable_load): Likewise.
4975 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4976 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4977 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4978 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4979 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4980 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4981 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4982 (evrp_dom_walker::before_dom_children): Likewise.
4983 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4984 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4985 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4986 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4987 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4988 (get_range_pos_neg): Likewise.
4989 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4990 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4991 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4992 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4993 * config/avr/avr.c (avr_fold_builtin): Likewise.
4994 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4995 * config/msp430/msp430.c (msp430_attr): Likewise.
4996 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4997 * config/powerpcspe/powerpcspe-c.c
4998 (altivec_resolve_overloaded_builtin): Likewise.
4999 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5000 (rs6000_expand_ternop_builtin): Likewise.
5001 * config/rs6000/rs6000-c.c
5002 (altivec_resolve_overloaded_builtin): Likewise.
5003 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5004 (rs6000_expand_ternop_builtin): Likewise.
5005 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5007 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5009 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5010 when copying loop nest with only one inner loop.
5012 2017-10-10 Richard Biener <rguenther@suse.de>
5014 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5015 blocks if SCEV is active.
5016 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5018 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5019 (scev_initialize): Assert we are not yet initialized.
5021 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5023 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5024 inner loop's exit stmt by making it always exit the loop, otherwise
5025 we would generate an infinite empty loop.
5027 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5029 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5030 renaming variables in new preheader if it's deleted.
5032 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5034 * tree-loop-distribution.c (struct partition): Remove unused field
5035 loops of the structure.
5036 (partition_alloc, partition_free): Ditto.
5037 (build_rdg_partition_for_vertex): Ditto.
5039 2017-10-09 Jeff Law <law@redhat.com>
5041 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5042 return type to match prototype and documentation.
5044 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5046 * config/rs6000/rs6000.c (processor_costs): Move to ...
5047 * config/rs6000/rs6000.h: ... here.
5048 (rs6000_cost): Declare.
5050 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5052 * except.c (setjmp_fn): New global variable.
5053 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5054 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5055 if DONT_USE_BUILTIN_SETJMP is defined.
5057 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5059 * target.def (insn_cost): New hook.
5060 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5061 * doc/tm.texi: Regenerate.
5062 * rtlanal.c (insn_cost): Use the new hook.
5064 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5066 * combine.c (combine_validate_cost): Compute the new insn_cost,
5067 not just pattern_cost.
5068 (try_combine): Adjust comment.
5070 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5072 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5074 * combine.c (uid_insn_cost): Adjust comment.
5075 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5077 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5078 * dse.c (find_shift_sequence): Ditto.
5079 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5080 (bb_valid_for_noce_process_p): Use pattern_cost.
5081 * rtl.h (insn_rtx_cost): Delete.
5082 (pattern_cost): New prototype.
5083 (insn_cost): New prototype.
5084 * rtlanal.c (insn_rtx_cost): Rename to...
5085 (pattern_cost): ... this.
5088 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5090 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5091 (*jcc<mode>_0_r_i387): Ditto.
5092 (*jccxf_r_i387): Ditto.
5093 (*jcc<mode>_r_i387): Ditto.
5094 (*jccu<mode>_r_i387): Ditto.
5095 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5096 (*jcc): Rename from *jcc_1.
5098 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5100 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5101 deferred rescans after the lvx/stvx recombination pre-pass.
5103 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5105 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5106 memory operation instruction support.
5107 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5108 (rs6000-ibm-aix[789]*): Likewise.
5109 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5110 Document new functions.
5112 2017-10-09 Richard Biener <rguenther@suse.de>
5114 PR tree-optimization/82397
5115 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5116 equality only for semantically equal trees.
5118 2017-10-09 Richard Biener <rguenther@suse.de>
5120 PR tree-optimization/82449
5121 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5122 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5123 allow constant addresses.
5124 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5127 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5129 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5132 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5135 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5138 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5141 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5143 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5146 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5147 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5149 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5151 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5152 (WI_BINARY_PREDICATE_RESULT): Likewise.
5153 (wi::binary_traits::operator_result): New type.
5154 (wi::binary_traits::predicate_result): Likewise.
5155 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5156 (generic_wide_int::operator==, generic_wide_int::operator!=)
5157 (generic_wide_int::operator&, generic_wide_int::and_not)
5158 (generic_wide_int::operator|, generic_wide_int::or_not)
5159 (generic_wide_int::operator^, generic_wide_int::operator+
5160 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5162 (operator~, unary operator-, operator==, operator!=, operator&)
5163 (operator|, operator^, operator+, binary operator-, operator*): New
5165 * expr.c (get_inner_reference): Use wi::bit_and_not.
5166 * fold-const.c (fold_binary_loc): Likewise.
5167 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5168 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5169 (bit_value_binop): Likewise.
5170 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5171 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5172 (extract_range_from_binary_expr_1): Likewise.
5173 (masked_increment): Likewise.
5174 (simplify_bit_ops_using_ranges): Likewise.
5176 2017-10-09 Martin Jambor <mjambor@suse.cz>
5179 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5180 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5181 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5183 (hsa_fixup_mov_insn_type): New function.
5184 (hsa_op_with_type::get_in_type): Use it.
5185 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5186 immediates in an assert.
5187 (hsa_op_with_type::extend_int_to_32bit): New method.
5188 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5189 types. Convert to dest type if necessary.
5190 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5191 (reg_for_gimple_ssa): Pass false as min32int to
5192 hsa_type_for_scalar_tree_type.
5193 (gen_hsa_addr): Fixup type when creating addresable temporary.
5194 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5195 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5196 necessary. Call hsa_fixup_mov_insn_type.
5197 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5198 extend operands and convert to dest type if necessary.
5199 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5200 to dest type if necessary.
5201 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5202 if conversion nt necessary and size matches.
5203 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5204 to dest type if necessary.
5205 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5206 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5208 (gen_hsa_clrsb): Likewise.
5209 (gen_hsa_ffs): Likewise.
5210 (gen_hsa_divmod): Extend operands and convert to dest type if
5212 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5214 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5216 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5217 Remove empty default arguments. Use a brace block as output
5219 (conditional return): Ditto.
5221 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5222 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5224 (group_ending_nop): Ditto.
5225 (doloop_end): Ditto.
5226 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5227 (splitters for those): Ditto.
5229 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5231 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5232 a conditional jump (and the compare for it) so that pc_rtx is the
5234 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5235 for the deleted and renamed ctr<mode>_internal[234] patterns.
5236 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5237 Delete second conditional return pattern.
5238 (ctr<mode>_internal2): Delete this second bdnz pattern.
5239 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5240 (ctr<mode>_internal4): Delete this second bdz pattern.
5242 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5244 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5245 (always_initialized_rtx_for_ssa_name_p): New predicate.
5246 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5247 (finish_out_of_ssa): Free new field of SA.
5248 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5249 * tree-ssa-coalesce.c: Include tree-ssa.h.
5250 (get_parm_default_def_partitions): Remove extern keyword.
5251 (get_undefined_value_partitions): New function.
5252 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5253 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5255 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5257 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5259 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5261 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5263 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5264 for targets that preffer 128bit.
5266 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5268 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5270 2017-10-08 Olivier Hainque <hainque@adacore.com>
5272 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5273 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5275 (thumb_set_return_address): Likewise.
5277 2017-10-08 Olivier Hainque <hainque@adacore.com>
5279 * common/config/arm/arm-common.c (arm_except_unwind_info):
5280 Handle DWARF2_UNWIND_INFO.
5282 2017-10-07 Michael Collison <michael.collison@arm.com>
5284 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5287 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5289 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5290 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5291 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5292 defined, force the creation of a new block for a dispatch label.
5294 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5296 * invoke.texi (Wsuggest-attribute=cold): Document.
5297 * common.opt (Wsuggest-attribute=cold): New
5298 * ipa-pure-const.c (warn_function_cold): New function.
5299 * predict.c (compute_function_frequency): Use it.
5300 * predict.h (warn_function_cold): Declare.
5302 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5304 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5307 2017-10-06 Martin Liska <mliska@suse.cz>
5309 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5310 keyword for member functions.
5311 (struct sanopt_tree_couple): New struct.
5312 (struct sanopt_tree_couple_hash): New function.
5313 (struct sanopt_ctx): Add new hash_map.
5314 (has_dominating_ubsan_ptr_check): New function.
5315 (record_ubsan_ptr_check_stmt): Likewise.
5316 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5317 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5318 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5320 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5323 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5324 aarch64_simd_valid_immediate on CONST_VECTORs.
5325 (aarch64_reg_or_bic_imm): Likewise.
5327 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5329 PR rtl-optimization/82396
5330 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5332 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5334 * graphite-dependences.c (scop_get_reads): Move code to...
5335 (scop_get_must_writes): Move code to...
5336 (scop_get_may_writes): Move code to...
5337 (scop_get_reads_and_writes): ... here.
5338 (scop_get_dependences): Call scop_get_reads_and_writes.
5340 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5342 PR tree-optimization/82434
5343 * fold-const.h (can_native_encode_type_p,
5344 can_native_encode_string_p): Remove.
5345 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5346 don't encode anything, just return what would be otherwise returned.
5347 (native_encode_fixed, native_encode_complex, native_encode_vector):
5349 (native_encode_string): Likewise. Inline by hand
5350 can_native_encode_string_p.
5351 (can_native_encode_type_p): Remove.
5352 (can_native_encode_string_p): Remove.
5353 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5354 STRING_CSTs using can_native_encode_string_p, test all
5355 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5356 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5357 argument from native_encode_expr.
5358 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5359 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5362 2017-10-06 Richard Biener <rguenther@suse.de>
5364 PR tree-optimization/82397
5365 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5366 operand_equal_p but rely on data_ref_compare_tree for detecting
5368 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5369 to match up with dr_group_sort_cmp.
5371 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5374 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5376 * config/s390/s390-builtin-types.def: Regenerate.
5378 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5381 * config/s390/s390-builtin-types.def: Regenerate.
5382 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5383 Change flag from B_VXE to B_VX.
5384 (s390_vec_min_dbl): Remove B_VXE flag.
5386 2017-10-06 Richard Biener <rguenther@suse.de>
5388 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5389 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5390 translate_isl_ast_to_gimple::is_valid_rename,
5391 translate_isl_ast_to_gimple::get_rename,
5392 translate_isl_ast_to_gimple::get_def_bb_for_const,
5393 translate_isl_ast_to_gimple::get_new_name,
5394 translate_isl_ast_to_gimple::collect_all_ssa_names,
5395 translate_isl_ast_to_gimple::copy_loop_phi_args,
5396 translate_isl_ast_to_gimple::collect_all_ssa_names,
5397 translate_isl_ast_to_gimple::copy_loop_phi_args,
5398 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5399 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5400 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5401 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5402 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5403 translate_isl_ast_to_gimple::copy_cond_phi_args,
5404 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5405 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5406 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5407 translate_isl_ast_to_gimple::rename_uses,
5408 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5409 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5410 (set_rename_for_each_def): Likewise.
5411 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5412 here. Handle rewriting SCEV analyzable uses here.
5413 (copy_bb_and_scalar_dependences): Generate code for PHI
5415 (graphite_regenerate_ast_isl): Adjust.
5416 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5417 (add_write, add_read): New functions.
5418 (build_cross_bb_scalars_def): Use it and simplify.
5419 (build_cross_bb_scalars_use): Likewise.
5420 (graphite_find_cross_bb_scalar_vars): Inline into...
5421 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5422 simulating out-of-SSA. Compute liveout and add dependencies.
5423 (build_scops): Force an empty entry block.
5424 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5426 (sese_build_liveouts): Declare.
5427 (sese_trivially_empty_bb_p): Likewise.
5428 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5429 compute liveout and debug_liveout.
5430 (sese_bad_liveouts_use): Remove.
5431 (sese_reset_debug_liveouts_bb): Likewise.
5432 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5433 (sese_build_liveouts): Build liveout and debug_liveout and store
5435 (new_sese_info): Adjust.
5436 (free_sese_info): Likewise.
5437 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5438 do not build liveout here.
5439 (move_sese_in_condition): Adjust region entry.
5440 (scev_analyzable_p): Match up with chrec_apply requirements.
5441 (sese_trivially_empty_bb_p): New.
5442 * tree-into-ssa.c (get_reaching_def): Properly support generating
5443 default-defs for incremental rewrite of anonymous names.
5445 2017-10-06 Richard Biener <rguenther@suse.de>
5447 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5448 precision do not perform modulo reduction.
5450 2017-10-06 Richard Biener <rguenther@suse.de>
5452 PR tree-optimization/82436
5453 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5454 conservatively choose the vectorization factor when checking
5455 whether we can perform the required load permutation.
5456 (vect_transform_slp_perm_load): Assert when we may not fail.
5458 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5460 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5461 message for incompatible -msdata=* and -mcall-* options.
5463 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5465 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5466 rate for post-reload scheduling.
5468 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5470 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5472 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5474 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5475 to improve monte carlo in scimark.
5477 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5479 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5480 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5481 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5482 pentium4_cost, nocona_cost): Set reassociation width to 1.
5483 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5484 width to 2 for fp operations and 1 otherwise.
5485 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5487 (atom_cost): Set reassociation width to 2.
5488 (slm_cost, generic_cost): Set fp reassociation width
5489 to 2 and 1 otherwise.
5490 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5491 (core_cost): Set fp reassociation width to 4 and vector to 2.
5492 (ix86_reassociation_width): Rewrite using cost table; special case
5493 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5494 and TARGET_AVX128_OPTIMAL.
5495 * config/i386/i386.h (processor_costs): Add
5496 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5497 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5498 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5499 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5500 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5501 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5503 2017-10-05 Nathan Sidwell <nathan@acm.org>
5505 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5507 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5509 * config/arm/arm.c (arm_test_fpu_data): New.
5510 (arm_run_selftests): Call arm_test_fpu_data.
5512 2017-10-04 Nathan Sidwell <nathan@acm.org>
5514 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5516 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5518 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5520 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5521 check type for aarch64_simd_valid_immediate.
5522 (aarch64_output_simd_mov_immediate): Update prototype.
5523 (aarch64_simd_valid_immediate): Update prototype.
5524 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5525 support for ORR-immediate.
5526 (and<mode>3): modified pattern to add support for BIC-immediate.
5527 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5528 now checks for valid immediate for BIC and ORR based on new enum
5530 (aarch64_output_simd_mov_immediate): Function now used to output
5531 BIC/ORR imm as well based on new enum argument.
5532 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5534 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5535 (aarch64_reg_or_bic_imm): Likewise.
5537 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5539 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5540 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5542 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5546 PR rtl-optimization/82396
5547 * haifa-sched.c (autopref_multipass_init): Simplify
5549 (autopref_rank_data): Simplify sort order.
5550 * sched-int.h (autopref_multipass_data_): Remove
5551 multi_mem_insn_p, min_offset and max_offset.
5553 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5555 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5557 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5559 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5562 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5564 * doc/sourcebuild.texi: Document vect_long_mult.
5566 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5568 PR tree-optimization/82413
5569 * fold-const.c (build_range_check): Use widest_int when comparing
5570 the maximum ETYPE value with HIGH.
5572 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5574 PR rtl-optimization/82396
5575 * haifa-sched.c (autopref_multipass_init): Simplify
5577 (autopref_rank_data): Simplify sort order.
5578 * sched-int.h (autopref_multipass_data_): Remove
5579 multi_mem_insn_p, min_offset and max_offset.
5581 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5583 PR tree-optimization/82381
5584 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5585 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5588 PR tree-optimization/82374
5589 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5590 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5591 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5592 current_function_decl to the new decl.
5594 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5596 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5597 helper macro for IEEE float128 hardware built-in functions.
5598 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5600 (TRUNCF128_ODD): Likewise.
5601 (ADDF128_ODD): Likewise.
5602 (SUBF128_ODD): Likewise.
5603 (MULF128_ODD): Likewise.
5604 (DIVF128_ODD): Likewise.
5605 (FMAF128_ODD): Likewise.
5606 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5607 UNSPEC_TRUNC_ROUND_TO_ODD.
5608 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5609 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5610 floating point round to odd instructions.
5611 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5612 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5613 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5614 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5615 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5616 (trunc<mode>sf2_hw): Change the truncate with round to odd
5617 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5618 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5619 to odd hardware instructions.
5620 (sub<mode>3_odd): Likewise.
5621 (mul<mode>3_odd): Likewise.
5622 (div<mode>3_odd): Likewise.
5623 (sqrt<mode>2_odd): Likewise.
5624 (fma<mode>4_odd): Likewise.
5625 (fms<mode>4_odd): Likewise.
5626 (nfma<mode>4_odd): Likewise.
5627 (nfms<mode>4_odd): Likewise.
5628 (trunc<mode>df2_odd): Change the truncate with round to odd
5629 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5631 * doc/extend.texi (PowerPC built-in functions): Update documentation
5632 for existing IEEE float128-bit built-in functions. Add built-in
5633 functions that generate the IEEE 128-bit floating point round to
5636 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5638 PR rtl-optimization/77729
5639 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5640 to (X&(C1&~C2))|C2 transformations.
5642 2017-10-03 Martin Jambor <mjambor@suse.cz>
5644 PR tree-optimization/82363
5645 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5646 mismatch, mark lacc written regardless of racc.
5648 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5650 PR tree-optimization/82381
5651 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5652 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5653 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5654 fallthrough into reassoc_stmt_dominates_stmt_p.
5657 * combine.c (combine_instructions): Don't combine in unreachable
5660 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5663 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5664 function to not use the have_cpu variable. Do not set cpu_index,
5665 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5667 (rs6000_valid_attribute_p): Remove duplicate initializations of
5668 old_optimize and func_optimize.
5669 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5670 (rs6000_activate_target_options): Make global.
5671 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5674 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5676 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5677 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5678 Set *poffset to 0 and *psize and *pmax_size to -1 if
5679 *poffset + *psize overflows in HOST_WIDE_INT.
5681 PR tree-optimization/82387
5682 PR tree-optimization/82388
5683 PR tree-optimization/82389
5684 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5685 instead of live_bytes non-NULL.
5687 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5690 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5693 2017-10-02 Richard Biener <rguenther@suse.de>
5695 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5696 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5697 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5699 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5701 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5702 requested precision matches the type's.
5703 * calls.c (alloc_max_size): Calculate the new candidate size as
5704 a widest_int and use wi::to_widest when comparing it with the
5705 current candidate size.
5706 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5707 zero rather than integer_zero_node.
5708 * match.pd: Check for a no-op conversion before using wi::add
5709 rather than after. Use tree_to_uhwi when summing small shift
5710 counts into an unsigned int.
5712 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5713 Alan Hayward <alan.hayward@arm.com>
5714 David Sherwood <david.sherwood@arm.com>
5717 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5718 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5719 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5720 POINTER_AND_FP_REGS.
5722 2017-10-02 Richard Biener <rguenther@suse.de>
5724 PR tree-optimization/82355
5725 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5726 a mapping for the enclosing loop but avoid generating one for
5728 (copy_bb_and_scalar_dependences): Remove premature codegen
5729 error on PHIs in blocks duplicated into multiple places.
5730 * graphite-scop-detection.c
5731 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5732 in the region use it as loop and nest to analyze the DR in.
5733 (try_generate_gimple_bb): Likewise.
5734 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5735 (add_loop_constraints): For blocks in a loop not in the region
5736 create a dimension with a single iteration.
5737 * sese.h (gbb_loop_at_index): Remove assert.
5739 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5741 * omp-expand.c (adjust_context_scope): New function.
5742 (expand_parallel_call): Call adjust_context_scope.
5744 2017-10-01 Jeff Law <law@redhat.com>
5746 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5747 dom_opt_dom_walker class with direct access to private members.
5748 Add comments. Call test_for_singularity.
5749 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5750 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5751 m_dummy_cond anymore.
5752 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5754 (pass_dominator:execute): Build the dummy_cond here and pass it
5755 to the dom_opt_dom_walker ctor.
5756 (test_for_singularity): New function.
5758 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5759 Maya Rashish <coypu@sdf.org>
5761 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5762 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5763 (alpha*-*-netbsd*) Use nbsd_tm_file.
5764 (arm*-*-netbsdelf*) Likewise.
5765 (i[34567]86-*-netbsdelf*) Likewise.
5766 (x86_64-*-netbsd*) Likewise.
5767 (mips*-*-netbsd*) Likewise.
5768 (powerpc-*-netbsd*) Likewise.
5769 (sh*-*-netbsd*) Likewise.
5770 (sparc-*-netbsdelf*) Likewise.
5771 (sparc64-*-netbsd*) Likewise.
5772 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5774 (vax-*-netbsdelf*) Likewise.
5775 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5776 (UINT_FAST8_TYPE) Likewise.
5777 (INT_FAST16_TYPE) Check CHAR_FAST16.
5778 (UINT_FAST16_TYPE) Likewise.
5780 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5783 * config/i386/i386.md
5784 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5785 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5786 *divmodsi4_zext_2): New define_insn_and_split.
5787 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5788 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5789 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5790 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5791 New define_insn_and_split.
5792 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5793 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5794 operands[1] having DImode when mode is SImode.
5796 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5797 always SImode for DIV and MOD in REG_EQUAL notes.
5799 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5802 * match.pd: Fix handling of NaNs in pattern.
5804 2017-09-29 Jeff Law <law@redhat.com>
5806 * sbitmap.c (bitmap_bit_in_range_p): New function.
5807 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5808 * tree-ssa-dse.c (live_bytes_read): New function.
5809 (dse_classify_store): Ignore reads of dead bytes.
5811 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5812 typos and whitespace errors.
5813 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5814 * config/s390/s390.c (s390_emit_prologue): Likewise.
5816 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5819 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5820 with a symbol for LRA.
5822 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5824 PR rtl-optimization/82338
5825 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5827 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5829 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5830 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5831 (qsort_chk): Declare.
5832 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5833 (qsort_chk): New function.
5835 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5837 PR tree-optimization/82337
5838 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5839 phi definition if the PHI result appears in an abnormal PHI.
5840 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5841 basis appears in an abnormal PHI.
5843 2017-09-29 Richard Biener <rguenther@suse.de>
5845 * graphite-isl-ast-to-gimple.c
5846 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5847 (binary_op_to_tree): Use it.
5848 (get_rename_from_scev): Likewise.
5849 (copy_loop_phi_nodes): Likewise.
5850 (copy_bb_and_scalar_dependences): Likewise.
5851 (translate_pending_phi_nodes): Likewise.
5853 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5856 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5857 for movabsq $(i32 << shift), r64.
5859 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5861 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5862 index when encoding %esp as %rsp to avoid 0x67 prefix.
5864 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5866 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5867 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5869 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5871 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5872 Extensions with more than 16 double VFP registers.
5873 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5874 to_clear_mask and all code related to it. Replace the remaining
5875 entry by a sbitmap and adapt code accordingly.
5877 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5879 * brig-builtins.def: Change pure attributes to const.
5881 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5883 * config.gcc (default_gnu_indirect_function): Default to yes for
5884 sparc*-*-linux* with glibc.
5886 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5888 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5889 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5890 when creating .init_array and .fini_array sections with priority
5893 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5896 * config/aarch64/aarch64.c
5897 (aarch64_builtin_support_vector_misalignment): Always return false
5898 when misalignment is unknown.
5900 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5902 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5903 this function to return false if the definition used by the swap
5904 instruction is artificial, or if the memory address from which the
5905 constant value is loaded is not represented by a base address held
5906 in a register or if the base address register is a frame or stack
5907 pointer. Additionally, return false if the base address of the
5908 loaded constant is a SYMBOL_REF but is not considered to be a
5910 (replace_swapped_load_constant): New function.
5911 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5912 loaded constant vector with a load of a swapped constant vector.
5914 2017-09-27 Carl Love <cel@us.ibm.com>
5916 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5917 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5918 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5921 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5923 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5924 first, always call autopref_rank_data otherwise.
5926 2017-09-27 Richard Biener <rguenther@suse.de>
5928 * graphite-scop-detection.c (find_scop_parameters): Move
5929 loop bound handling ...
5930 (gather_bbs::before_dom_children): ... here, avoiding the need
5931 to build scop_info->loop_nest.
5932 (record_loop_in_sese): Remove.
5933 * sese.h (sese_info_t::loop_nest): Remove.
5934 * sese.c (new_sese_info): Do not allocate loop_nest.
5935 (free_sese_info): Do not free loop_nest.
5937 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5940 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5941 lhs from calls if the lhs has addressable type.
5943 2017-09-27 Richard Biener <rguenther@suse.de>
5945 * graphite.h (scop::max_alias_set): New member.
5946 * graphite-scop-detection.c: Remove references to non-existing
5947 --param in comments.
5948 (build_alias_sets): Record the maximum alias set used for drs.
5949 (build_scops): Support zero as unlimited for
5950 --param graphite-max-arrays-per-scop.
5951 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5953 (build_poly_sr_1): ... here. Compute alias set based on the
5954 maximum alias set used for drs rather than
5955 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5957 2017-09-27 Richard Biener <rguenther@suse.de>
5959 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5960 --param loop-block-tile-size=0 to disable tiling.
5962 2017-09-27 Richard Biener <rguenther@suse.de>
5964 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5965 (graphite-max-nb-scop-params): Document special value zero.
5966 * domwalk.h (dom_walker::STOP): New symbolical constant.
5967 (dom_walker::dom_walker): Add optional parameter for bb to
5969 (dom_walker::~dom_walker): Declare.
5970 (dom_walker::before_dom_children): Document STOP return value.
5971 (dom_walker::m_user_bb_to_rpo): New member.
5972 (dom_walker::m_bb_to_rpo): Likewise.
5973 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5974 mapping here if not provided by the user.
5975 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5976 provided by the user.
5977 (dom_walker::STOP): Define.
5978 (dom_walker::walk): Do not compute bb to RPO mapping here.
5979 Support STOP return value from before_dom_children to stop
5981 * graphite-optimize-isl.c (optimize_isl): If the schedule
5982 is the same still generate code if -fgraphite-identity
5983 or -floop-parallelize-all are given.
5984 * graphite-scop-detection.c: Include cfganal.h.
5985 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5987 (gather_bbs::before_dom_children): Return STOP for BBs
5989 (build_scops): Compute bb to RPO mapping and pass it to
5990 the domwalk. Treat --param graphite-max-nb-scop-params=0
5991 as not limiting the number of params.
5992 * graphite.c (graphite_initialize): Remove limit on the
5993 number of basic-blocks in a function.
5994 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5995 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5996 default value of 10.
5998 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6000 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6001 Adjust code to eliminate needing to do the shift right 32-bits
6002 operation after XSCVDPSPN.
6004 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6006 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6007 ((X / Y) != 0 -> X >= Y): Likewise.
6009 2017-09-26 Carl Love <cel@us.ibm.com>
6011 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6012 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6013 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6014 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6015 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6016 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6017 definitions and overloading.
6018 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6019 statement for P9V_BUILTIN_XST_LEN_R.
6020 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6021 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6022 define_expand and define_insn for the instructions and builtins.
6023 * doc/extend.texi: Update the built-in documentation file for the new
6025 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6026 define_insn for the instructions
6028 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6031 * gcc/config/netbsd-protos.h: New file.
6032 * gcc/config/netbsd.c: New file.
6033 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6034 * gcc/config/t-netbsd: New file.
6035 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6036 (tmake_file) Add t-netbsd.
6037 (extra_objs) Add netbsd.o.
6039 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6043 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6045 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6047 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6048 sign extension from a vector register to a GPR by doing a 32-bit
6049 direct move and then an EXTSW.
6050 (extendsi<mode>2 splitter): Likewise.
6051 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6052 right or vector extract after doing XSCVDPSPN. Use
6053 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6055 (movdi_from_sf_zero_ext): Likewise.
6056 (reload_gpr_from_vsxsf): Likewise.
6057 (p8_mfvsrd_4_disf): Delete, no longer used.
6058 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6059 then needing to move the SFmode to a GPR to use the XSCVDPSP
6060 instruction instead of FRSP and XSCVDPSPN.
6061 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6062 it is adjacent to the other XSCVSPDP insns.
6063 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6064 SFmode to be in traditional Altivec registers.
6065 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6066 (vsx_xscvspdpn): Likewise.
6067 (vsx_xscvspdpn_scalar): Likewise.
6069 2017-09-26 Martin Jambor <mjambor@suse.cz>
6071 * tree-sra.c (compare_access_positions): Put integral types first,
6072 stabilize sorting of integral types, remove conditions putting
6073 non-full-precision integers last.
6074 (sort_and_splice_var_accesses): Disable scalarization if a
6075 non-integert would be represented by a non-full-precision integer.
6077 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6079 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6080 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6081 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6082 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6083 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6084 conditionals inside the function instead of around it. Call
6085 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6086 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6088 2017-09-26 Richard Biener <rguenther@suse.de>
6090 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6092 (scop_detection::build_scop_breadth): ... this. Removed.
6093 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6094 (scop_detection::harmful_stmt_in_bb): Likewise.
6095 (scop_detection::graphite_can_represent_stmt): Likewise.
6096 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6097 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6098 (scop_detection::can_represent_loop_1): ... this. Removed.
6099 (scop_detection::harmful_loop_in_region): Simplify after inlining
6100 the above and remove more quadraticness.
6101 (build_scops): Adjust.
6102 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6105 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6108 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6109 REGNO (base) == SP_REG if base is a REG.
6112 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6113 if it is different SSA_NAME.
6114 (optimize_range_tests_cmp_bitwise): New function.
6115 (optimize_range_tests): Call it.
6117 2017-09-26 Richard Biener <rguenther@suse.de>
6119 PR tree-optimization/82321
6120 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6121 for the def being inside the loop.
6123 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6125 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6127 * config/s390/s390-builtins.def: Fix constraint on op4.
6129 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6131 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6132 independent expanders.
6133 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6134 ("vec_ordered", "vec_unordered"): New expanders.
6136 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6138 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6141 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6143 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6144 vec_unpacks_lo_v16qi.
6145 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6147 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6149 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6150 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6151 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6153 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6155 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6157 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6159 ("*vec_slb<mode>"): New insn pattern.
6160 ("vec_shr_<mode>"): New expander.
6161 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6162 and force the shift count operand to V16QImode.
6163 ("vec_srb<mode>"): Set shift count mode to V16QI.
6165 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6167 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6168 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6169 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6171 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6174 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6176 2017-09-26 Richard Biener <rguenther@suse.de>
6178 PR tree-optimization/82320
6179 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6182 2017-09-25 Jeff Law <law@redhat.com>
6184 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6185 prototype for new argument.
6186 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6187 mostly extracted from rs6000_emit_allocate_stack.
6188 (rs6000_emit_probe_stack_range_stack_clash): New function.
6189 (rs6000_emit_allocate_stack): Call
6190 rs6000_emit_probe_stack_range_stack_clash as needed.
6191 (rs6000_emit_probe_stack_range): Add additional argument
6192 to call to gen_probe_stack_range{si,di}.
6193 (output_probe_stack_range): New.
6194 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6195 (output_probe_stack_range_stack_clash): New.
6196 (rs6000_emit_prologue): Emit notes into dump file as requested.
6197 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6198 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6199 Add additional operand and pass it to output_probe_stack_range.
6201 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6203 PR tree-optimization/82163
6204 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6205 (checking_verify_loop_closed_ssa): New parameter.
6206 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6207 (check_loop_closed_ssa_stmt): Delete.
6208 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6209 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6210 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6213 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6215 * brig-builtins.def: Treat HSAIL barrier builtins as
6216 setjmp/longjump style functions.
6218 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6220 * target.def (constant_alignment): New hook.
6221 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6222 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6223 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6224 * doc/tm.texi: Regenerate.
6225 * targhooks.h (default_constant_alignment): Declare.
6226 (constant_alignment_word_strings): Likewise.
6227 * targhooks.c (default_constant_alignment): New function.
6228 (constant_alignment_word_strings): Likewise.
6229 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6230 instead of CONSTANT_ALIGNMENT.
6231 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6232 (force_const_mem): Likewise.
6233 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6234 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6235 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6236 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6237 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6239 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6240 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6241 constant_alignment_word_strings.
6242 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6243 (CONSTANT_ALIGNMENT): Likewise.
6244 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6245 (arm_constant_alignment): New function.
6246 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6247 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6248 constant_alignment_word_strings.
6249 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6250 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6251 constant_alignment_word_strings.
6252 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6253 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6254 (cris_constant_alignment): New function.
6255 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6256 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6257 (epiphany_constant_alignment): New function.
6258 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6259 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6260 constant_alignment_word_strings.
6261 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6262 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6263 constant_alignment_word_strings.
6264 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6265 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6266 constant_alignment_word_strings.
6267 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6268 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6269 * config/i386/i386.c (ix86_constant_alignment): Make static.
6270 Use the same interface as the target hook.
6271 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6272 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6273 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6274 constant_alignment_word_strings.
6275 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6276 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6277 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6278 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6279 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6280 constant_alignment_word_strings.
6281 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6282 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6283 constant_alignment_word_strings.
6284 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6285 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6286 constant_alignment_word_strings.
6287 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6288 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6290 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6291 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6292 * config/mips/mips.c (mips_constant_alignment): New function.
6293 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6294 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6295 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6296 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6297 (mmix_constant_alignment): Make static. Use the same interface
6299 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6300 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6301 constant_alignment_word_strings.
6302 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6303 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6304 constant_alignment_word_strings.
6305 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6306 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6307 constant_alignment_word_strings.
6308 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6309 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6310 (rs6000_constant_alignment): New function.
6311 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6312 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6313 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6314 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6315 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6316 (rs6000_constant_alignment): New function.
6317 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6318 * config/s390/s390.c (s390_constant_alignment): New function.
6319 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6320 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6321 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6322 constant_alignment_word_strings.
6323 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6324 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6325 (sparc_constant_alignment): New function.
6326 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6327 * config/spu/spu.c (spu_constant_alignment): New function.
6328 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6329 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6330 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6331 constant_alignment_word_strings.
6332 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6333 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6334 constant_alignment_word_strings.
6335 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6336 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6337 constant_alignment_word_strings.
6338 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6339 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6340 (visium_constant_alignment): New function.
6341 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6342 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6343 (xtensa_constant_alignment): New function.
6344 * system.h (CONSTANT_ALIGNMENT): Poison.
6346 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6348 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6349 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6350 (rs6000_builtin_valid_without_lhs): New helper function.
6351 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6352 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6354 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6356 * target.h (vec_perm_indices): Use unsigned short rather than
6358 (auto_vec_perm_indices): Likewise.
6359 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6360 Use unsigned int rather than unsigned char.
6361 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6363 2017-09-25 Richard Biener <rguenther@suse.de>
6365 * cfgloop.h (sort_sibling_loops): Declare.
6366 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6367 (sort_sibling_loops): New function sorting the sibling loop list
6369 * graphite.c (graphite_transform_loops): Sort sibling loops.
6371 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6373 * target.def (vec_perm_const_ok): Change sel parameter to
6375 * optabs-query.c (can_vec_perm_p): Update accordingly.
6376 * doc/tm.texi: Regenerate.
6377 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6378 auto_vec_perm_indices and remove separate nelt field.
6379 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6380 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6381 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6382 (aarch64_expand_vec_perm_const): Update accordingly.
6383 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6384 to vec_perm_indices.
6385 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6386 auto_vec_perm_indices and remove separate nelt field.
6387 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6388 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6389 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6391 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6392 to vec_perm_indices.
6393 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6394 sel to vec_perm_indices.
6395 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6396 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6397 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6399 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6402 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6405 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6406 on the FUNCTION_DECL function context if it has a DIE that is a
6409 2017-09-25 Richard Biener <rguenther@suse.de>
6411 PR tree-optimization/82285
6412 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6415 2017-09-25 Tom de Vries <tom@codesourcery.com>
6419 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6422 2017-09-25 Richard Biener <rguenther@suse.de>
6424 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6425 ISL errors other than isl_error_quota happen. Dump if the
6426 schedule is the same.
6427 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6428 errors instead of aborting inside ISL.
6430 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6433 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6434 of libgcc_eh for m64.
6435 * config/i386/darwin64.h: Likewise.
6437 2017-09-25 Richard Biener <rguenther@suse.de>
6440 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6441 attribute for incomplete types nor twice for complete ones.
6443 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6446 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6447 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6450 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6453 * config/i386/i386.opt (mprefer-avx256): Use
6454 ix86_target_flags variable.
6455 * config/i386/i386.c (ix86_target_string): Move
6456 -mprefer-avx256 to flag2_opts.
6458 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6461 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6462 and x != -1 | y != -1 into (x & y) != -1.
6464 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6466 * config.gcc: Add new case statement to set
6467 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6468 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6469 s390x-*-linux* case statements. Added aarch64 to the list of
6470 supported architectures.
6472 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6474 PR tree-optimization/82289
6475 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6476 STMT_VINFO_RELEVANT_P.
6478 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6479 Alan Hayward <alan.hayward@arm.com>
6480 David Sherwood <david.sherwood@arm.com>
6482 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6483 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6484 (extract_range_from_binary_expr_1): Don't call
6485 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6487 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6488 Alan Hayward <alan.hayward@arm.com>
6489 David Sherwood <david.sherwood@arm.com>
6491 * target.def (preferred_vector_alignment): New hook.
6492 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6494 * doc/tm.texi: Regenerate.
6495 * targhooks.h (default_preferred_vector_alignment): Declare.
6496 * targhooks.c (default_preferred_vector_alignment): New function.
6497 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6499 (DR_TARGET_ALIGNMENT): New macro.
6500 (aligned_access_p): Update commentary.
6501 (vect_known_alignment_in_bytes): New function.
6502 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6504 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6505 Calculate the misalignment based on the target alignment rather than
6507 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6508 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6509 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6510 the target alignment, rather than masking the element misalignment
6511 with the number of elements in a vector. Also use the target
6512 alignment when calculating the maximum number of peels.
6513 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6514 instead of TYPE_ALIGN_UNIT.
6515 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6516 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6517 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6518 (vect_create_data_ref_ptr): Likewise.
6519 (vect_setup_realignment): Realign by ANDing with
6520 -DR_TARGET_MISALIGNMENT.
6521 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6522 the number of peels based on DR_TARGET_ALIGNMENT.
6523 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6524 with the guaranteed alignment boundary when deciding whether
6526 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6527 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6528 (ensure_base_align): Remove stmt_info parameter. Get the
6529 target base alignment from DR_TARGET_ALIGNMENT.
6530 (vectorizable_store): Update call accordingly. Interpret
6531 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6533 (vectorizable_load): Likewise.
6535 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6536 Alan Hayward <alan.hayward@arm.com>
6537 David Sherwood <david.sherwood@arm.com>
6539 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6540 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6541 (vect_enhance_data_refs_alignment): Likewise.
6543 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6545 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6546 error. Only quit immediately if parsing is complete.
6547 (BEGIN): Initialize fatal_err and parse_done.
6548 (begin fpu, end fpu): Check number of arguments.
6549 (begin arch, end arch): Likewise.
6550 (begin cpu, end cpu): Likewise.
6551 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6552 (optalias): Likewise.
6554 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6556 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6557 * config/arm/arm-isa.h: Delete. Move definitions to ...
6558 * arm-cpus.in: ... here. Use new feature and fgroup values.
6559 * config/arm/arm.c (arm_option_override): Use lower case for feature
6561 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6562 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6563 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6565 (print_isa_bits_for): New function.
6566 (gen_isa): New function.
6567 (gen_comm_data): Use print_isa_bits_for.
6568 (define feature): New keyword.
6569 (define fgroup): New keyword.
6570 * config/arm/t-arm (TM_H): Remove.
6571 (GTM_H): Add arm-isa.h.
6572 (arm-isa.h): Add rule to generate file.
6573 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6574 case for feature bit names.
6576 2017-09-22 Richard Biener <rguenther@suse.de>
6578 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6580 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6581 print of no dependency loops ...
6582 * graphite.c (graphite_transform_loops): ... here.
6583 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6585 (same_close_phi_node, remove_duplicate_close_phi,
6586 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6587 (canonicalize_loop_closed_ssa): ... here and simplify.
6588 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6589 (optimize_isl): Use dump_printf_loc to tell when we stopped
6590 optimizing because of an ISL timeout.
6592 2017-09-22 Richard Biener <rguenther@suse.de>
6594 PR tree-optimization/82291
6595 * tree-if-conv.c (predicate_mem_writes): Make sure to
6596 remove writes in blocks predicated with false.
6598 2017-09-22 Richard Biener <rguenther@suse.de>
6600 * sese.c: Include cfganal.h.
6601 (if_region_set_false_region): Remove.
6602 (create_if_region_on_edge): Likewise.
6603 (move_sese_in_condition): Re-implement without destroying
6606 2017-09-22 Richard Biener <rguenther@suse.de>
6608 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6609 Verify both BBs contain loop PHI nodes before dispatching to
6611 (graphite_regenerate_ast_isl): Do not recompute dominators,
6612 do not verify three times. Restructure for clarity.
6613 * graphite-scop-detection.c (same_close_phi_node,
6614 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6615 defined_in_loop_p, canonicalize_loop_closed_ssa,
6616 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6617 checking and SSA rewrite, move to ...
6618 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6619 (graphite_initialize): Do not pass in ctx, do not reset the
6620 SCEV cache, compute only dominators.
6621 (graphite_transform_loops): Allocate ISL ctx after
6622 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6623 Maintain post-dominators only around build_scops.
6624 * sese.c (if_region_set_false_region): Make static. Free
6625 and recompute dominators.
6626 (move_sese_in_condition): Assert we don't get called with
6627 post-dominators computed.
6628 * sese.h (if_region_set_false_region): Remove.
6630 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6632 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6633 mode attribute for TARGET_AVX512VL.
6635 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6637 * config/i386/i386.opt (mprefer-avx256): New option.
6638 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6640 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6641 for TARGET_PREFER_AVX256.
6642 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6644 2017-09-21 Jeff Law <law@redhat.com>
6646 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6647 Fix dump output if the only stack space is for pushed registers.
6649 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6651 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6654 2017-09-21 Martin Sebor <msebor@redhat.com>
6657 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6658 code (in C++) or code that triggers warnings.
6660 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6662 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6664 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6666 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6667 * sched-rgn.c: Ditto.
6668 * sel-sched-ir.c: Ditto.
6670 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6672 * toplev.h (set_random_seed): Adjust return type.
6673 * toplev.c (init_local_tick): Move eager initialization of random_seed
6674 to get_random_seed. Adjust comment.
6675 (init_random_seed): Inline to get_random_seed, delete.
6676 (get_random_seed): Initialize random_seed lazily.
6677 (set_random_seed): Do not return previous value.
6678 (print_switch_value): Do not call get_random_seed.
6680 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6682 * cgraph.c (delete_function_version): New, broken out from...
6683 (cgraph_node::delete_function_version): ...here. Rename to
6684 cgraph_node::delete_function_version_by_decl. Update all uses.
6685 (cgraph_node::remove): Call delete_function_version.
6687 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6690 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6691 VAR_DECLs to which addressable non-volatile parameters are mapped
6692 and for id->retvar after the return value assignment. Clear
6693 id->retval and id->retbnd after inlining.
6695 2017-09-21 Richard Biener <rguenther@suse.de>
6697 PR tree-optimization/82276
6698 PR tree-optimization/82244
6699 * tree-vrp.c (build_assert_expr_for): Set
6700 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6702 (remove_range_assertions): Revert earlier change.
6704 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6707 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6709 2017-09-21 Richard Biener <rguenther@suse.de>
6711 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6712 Restore valid IL after code generation errors.
6713 * graphite.c (graphite_transform_loops): Diagnose code
6714 generation issues as MSG_MISSED_OPTIMIZATION and continue
6715 with processing SCOPs.
6717 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6718 Alan Hayward <alan.hayward@arm.com>
6719 David Sherwood <david.sherwood@arm.com>
6721 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6722 rather than choosing between plus_constant and gen_rtx_<CODE>.
6723 * expr.c (emit_push_insn): Likewise.
6724 (expand_expr_real_2): Likewise.
6726 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6727 Alan Hayward <alan.hayward@arm.com>
6728 David Sherwood <david.sherwood@arm.com>
6730 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6732 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6733 Alan Hayward <alan.hayward@arm.com>
6734 David Sherwood <david.sherwood@arm.com>
6736 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6737 calling tree_to_uhwi.
6739 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6740 Alan Hayward <alan.hayward@arm.com>
6741 David Sherwood <david.sherwood@arm.com>
6743 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6744 INTEGER_CST rather than a negative test for ADDR_EXPR.
6746 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6747 Alan Hayward <alan.hayward@arm.com>
6748 David Sherwood <david.sherwood@arm.com>
6750 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6751 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6753 2017-09-21 Richard Biener <rguenther@suse.de>
6755 PR tree-optimization/71351
6756 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6757 graphite_create_new_loop_guard): Remove, fold remaining parts
6759 (translate_isl_ast_node_for): ... here and simplify.
6761 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6764 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6765 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6766 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6767 alternative always use QI mode, for -Os imov (=R,R) alternative
6768 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6771 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6772 Jeff Law <law@redhat.com>
6774 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6775 (allocate_stack_space): New function, partially extracted from
6777 (s390_emit_prologue): Track offset to most recent stack probe.
6778 Code to allocate space moved into allocate_stack_space.
6779 Dump actions when no stack is allocated.
6780 (s390_prologue_plus_offset): New function.
6781 (s390_emit_stack_probe): Likewise.
6783 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6785 * common.opt (Wa, Wl, Wp, g, gz=): Add
6787 (gno-column-info): Remove.
6788 (gcolumn-info): Drop RejectNegative.
6790 (gno-record-gcc-switches): Remove.
6791 (grecord-gcc-switches): Drop RejectNegative.
6792 (gno-split-dwarf): Remove.
6793 (gsplit-dwarf): Drop RejectNegative.
6794 (gno-strict-dwarf): Remove.
6795 (gstrict-dwarf): Drop RejectNegative.
6796 * config/darwin.opt (gfull, gused): Add RejectNegative.
6797 * dwarf2out.c (gen_producer_string): Drop
6798 gno-record-gcc-switches handler.
6799 * optc-gen.awk: Add g to prefixes with negative forms.
6800 * opts-common.c (remapping_prefix_p): New.
6801 (find_opt): Check it.
6802 (generate_canonical_option): Test g prefix.
6803 (option_map): Add -gno- mapping.
6804 (add_misspelling_candidates): Check remapping_prefix_p.
6806 2017-09-20 Jeff Law <law@redhat.com>
6808 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6809 thinko in stack clash protection support.
6811 * explow.c (compute_stack_clash_protection_loop_data): Use
6812 CONST_INT_P instead of explicit test. Verify object is a
6813 CONST_INT_P before looking at INTVAL.
6814 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6815 instead of explicit test.
6817 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6820 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6821 address instead of to r1 and r11.
6823 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6825 * config.gcc: Support "knm".
6826 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6827 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6829 * config/i386/i386.c (m_KNM): Define.
6830 (processor_target_table): Add "knm".
6832 (ix86_option_override_internal): Add "knm".
6833 (ix86_issue_rate): Add PROCESSOR_KNM.
6834 (ix86_adjust_cost): Ditto.
6835 (ia32_multipass_dfa_lookahead): Ditto.
6836 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6837 (fold_builtin_cpu): Add M_INTEL_KNM.
6838 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6839 (processor_type): Add PROCESSOR_KNM.
6840 * config/i386/x86-tune.def: Add m_KNM.
6841 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6843 2017-09-20 Richard Biener <rguenther@suse.de>
6845 PR tree-optimization/80213
6846 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6847 are allowed in empty BBs as well.
6848 (canonicalize_loop_closed_ssa): Also look for other complex
6850 (scop_detection::get_sese): Include the loop-closed PHI block
6852 (scop_detection::merge_sese): Remove code adding extra blocks.
6853 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6854 (build_scops): Assert the final returned scop is invalid.
6856 2017-09-20 Richard Biener <rguenther@suse.de>
6858 PR tree-optimization/82264
6859 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6861 (vn_phi_lookup): Likewise.
6862 (vn_phi_insert): Likewise.
6864 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6866 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6867 that fits into uhwi or shwi, add DW_AT_const_value regardless
6868 of early_dwarf without going through RTL, using add_AT_unsigned
6871 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6872 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6873 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6874 (DEBUG_MACRO_SECTION): Likewise.
6875 (DEBUG_LTO_MACRO_SECTION): Likewise.
6876 (DEBUG_STR_DWO_SECTION): Likewise.
6877 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6878 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6879 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6880 (DEBUG_LTO_LINE_STR_SECTION): Define.
6881 (init_sections_and_labels): Initialize debug_line_str_section
6882 variable. Initialize debug_loc_section for -gdwarf-5 to
6883 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6885 2017-09-20 Richard Biener <rguenther@suse.de>
6887 * graphite-sese-to-poly.c (extract_affine): Properly handle
6888 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6890 2017-09-20 Richard Biener <rguenther@suse.de>
6892 PR tree-optimization/81373
6893 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6894 Force SESE live-out defs to be handled even if they are
6897 2017-09-19 Jeff Law <law@redhat.com>
6899 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6900 nothing for stack adjustments with REG_STACK_CHECK.
6901 * sched-deps.c (parse_add_or_inc): Reject insns with
6902 REG_STACK_CHECK from dependency breaking.
6903 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6904 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6905 * reg-notes.def (STACK_CHECK): New note.
6907 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6908 (ix86_expand_prologue): Dump stack clash info as needed.
6909 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6911 * function.c (dump_stack_clash_frame_info): New function.
6912 * function.h (dump_stack_clash_frame_info): Prototype.
6913 (enum stack_clash_probes): New enum.
6915 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6916 flag_stack_clash_protection.
6917 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6918 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6919 (arm_frame_pointer_required): Likewise.
6920 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6921 (ia64_expand_prologue): Likewise.
6922 * config/mips/mips.c (mips_expand_prologue): Likewise.
6923 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6924 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6925 (sparc_flat_expand_prologue): Likewise.
6926 * config/spu/spu.c (spu_expand_prologue): Likewise.
6928 * explow.c: Include "params.h".
6929 (anti_adjust_stack_and_probe_stack_clash): New function.
6930 (get_stack_check_protect): Likewise.
6931 (compute_stack_clash_protection_loop_data): Likewise.
6932 (emit_stack_clash_protection_loop_start): Likewise.
6933 (emit_stack_clash_protection_loop_end): Likewise.
6934 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6935 Use anti_adjust_stack_and_probe_stack_clash.
6936 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6937 (emit_stack_clash_protection_loop_start): Likewise.
6938 (emit_stack_clash_protection_loop_end): Likewise.
6939 * rtl.h (get_stack_check_protect): Prototype.
6940 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6941 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6942 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6944 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6946 * doc/tm.texi: Rebuilt.
6947 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6948 get_stack_check_protect.
6949 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6950 * config/arm/arm.c (arm_expand_prologue): Likewise.
6951 (arm_frame_pointer_required): Likewise.
6952 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6953 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6954 * config/mips/mips.c (mips_expand_prologue): Likewise.
6955 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6956 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6957 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6958 (sparc_flat_expand_prologue): Likewise.
6960 * common.opt (-fstack-clash-protection): New option.
6961 * flag-types.h (enum stack_check_type): Note difference between
6962 -fstack-check= and -fstack-clash-protection.
6963 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6964 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6965 * toplev.c (process_options): Issue warnings/errors for cases
6966 not handled with -fstack-clash-protection.
6967 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6968 (-fstack-check): Note additional problem with -fstack-check=generic.
6969 Note that -fstack-check is primarily for Ada and refer users
6970 to -fstack-clash-protection for stack-clash-protection.
6971 Document new params for stack clash protection.
6973 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6975 * config/i386/i386.c (ix86_split_long_move): Do not handle
6976 address used for LEA in a special way.
6978 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6980 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6982 2017-09-19 Martin Sebor <msebor@redhat.com>
6985 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6986 of incompatible types.
6988 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6990 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6991 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6992 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6993 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6995 2017-09-19 Richard Biener <rguenther@suse.de>
6997 PR tree-optimization/82244
6998 * tree-vrp.c (remove_range_assertions): Do not propagate
6999 a constant to abnormals but replace the assert with a copy.
7001 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7003 PR rtl-optimization/57878
7004 PR rtl-optimization/68988
7005 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7006 avoidance test involving non_reload_pseudos. Move frequency test
7007 below the general fragmentation avoidance test.
7009 2017-09-19 Richard Biener <rguenther@suse.de>
7011 PR tree-optimization/69728
7012 * graphite-sese-to-poly.c (schedule_error): New global.
7013 (add_loop_schedule): Handle empty domain by failing the
7015 (build_original_schedule): Handle schedule_error.
7017 2017-09-19 Richard Biener <rguenther@suse.de>
7019 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7020 Do not iterate to sibling loops but only to siblings of inner
7023 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7026 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7027 that the registers are different.
7029 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7031 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7032 to processor_model and "amdfam17h" to arch_names_table.
7033 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7035 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7038 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7040 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7041 Alan Hayward <alan.hayward@arm.com>
7042 David Sherwood <david.sherwood@arm.com>
7044 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7046 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7048 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7049 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7050 vect_schedule_slp_instance.
7051 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7052 Update call to vect_slp_analyze_node_operations. Simplify return
7054 (vect_slp_analyze_bb_1): Update call accordingly.
7055 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7056 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7057 (vect_schedule_slp): Update call accordingly.
7059 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7060 Alan Hayward <alan.hayward@arm.com>
7061 David Sherwood <david.sherwood@arm.com>
7063 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7064 with types that aren't in fact scalar.
7066 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7068 * tree-vect-slp.c (vect_record_max_nunits): New function,
7070 (vect_build_slp_tree_1): ...here.
7071 (vect_build_slp_tree_2): Call it for phis too.
7073 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7075 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7076 to vect_get_vec_def_for_operand when getting the mask operand.
7078 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7079 Alan Hayward <alan.hayward@arm.com>
7080 David Sherwood <david.sherwood@arm.com>
7082 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7085 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7086 Alan Hayward <alan.hayward@arm.com>
7087 David Sherwood <david.sherwood@arm.com>
7089 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7090 calculation for vector booleans.
7092 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7093 Alan Hayward <alan.hayward@arm.com>
7094 David Sherwood <david.sherwood@arm.com>
7096 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7098 (vect_transform_stmt): ...here.
7099 (vect_analyze_stmt): Use it instead of calling
7100 vectorizable_live_operation directly.
7102 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7104 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7105 non-SIMT targets in acc vector loops.
7107 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7109 * configure.ac: Add arc and check if assembler supports gdwarf2.
7110 * configure: Regenerate.
7112 2017-09-18 Richard Biener <rguenther@suse.de>
7114 PR tree-optimization/82220
7115 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7116 epilogue niters from the min_profitable_iters compute.
7118 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7121 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7122 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7123 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7125 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7128 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7129 switching to a new text section.
7131 2017-09-18 Richard Biener <rguenther@suse.de>
7133 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7135 (build_alias_set): Reject aliases with no access function.
7137 2017-09-18 Richard Biener <rguenther@suse.de>
7139 PR tree-optimization/79622
7140 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7142 (build_cross_bb_scalars_use): Likewise.
7144 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7146 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7148 2017-09-18 Alan Modra <amodra@gmail.com>
7151 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7152 stack_pointer_rtx for count 0. Update comments. Break up
7153 large rtl expression.
7155 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7157 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7158 Increase to 20 bytes.
7159 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7160 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7161 or avx version of the stub.
7163 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7166 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7167 compute the minimum stack alignment. Also update preferred stack
7168 boundary for leaf functions.
7170 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7172 PR tree-optimization/82228
7173 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7176 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7178 * common/config/nds32/nds32-common.c
7179 (nds32_option_optimization_table): Refine formatting.
7180 (nds32_option_optimization_table): Use -fsched-pressure and
7181 -fomit-frame-pointer for specific optimization level.
7183 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7185 * config/nds32/nds32.c: Refine formatting and comments.
7186 * config/nds32/nds32.h: Likewise.
7187 * config/nds32/nds32.md: Likewise.
7188 * config/nds32/nds32-cost.c: Likewise.
7189 * config/nds32/nds32-isr.c: Likewise.
7190 * config/nds32/nds32-md-auxiliary.c: Likewise.
7191 * config/nds32/nds32-multiple.md: Likewise.
7192 * config/nds32/nds32-predicates.c: Likewise.
7194 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7195 Jakub Jelinek <jakub@redhat.com>
7197 Add support for -std=c++2a.
7198 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7200 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7202 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7205 * doc/extend.texi (Common Function Attributes): Add
7206 references to ARM, AArch64, and S/390 specific attributes.
7207 (Function Specific Option Pragmas): Add AArch64 and S/390
7208 to list of back ends that support the target pragma.
7210 2017-09-15 Nathan Sidwell <nathan@acm.org>
7212 * doc/standards.texi: Fix C++17 description. Update URLs for
7215 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7217 * common.opt (Wcast-align=strict): New warning option.
7218 * doc/invoke.texi: Document -Wcast-align=strict.
7220 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7222 * cgraph.h (cgraph_thunk_info): Add comments.
7223 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7224 assert for VIRTUAL_* arguments stricter.
7226 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7228 PR tree-optimization/71026
7229 * match.pd: Move RDIV patterns from fold-const.c
7230 * fold-const.c (distribute_real_division): Removed.
7231 (fold_binary_loc): Remove calls to distribute_real_divison.
7233 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7235 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7236 c++1z and gnu++1z as deprecated. Change other references to
7237 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7238 Change -Wc++1z-compat to -Wc++17-compat.
7239 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7240 * dwarf2out.c (highest_c_language): Handle C++17.
7241 (gen_compile_unit_die): Likewise.
7243 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7245 PR rtl-optimization/82192
7246 * combine.c (make_extraction): Don't look through non-paradoxical
7247 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7250 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7251 Alan Hayward <alan.hayward@arm.com>
7252 David Sherwood <david.sherwood@arm.com>
7254 * target.def (function_arg_offset): New hook.
7255 * targhooks.h (default_function_arg_offset): Declare.
7256 * targhooks.c (default_function_arg_offset): New function.
7257 * function.c (locate_and_pad_parm): Use
7258 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7259 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7260 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7261 * doc/tm.texi: Regenerate.
7262 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7263 * config/spu/spu.c (spu_function_arg_offset): New function.
7264 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7265 * system.h (FUNCTION_ARG_OFFSET): Poison.
7267 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7268 Alan Hayard <alan.hayward@arm.com>
7269 David Sherwood <david.sherwood@arm.com>
7271 * target.def (truly_noop_truncation): New hook.
7272 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7273 than TRULY_NOOP_TRUNCATION.
7274 * hooks.h (hook_bool_uint_uint_true): Declare.
7275 * hooks.c (hook_bool_uint_uint_true): New function.
7276 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7277 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7278 * doc/tm.texi: Regenerate.
7279 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7280 rather than TRULY_NOOP_TRUNCATION in comments.
7281 (simplify_comparison): Likewise.
7282 (record_truncated_value): Likewise.
7283 * expmed.c (extract_bit_field_1): Likewise.
7284 (extract_split_bit_field): Likewise.
7285 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7286 instead of TRULY_NOOP_TRUNCATION.
7287 * function.c (assign_parm_setup_block): Likewise.
7288 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7289 * rtlhooks.c: Include target.h.
7290 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7291 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7292 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7293 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7294 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7295 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7296 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7297 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7298 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7299 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7300 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7301 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7302 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7303 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7304 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7305 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7306 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7307 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7308 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7309 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7310 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7311 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7312 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7313 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7314 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7315 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7316 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7317 rather than TRULY_NOOP_TRUNCATION in comments.
7318 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7319 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7320 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7321 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7322 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7323 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7324 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7325 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7326 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7327 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7328 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7329 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7330 rather than TRULY_NOOP_TRUNCATION in comments.
7331 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7332 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7333 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7334 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7335 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7336 TRULY_NOOP_TRUNCATION condition.
7337 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7338 (TRULY_NOOP_TRUNCATION): Delete.
7339 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7340 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7341 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7342 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7343 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7344 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7345 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7346 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7347 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7348 rather than TRULY_NOOP_TRUNCATION in comments.
7349 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7350 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7351 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7352 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7353 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7354 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7356 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7359 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7360 (*cmp_ior): Likewise.
7361 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7362 (*ior_scc_scc_cmp): Likewise.
7363 (*and_scc_scc): Likewise.
7364 (*and_scc_scc_cmp): Likewise.
7366 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7367 Alan Hayard <alan.hayward@arm.com>
7368 David Sherwood <david.sherwood@arm.com>
7370 * target.def (can_change_mode_class): New hook.
7371 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7372 (hard_regno_nregs): Likewise.
7373 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7374 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7375 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7376 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7377 (LOAD_EXTEND_OP): Update accordingly.
7378 * doc/tm.texi: Regenerate.
7379 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7380 CANNOT_CHANGE_MODE_CLASS.
7381 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7382 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7383 * combine.c (simplify_set): Update accordingly.
7384 * emit-rtl.c (validate_subreg): Likewise.
7385 * recog.c (general_operand): Likewise.
7386 * regcprop.c (mode_change_ok): Likewise.
7387 * reload1.c (choose_reload_regs): Likewise.
7388 (inherit_piecemeal_p): Likewise.
7389 * rtlanal.c (simplify_subreg_regno): Likewise.
7390 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7391 instead of CANNOT_CHANGE_MODE_CLASS.
7392 (reload_cse_simplify_operands): Likewise.
7393 * reload.c (push_reload): Use targetm.can_change_mode_class
7394 instead of CANNOT_CHANGE_MODE_CLASS.
7395 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7396 REG_CANNOT_CHANGE_MODE_P.
7397 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7398 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7399 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7400 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7401 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7402 (arm_can_change_mode_class): New function.
7403 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7404 than CANNOT_CHANGE_MODE_CLASS in comments.
7405 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7406 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7407 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7408 (ix86_can_change_mode_class): ...this new function, inverting the
7409 sense of the return value.
7410 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7411 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7412 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7413 (ia64_can_change_mode_class): New function.
7414 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7415 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7416 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7417 (m32c_can_change_mode_class): ...this new function, inverting the
7418 sense of the return value.
7419 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7420 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7421 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7422 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7423 (mips_can_change_mode_class): ...this new function, inverting the
7424 sense of the return value.
7425 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7426 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7427 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7428 (msp430_can_change_mode_class): New function.
7429 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7430 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7431 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7432 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7433 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7434 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7435 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7436 (pa_cannot_change_mode_class): Replace with...
7437 (pa_can_change_mode_class): ...this new function, inverting the
7438 sense of the return value.
7439 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7440 than CANNOT_CHANGE_MODE_CLASS in comments.
7441 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7442 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7443 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7444 (pdp11_cannot_change_mode_class): Replace with...
7445 (pdp11_can_change_mode_class): ...this new function, inverting the
7446 sense of the return value.
7447 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7448 * config/powerpcspe/powerpcspe-protos.h
7449 (rs6000_cannot_change_mode_class_ptr): Delete.
7450 * config/powerpcspe/powerpcspe.c
7451 (rs6000_cannot_change_mode_class_ptr): Delete.
7452 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7453 (rs6000_option_override_internal): Assign to
7454 targetm.can_change_mode_class instead of
7455 rs6000_cannot_change_mode_class_ptr.
7456 (rs6000_cannot_change_mode_class): Replace with...
7457 (rs6000_can_change_mode_class): ...this new function, inverting the
7458 sense of the return value.
7459 (rs6000_debug_cannot_change_mode_class): Replace with...
7460 (rs6000_debug_can_change_mode_class): ...this new function.
7461 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7462 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7463 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7464 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7465 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7467 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7468 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7469 (rs6000_option_override_internal): Assign to
7470 targetm.can_change_mode_class instead of
7471 rs6000_cannot_change_mode_class_ptr.
7472 (rs6000_cannot_change_mode_class): Replace with...
7473 (rs6000_can_change_mode_class): ...this new function, inverting the
7474 sense of the return value.
7475 (rs6000_debug_cannot_change_mode_class): Replace with...
7476 (rs6000_debug_can_change_mode_class): ...this new function.
7477 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7478 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7479 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7480 (s390_can_change_mode_class): ...this new function, inverting the
7481 sense of the return value.
7482 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7483 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7484 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7485 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7486 (sh_cannot_change_mode_class): Replace with...
7487 (sh_can_change_mode_class): ...this new function, inverting the
7488 sense of the return value.
7489 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7490 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7491 (sparc_can_change_mode_class): New function.
7492 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7493 * config/spu/spu.c (spu_can_change_mode_class): New function.
7494 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7495 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7496 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7497 (visium_can_change_mode_class): New function.
7498 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7500 2017-09-15 Richard Biener <rguenther@suse.de>
7502 PR tree-optimization/82217
7503 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7504 but not undefined case.
7506 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7509 * postreload.c (reload_cse_simplify_operands): Skip
7510 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7512 2017-09-15 Richard Biener <rguenther@suse.de>
7514 PR tree-optimization/68823
7515 * graphite-scop-detection.c (build_alias_set): If we have a
7516 possible dependence check whether we can handle them by just
7517 looking at the DRs DR_ACCESS_FNs.
7518 (build_scops): If build_alias_set fails, fail the SCOP.
7520 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7522 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7523 to support float128 built-in functions that require the ISA 3.0
7525 (BU_FLOAT128_3_HW): Likewise.
7526 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7528 (FMAF128): Likewise.
7530 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7531 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7532 floating point instructions.
7533 (rs6000_invalid_builtin): Likewise.
7534 (rs6000_builtin_mask_names): Likewise.
7535 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7536 (RS6000_BTM_FLOAT128_HW): Likewise.
7537 (RS6000_BTM_COMMON): Likewise.
7538 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7540 * doc/extend.texi (RS/6000 built-in functions): Document the
7541 IEEE 128-bit floating point square root and fused multiply-add
7544 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7546 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7547 reg (r2) isn't in the set of registers defined in the prologue.
7549 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7550 Alan Hayward <alan.hayward@arm.com>
7551 David Sherwood <david.sherwood@arm.com>
7553 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7554 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7555 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7556 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7557 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7559 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7560 max_vectorization_factor.
7561 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7563 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7564 Alan Hayward <alan.hayward@arm.com>
7565 David Sherwood <david.sherwood@arm.com>
7567 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7568 (vect_worthwhile_without_simd_p): Declare.
7569 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7570 (vectorizable_reduction): Use it.
7571 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7572 (vectorizable_operation): Likewise.
7574 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7575 Alan Hayward <alan.hayward@arm.com>
7576 David Sherwood <david.sherwood@arm.com>
7578 * tree-vectorizer.h (vect_get_num_copies): New function.
7579 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7580 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7581 (vectorizable_induction): Likewise.
7582 (vectorizable_live_operation): Likewise.
7583 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7584 (vectorizable_bswap): Likewise.
7585 (vectorizable_call): Likewise.
7586 (vectorizable_conversion): Likewise.
7587 (vectorizable_assignment): Likewise.
7588 (vectorizable_shift): Likewise.
7589 (vectorizable_operation): Likewise.
7590 (vectorizable_store): Likewise.
7591 (vectorizable_load): Likewise.
7592 (vectorizable_condition): Likewise.
7593 (vectorizable_comparison): Likewise.
7594 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7596 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7597 Alan Hayward <alan.hayward@arm.com>
7598 David Sherwood <david.sherwood@arm.com>
7600 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7601 of vect_init_vector.
7603 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7604 Alan Hayward <alan.hayward@arm.com>
7605 David Sherwood <david.sherwood@arm.com>
7607 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7608 an inline wrapper that provides a location.
7609 (gimple_build_vector): Likewise.
7610 * gimple-fold.c (gimple_build_vector_from_val): New function.
7611 (gimple_build_vector): Likewise.
7612 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7613 functions to build the initial value. Always return a gimple value.
7614 (get_initial_defs_for_reduction): Likewise. Only compute
7616 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7617 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7618 (vectorizable_induction): Use gimple_build_vector rather than
7621 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7622 Alan Hayward <alan.hayward@arm.com>
7623 David Sherwood <david.sherwood@arm.com>
7625 * target.h (vec_perm_indices): New typedef.
7626 (auto_vec_perm_indices): Likewise.
7627 * optabs-query.h: Include target.h
7628 (can_vec_perm_p): Take a vec_perm_indices *.
7629 * optabs-query.c (can_vec_perm_p): Likewise.
7630 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7631 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7632 * tree-vect-generic.c (lower_vec_perm): Likewise.
7633 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7634 (vect_grouped_load_supported): Likewise.
7635 (vect_shift_permute_load_chain): Likewise.
7636 (vect_permute_store_chain): Use auto_vec_perm_indices.
7637 (vect_permute_load_chain): Likewise.
7638 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7639 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7640 Update uses of can_vec_perm_p.
7641 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7642 mode with a number of elements. Take a vec_perm_indices *.
7643 (vect_create_epilog_for_reduction): Update accordingly.
7644 Use auto_vec_perm_indices.
7645 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7646 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7647 (vect_transform_slp_perm_load): Likewise.
7648 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7649 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7650 (vect_gen_perm_mask_checked): Likewise.
7651 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7652 (vect_gen_perm_mask_checked): Likewise.
7653 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7654 (vectorizable_store): Likewise.
7655 (vectorizable_load): Likewise.
7656 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7657 (vectorizable_bswap): Likewise.
7659 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7660 Alan Hayward <alan.hayward@arm.com>
7661 David Sherwood <david.sherwood@arm.com>
7663 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7664 * tree.c (build_vector): Likewise.
7665 (build_vector_from_ctor): Update accordingly.
7666 (build_vector_from_val): Likewise.
7667 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7668 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7669 * tree-vect-generic.c (add_rshift): Likewise.
7670 (expand_vector_divmod): Likewise.
7671 (optimize_vector_constructor): Likewise.
7672 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7673 (vect_transform_slp_perm_load): Likewise.
7674 (vect_schedule_slp_instance): Likewise.
7675 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7676 (vectorizable_call): Likewise.
7677 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7678 * expmed.c (make_tree): Likewise.
7679 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7680 a vector passed to build_vector.
7681 (fold_convert_const): Likewise.
7682 (exact_inverse): Likewise.
7683 (fold_ternary_loc): Likewise.
7684 (fold_relational_const): Likewise.
7685 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7686 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7687 (const_unop): Likewise. Store the reduction accumulator in a
7688 variable rather than an array.
7689 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7690 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7691 the new vector, rather than constructing it after the input arrays.
7692 (native_interpret_vector): Use auto_vec<tree> when building
7693 a vector passed to build_vector. Add elements in order.
7694 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7695 auto_vec<tree> when building a vector passed to build_vector.
7696 (vect_create_epilog_for_reduction): Likewise.
7697 (vectorizable_induction): Likewise.
7698 (get_initial_def_for_reduction): Likewise. Fix indentation of
7700 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7702 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7703 passed to build_vector.
7705 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7706 Alan Hayward <alan.hayward@arm.com>
7707 David Sherwood <david.sherwood@arm.com>
7709 * tree-core.h (tree_base::u): Add an "nelts" field.
7710 (tree_vector): Use VECTOR_CST_NELTS as the length.
7711 * tree.c (tree_size): Likewise.
7712 (make_vector): Initialize VECTOR_CST_NELTS.
7713 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7714 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7715 TYPE_VECTOR_SUBPARTS.
7716 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7717 as the number of units, setting it from VECTOR_CST_NELTS.
7718 (const_vector_from_tree): Likewise.
7719 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7720 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7721 (fold_negate_expr_1): Likewise.
7722 (fold_convert_const): Likewise.
7723 (const_binop): Likewise. Differentiate the number of output and
7725 (const_unop): Likewise.
7726 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7727 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7728 in cases that did the opposite.
7730 2017-09-14 Richard Biener <rguenther@suse.de>
7732 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7735 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7737 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7739 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7742 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7743 if and where to split a bb, except for splitting before debug insn
7744 sequences followed by non-label real insn. Delete debug insns
7745 in between basic blocks.
7747 * combine.c (make_compound_operation_int): Formatting fixes.
7749 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7750 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7751 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7752 * config/sol2.h (LINK_EH_SPEC): Likewise.
7753 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7754 * config/s390/linux.h (LINK_SPEC): Likewise.
7755 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7756 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7757 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7758 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7759 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7760 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7761 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7762 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7764 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7766 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7768 (ENDFILE_LINUX_SPEC): Likewise.
7769 (LINK_EH_SPEC): Likewise.
7770 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7771 (LINK_OS_LINUX_SPEC32): Likewise.
7772 (LINK_OS_LINUX_SPEC64): Likewise.
7773 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7774 (LINK_OS_LINUX_SPEC): Likewise.
7776 2017-09-13 Martin Liska <mliska@suse.cz>
7779 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7780 CASE_HIGH is NULL_TREE.
7782 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7783 Alan Hayward <alan.hayward@arm.com>
7784 David Sherwood <david.sherwood@arm.com>
7786 * target.def (secondary_memory_needed): New hook.
7787 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7788 instead of SECONDARY_MEMORY_NEEDED.
7789 (secondary_memory_needed_mode): Likewise.
7790 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7791 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7792 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7793 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7794 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7795 * doc/tm.texi: Regenerate.
7796 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7797 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7798 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7799 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7800 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7801 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7802 mode argument first and change the reg_class arguments to reg_class_t.
7803 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7804 Make static. Update the call to inline_secondary_memory_needed.
7805 (ix86_register_move_cost): Update the call to
7806 inline_secondary_memory_needed.
7807 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7808 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7810 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7811 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7813 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7814 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7815 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7816 and match hook interface. Add comment from mips.h.
7817 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7818 * config/mmix/mmix.md (truncdfsf2): Refer to
7819 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7821 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7822 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7823 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7824 (pa_secondary_memory_needed): New function.
7825 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7826 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7827 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7828 (pdp11_secondary_memory_needed): Make static and match hook interface.
7829 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7830 * config/powerpcspe/powerpcspe-protos.h
7831 (rs6000_secondary_memory_needed_ptr): Delete.
7832 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7834 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7835 (rs6000_option_override_internal): Assign to
7836 targetm.secondary_memory_needed rather than
7837 rs6000_secondary_memory_needed_ptr.
7838 (rs6000_secondary_memory_needed): Match hook interface.
7839 (rs6000_debug_secondary_memory_needed): Likewise.
7840 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7841 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7842 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7843 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7844 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7845 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7847 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7848 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7849 (rs6000_option_override_internal): Assign to
7850 targetm.secondary_memory_needed rather than
7851 rs6000_secondary_memory_needed_ptr.
7852 (rs6000_secondary_memory_needed): Match hook interface.
7853 (rs6000_debug_secondary_memory_needed): Likewise.
7854 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7855 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7856 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7857 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7858 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7859 (sparc_secondary_memory_needed): New function.
7860 * lra-constraints.c (check_and_process_move): Refer to
7861 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7863 (curr_insn_transform): Likewise.
7864 (process_alt_operands): Use targetm.secondary_memory_needed
7865 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7866 (check_secondary_memory_needed_p): Likewise.
7867 (choose_split_class): Likewise.
7868 * reload.c: Unconditionally include code that was previously
7869 conditional on SECONDARY_MEMORY_NEEDED.
7870 (push_secondary_reload): Use targetm.secondary_memory_needed
7871 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7872 (push_reload): Likewise.
7873 * reload1.c: Unconditionally include code that was previously
7874 conditional on SECONDARY_MEMORY_NEEDED.
7875 (choose_reload_regs): Use targetm.secondary_memory_needed
7876 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7877 (gen_reload): Likewise.
7878 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7880 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7881 Alan Hayward <alan.hayward@arm.com>
7882 David Sherwood <david.sherwood@arm.com>
7884 * target.def (secondary_memory_needed_mode): New hook:
7885 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7886 * targhooks.h (default_secondary_memory_needed_mode): New function.
7887 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7888 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7889 * doc/tm.texi: Regenerate.
7890 * lra-constraints.c (check_and_process_move): Use
7891 targetm.secondary_memory_needed_mode instead of
7892 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7893 (curr_insn_transform): Likewise.
7894 * reload.c (get_secondary_mem): Likewise.
7895 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7896 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7898 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7899 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7900 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7901 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7902 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7904 * config/powerpcspe/powerpcspe-protos.h
7905 (rs6000_secondary_memory_needed_mode): Delete.
7906 * config/powerpcspe/powerpcspe.c
7907 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7908 (rs6000_secondary_memory_needed_mode): Make static.
7909 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7910 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7912 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7914 (rs6000_secondary_memory_needed_mode): Make static.
7915 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7916 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7917 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7918 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7919 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7921 (sparc_secondary_memory_needed_mode): New function.
7922 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7924 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7926 * config/aarch64/constraints.md (Umq): New constraint.
7927 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7929 (mov<mode>): Update condition.
7931 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7933 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7934 when bitposition is the same.
7936 2017-09-13 Richard Biener <rguenther@suse.de>
7938 * dwarf2out.c (output_die_symbol): Remove.
7939 (output_die): Do not output a DIEs symbol.
7941 2017-09-13 Richard Biener <rguenther@suse.de>
7944 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7945 default-def to avoid breaking iterator update with the weird
7946 interaction with cgraph_update_edges_for_call_stmt_node.
7948 2017-09-13 Richard Biener <rguenther@suse.de>
7950 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7951 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7952 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7953 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7955 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7957 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7958 Disable pc relative literal load irrespective of
7959 TARGET_FIX_ERR_A53_84341 for default.
7961 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7963 * config/sparc/sparc.c (output_return): Output the source location of
7964 the insn in the delay slot, if any.
7965 (output_sibcall): Likewise.
7967 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7970 * common.opt (-static-pie): New alias.
7971 (shared): Negate static-pie.
7972 (-no-pie): Update help text.
7974 (static-pie): New option.
7975 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7976 -static-pie support.
7977 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7978 (LINK_EH_SPEC): Likewise.
7979 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7980 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7981 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7982 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7983 (init_gcc_specs): Likewise.
7984 (init_spec): Likewise.
7985 (display_help): Update help message for -pie.
7986 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7989 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7991 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7992 (movdi_aarch64): Likewise.
7993 (movti_aarch64): Likewise.
7995 2017-09-12 Simon Wright <simon@pushface.org>
7998 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7999 calculation of the minor version, always output as 0.
8001 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8004 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8005 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8006 on it early, rather than manual conversion late. For
8007 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8008 instead of performing manual conversion.
8010 2017-09-12 Carl Love <cel@us.ibm.com>
8012 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8013 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8015 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8016 VMULOSW): Add definitions.
8017 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8018 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8019 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8020 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8021 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8023 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8025 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8027 (movti_aarch64): Likewise.
8028 (movdf_aarch64): Likewise.
8029 (movtf_aarch64): Likewise.
8030 (load_pairdi): Likewise.
8031 (store_pairdi): Likewise.
8032 (load_pairdf): Likewise.
8033 (store_pairdf): Likewise.
8034 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8035 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8036 (ldr_got_small_<mode>): Likewise.
8037 (ldr_got_small_28k_<mode>): Likewise.
8038 (ldr_got_tiny): Likewise.
8039 * config/aarch64/iterators.md (ldst_sz): New.
8040 (ldpstp_sz): Likewise.
8041 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8043 (thunderx_load): Split load_8 to load_16.
8044 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8046 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8047 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8048 (xgene1_store_pair): Split store_8 to store_16.
8049 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8050 (falkor_st_0_st_sd): Split store_8 to store_16.
8052 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8054 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8055 and store1/2/3/4 to store_4/8/12/16.
8056 * config/aarch64/aarch64.md: Update for rename.
8057 * config/arm/arm.md: Likewise.: Likewise.
8058 * config/arm/arm.c: Likewise.
8059 * config/arm/thumb1.md: Likewise.
8060 * config/arm/thumb2.md: Likewise.
8061 * config/arm/vfp.md: Likewise.
8062 * config/arm/arm-generic.md: Likewise.
8063 * config/arm/arm1020e.md: Likewise.
8064 * config/arm/arm1026ejs.md: Likewise.
8065 * config/arm/arm1136jfs.md: Likewise.
8066 * config/arm/arm926ejs.md: Likewise.
8067 * config/arm/cortex-a15.md: Likewise.
8068 * config/arm/cortex-a17.md: Likewise.
8069 * config/arm/cortex-a5.md: Likewise.
8070 * config/arm/cortex-a53.md: Likewise.
8071 * config/arm/cortex-a57.md: Likewise.
8072 * config/arm/cortex-a7.md: Likewise.
8073 * config/arm/cortex-a8.md: Likewise.
8074 * config/arm/cortex-a9.md: Likewise.
8075 * config/arm/cortex-m4.md: Likewise.
8076 * config/arm/cortex-m7.md: Likewise.
8077 * config/arm/cortex-r4.md: Likewise.
8078 * config/arm/exynos-m1.md: Likewise.
8079 * config/arm/fa526.md: Likewise.
8080 * config/arm/fa606te.md: Likewise.
8081 * config/arm/fa626te.md: Likewise.
8082 * config/arm/fa726te.md: Likewise.
8083 * config/arm/fmp626.md: Likewise.
8084 * config/arm/iwmmxt.md: Likewise.
8085 * config/arm/ldmstm.md: Likewise.
8086 * config/arm/marvell-pj4.md: Likewise.
8087 * config/arm/xgene1.md: Likewise.
8088 * config/aarch64/thunderx.md: Likewise.
8089 * config/aarch64/thunderx2t99.md: Likewise.
8090 * config/aarch64/falkor.md: Likewise.
8092 2017-09-12 Martin Liska <mliska@suse.cz>
8094 * attribs.c (private_lookup_attribute): New function.
8095 * attribs.h (private_lookup_attribute): Declared here.
8096 (lookup_attribute): Called from this place.
8098 2017-09-12 Richard Biener <rguenther@suse.de>
8100 PR tree-optimization/82157
8101 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8102 stmts with side-effects.
8104 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8105 Alan Hayward <alan.hayward@arm.com>
8106 David Sherwood <david.sherwood@arm.com>
8108 * target.def (hard_regno_nregs): New hook.
8109 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8110 * targhooks.h (default_hard_regno_nregs): Declare.
8111 * targhooks.c (default_hard_regno_nregs): New function.
8112 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8113 (TARGET_HARD_REGNO_NREGS): ...this hook.
8114 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8115 (CLASS_MAX_NREGS): Likewise.
8116 * doc/tm.texi: Regenerate.
8117 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8118 instead of HARD_REGNO_NREGS.
8119 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8120 HARD_REGNO_NREGS in the comment.
8121 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8122 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8123 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8124 Return an unsigned int.
8125 (TARGET_HARD_REGNO_NREGS): Redefine.
8126 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8127 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8128 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8129 (arc_hard_regno_nregs): New function.
8130 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8131 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8132 (arm_hard_regno_nregs): New function.
8133 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8134 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8135 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8136 (TARGET_HARD_REGNO_NREGS): Redefine.
8137 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8138 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8139 (HARD_REGNO_NREGS): Delete.
8140 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8141 (cr16_hard_regno_nregs): New function.
8142 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8143 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8144 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8145 (cris_hard_regno_nregs): New function.
8146 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8147 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8148 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8149 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8150 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8151 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8152 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8153 (frv_hard_regno_nregs): Make static. Take and return an
8155 (frv_class_max_nregs): Remove outdated copy of documentation.
8156 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8157 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8158 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8159 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8160 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8161 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8162 (TARGET_HARD_REGNO_NREGS): Redefine.
8163 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8164 (CLASS_MAX_NREGS): Update comment.
8165 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8166 (ia64_hard_regno_nregs): New function.
8167 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8168 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8169 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8170 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8171 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8173 (m32c_hard_regno_nregs): Likewise. Make static.
8174 (TARGET_HARD_REGNO_NREGS): Redefine.
8175 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8176 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8177 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8178 (m68k_hard_regno_nregs): New function.
8179 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8180 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8181 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8182 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8183 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8184 Take and return an unsigned int.
8185 (TARGET_HARD_REGNO_NREGS): Redefine.
8186 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8187 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8188 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8189 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8190 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8191 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8192 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8193 (msp430_hard_regno_nregs): Make static. Take and return an
8195 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8196 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8197 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8198 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8199 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8200 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8201 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8202 (TARGET_HARD_REGNO_NREGS): Redefine.
8203 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8204 (PA_HARD_REGNO_NREGS): ...this.
8205 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8206 (PA_HARD_REGNO_NREGS): ...this.
8207 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8208 (pa_hard_regno_nregs): New function.
8209 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8210 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8211 (pdp11_hard_regno_nregs): New function.
8212 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8213 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8214 (rs6000_hard_regno_nregs_hook): New function.
8215 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8216 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8217 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8218 Take and return an unsigned int. Move earlier in file.
8219 (TARGET_HARD_REGNO_NREGS): Redefine.
8220 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8221 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8222 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8223 (rl78_hard_regno_nregs): Make static. Take and return an
8225 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8226 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8227 (rs6000_hard_regno_nregs_hook): New function.
8228 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8229 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8230 (TARGET_HARD_REGNO_NREGS): Redefine.
8231 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8232 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8233 instead of HARD_REGNO_NREGS.
8234 (s390_hard_regno_nregs): New function.
8235 (s390_hard_regno_mode_ok): Add comment from s390.h.
8236 (TARGET_HARD_REGNO_NREGS): Redefine.
8237 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8238 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8239 (sh_hard_regno_nregs): New function.
8240 (sh_pass_in_reg_p): Use it.
8241 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8242 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8243 (sparc_hard_regno_nregs): New function.
8244 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8245 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8246 (spu_function_arg_advance): Use it, supplying a valid register number.
8247 (TARGET_HARD_REGNO_NREGS): Redefine.
8248 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8249 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8250 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8251 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8252 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8253 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8254 (CLASS_MAX_NREGS): Remove copy of old documentation.
8255 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8256 (visium_hard_regno_nregs): New function.
8257 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8258 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8259 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8260 xtensa_hard_regno_nregs): New function.
8261 * system.h (HARD_REGNO_NREGS): Poison.
8263 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8265 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8266 hard_regno_nregs instead of HARD_REGNO_NREGS.
8267 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8268 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8269 (c6x_expand_epilogue): Likewise.
8270 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8271 (frv_read_iacc_argument): Likewise.
8272 * config/sh/sh.c: Include regs.h.
8273 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8274 (regs_used): Likewise.
8275 (output_stack_adjust): Likewise.
8276 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8277 * expmed.c: Include regs.h.
8278 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8279 * ree.c: Include regs.h.
8280 (combine_reaching_defs): Use hard_regno_nregs instead of
8282 (add_removable_extension): Likewise.
8284 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8286 * regs.h (hard_regno_nregs): Turn into a function.
8287 (end_hard_regno): Update accordingly.
8288 * caller-save.c (setup_save_areas): Likewise.
8289 (save_call_clobbered_regs): Likewise.
8290 (replace_reg_with_saved_mem): Likewise.
8291 (insert_restore): Likewise.
8292 (insert_save): Likewise.
8293 * combine.c (can_change_dest_mode): Likewise.
8294 (move_deaths): Likewise.
8295 (distribute_notes): Likewise.
8296 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8297 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8298 (rs6000_split_multireg_move): Likewise.
8299 (rs6000_register_move_cost): Likewise.
8300 (rs6000_memory_move_cost): Likewise.
8301 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8302 (rs6000_split_multireg_move): Likewise.
8303 (rs6000_register_move_cost): Likewise.
8304 (rs6000_memory_move_cost): Likewise.
8305 * cselib.c (cselib_reset_table): Likewise.
8306 (cselib_lookup_1): Likewise.
8307 * emit-rtl.c (set_mode_and_regno): Likewise.
8308 * function.c (aggregate_value_p): Likewise.
8309 * ira-color.c (setup_profitable_hard_regs): Likewise.
8310 (check_hard_reg_p): Likewise.
8311 (calculate_saved_nregs): Likewise.
8312 (assign_hard_reg): Likewise.
8313 (improve_allocation): Likewise.
8314 (calculate_spill_cost): Likewise.
8315 * ira-emit.c (modify_move_list): Likewise.
8316 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8317 (ira_hard_reg_in_set_p): Likewise.
8318 * ira.c (setup_reg_mode_hard_regset): Likewise.
8319 (clarify_prohibited_class_mode_regs): Likewise.
8320 (check_allocation): Likewise.
8321 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8322 (lra_setup_reg_renumber): Likewise.
8323 (setup_try_hard_regno_pseudos): Likewise.
8324 (spill_for): Likewise.
8325 (assign_hard_regno): Likewise.
8326 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8327 * lra-constraints.c (in_class_p): Likewise.
8328 (lra_constraint_offset): Likewise.
8329 (simplify_operand_subreg): Likewise.
8330 (lra_constraints): Likewise.
8331 (split_reg): Likewise.
8332 (split_if_necessary): Likewise.
8333 (invariant_p): Likewise.
8334 (inherit_in_ebb): Likewise.
8335 * lra-lives.c (process_bb_lives): Likewise.
8336 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8337 (get_hard_regs): Likewise.
8338 (do_remat): Likewise.
8339 * lra-spills.c (assign_spill_hard_regs): Likewise.
8340 * mode-switching.c (create_pre_exit): Likewise.
8341 * postreload.c (reload_combine_recognize_pattern): Likewise.
8342 * recog.c (peep2_find_free_register): Likewise.
8343 * regcprop.c (kill_value_regno): Likewise.
8344 (set_value_regno): Likewise.
8345 (copy_value): Likewise.
8346 (maybe_mode_change): Likewise.
8347 (find_oldest_value_reg): Likewise.
8348 (copyprop_hardreg_forward_1): Likewise.
8349 * regrename.c (check_new_reg_p): Likewise.
8350 (regrename_do_replace): Likewise.
8351 * reload.c (push_reload): Likewise.
8352 (combine_reloads): Likewise.
8353 (find_dummy_reload): Likewise.
8354 (operands_match_p): Likewise.
8355 (find_reloads): Likewise.
8356 (find_equiv_reg): Likewise.
8357 (reload_adjust_reg_for_mode): Likewise.
8358 * reload1.c (count_pseudo): Likewise.
8359 (count_spilled_pseudo): Likewise.
8360 (find_reg): Likewise.
8361 (clear_reload_reg_in_use): Likewise.
8362 (free_for_value_p): Likewise.
8363 (allocate_reload_reg): Likewise.
8364 (choose_reload_regs): Likewise.
8365 (reload_adjust_reg_for_temp): Likewise.
8366 (emit_reload_insns): Likewise.
8367 (delete_output_reload): Likewise.
8368 * rtlanal.c (subreg_get_info): Likewise.
8369 * sched-deps.c (sched_analyze_reg): Likewise.
8370 * sel-sched.c (init_regs_for_mode): Likewise.
8371 (mark_unavailable_hard_regs): Likewise.
8372 (choose_best_reg_1): Likewise.
8373 (verify_target_availability): Likewise.
8374 * valtrack.c (dead_debug_insert_temp): Likewise.
8375 * var-tracking.c (track_loc_p): Likewise.
8376 (emit_note_insn_var_location): Likewise.
8377 * varasm.c (make_decl_rtl): Likewise.
8378 * reginfo.c (choose_hard_reg_mode): Likewise.
8379 (init_reg_modes_target): Refer directly to
8380 this_target_regs->x_hard_regno_nregs.
8382 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8384 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8385 instead of hard_regno_nregs.
8387 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8389 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8390 end_hard_regno instead of hard_regno_nregs.
8391 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8392 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8393 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8394 * ira-color.c (improve_allocation): Likewise.
8395 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8396 * lra-lives.c (mark_regno_live): Likewise.
8397 (mark_regno_dead): Likewise.
8398 * lra-remat.c (operand_to_remat): Likewise.
8399 * lra.c (collect_non_operand_hard_regs): Likewise.
8400 * postreload.c (reload_combine_note_store): Likewise.
8401 (move2add_valid_value_p): Likewise.
8402 * reload.c (regno_clobbered_p): Likewise.
8404 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8406 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8408 * config/v850/v850.c (v850_reorg): Likewise.
8409 * reload.c (refers_to_regno_for_reload_p): Likewise.
8410 (find_equiv_reg): Likewise.
8411 * reload1.c (reload_reg_reaches_end_p): Likewise.
8413 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8415 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8417 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8418 * config/arm/arm.c (output_move_neon): Likewise.
8419 (arm_attr_length_move_neon): Likewise.
8420 (neon_split_vcombine): Likewise.
8421 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8422 (c6x_mark_reg_written): Likewise.
8423 (c6x_dwarf_register_span): Likewise.
8424 * config/i386/i386.c (ix86_save_reg): Likewise.
8425 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8426 (rws_access_reg): Likewise.
8427 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8428 * mode-switching.c (create_pre_exit): Likewise.
8429 * ree.c (combine_reaching_defs): Likewise.
8430 (add_removable_extension): Likewise.
8431 * regcprop.c (find_oldest_value_reg): Likewise.
8432 (copyprop_hardreg_forward_1): Likewise.
8433 * reload.c (reload_inner_reg_of_subreg): Likewise.
8434 (push_reload): Likewise.
8435 (combine_reloads): Likewise.
8436 (find_dummy_reload): Likewise.
8437 (reload_adjust_reg_for_mode): Likewise.
8438 * reload1.c (find_reload_regs): Likewise.
8439 (forget_old_reloads_1): Likewise.
8440 (reload_reg_free_for_value_p): Likewise.
8441 (reload_adjust_reg_for_temp): Likewise.
8442 (emit_reload_insns): Likewise.
8443 (delete_output_reload): Likewise.
8444 * sel-sched.c (choose_best_reg_1): Likewise.
8445 (choose_best_pseudo_reg): Likewise.
8447 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8448 Alan Hayward <alan.hayward@arm.com>
8449 David Sherwood <david.sherwood@arm.com>
8451 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8452 * target.def (slow_unaligned_access): New hook.
8453 * targhooks.h (default_slow_unaligned_access): Declare.
8454 * targhooks.c (default_slow_unaligned_access): New function.
8455 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8456 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8457 * doc/tm.texi: Regenerate.
8458 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8459 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8460 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8462 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8463 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8465 (rs6000_slow_unaligned_access): New function.
8466 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8467 (expand_block_compare): Likewise.
8468 (expand_strn_compare): Likewise.
8469 (rs6000_rtx_costs): Likewise.
8470 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8471 (riscv_slow_unaligned_access): Likewise.
8472 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8473 (riscv_slow_unaligned_access_p): ...this and make static.
8474 (riscv_option_override): Update accordingly.
8475 (riscv_slow_unaligned_access): New function.
8476 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8477 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8478 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8479 (rs6000_slow_unaligned_access): New function.
8480 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8481 (rs6000_rtx_costs): Likewise.
8482 * config/rs6000/rs6000-string.c (expand_block_compare)
8483 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8484 of SLOW_UNALIGNED_ACCESS.
8485 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8486 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8487 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8488 of SLOW_UNALIGNED_ACCESS.
8489 * expmed.c (simple_mem_bitfield_p): Likewise.
8490 * expr.c (alignment_for_piecewise_move): Likewise.
8491 (emit_group_load_1): Likewise.
8492 (emit_group_store): Likewise.
8493 (copy_blkmode_from_reg): Likewise.
8494 (emit_push_insn): Likewise.
8495 (expand_assignment): Likewise.
8496 (store_field): Likewise.
8497 (expand_expr_real_1): Likewise.
8498 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8499 * lra-constraints.c (simplify_operand_subreg): Likewise.
8500 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8501 * gimple-ssa-store-merging.c: Likewise in block comment at start
8503 * tree-ssa-strlen.c: Include target.h.
8504 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8505 of SLOW_UNALIGNED_ACCESS.
8506 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8508 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8510 PR rtl-optimization/82185
8511 * expmed.c (emit_store_flag_int): Only test tem if it has been
8514 2017-09-12 Richard Biener <rguenther@suse.de>
8517 * match.pd ((FTYPE) N CMP CST): Fix typo.
8519 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8521 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8523 (mips_near_type_p): Add 'short_call' attribute as a synonym
8525 * doc/extend.texi (short_call): Document new function attribute.
8527 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8530 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8531 assertion check that in the condition.
8532 (get_atomic_generic_size): Likewise. Before testing if parameter
8533 has pointer type, if it has array type, call for C++
8534 default_conversion to perform array-to-pointer conversion.
8536 2017-09-12 Richard Biener <rguenther@suse.de>
8538 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8539 for operations we cannot scalarize.
8541 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8543 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8544 vectors heap vectors. Clean up comments.
8545 Make visited_bbs a reference.
8546 (profitable_jump_thread_path): Make GC
8547 vectors heap vectors. Clean up comments.
8549 (convert_and_register_jump_thread_path): Make GC vectors heap
8551 (check_subpath_and_update_thread_path): Same. Clean up comments.
8552 Make visited_bbs a reference.
8553 (handle_phi): Abstract common code to to
8554 register_jump_thread_path_if_profitable.
8555 Rename VAR_BB to DEF_BB.
8557 Make GC vectors heap vectors.
8558 Make visited_bbs a reference.
8559 (handle_assignment): Same.
8560 (register_jump_thread_path_if_profitable): New.
8561 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8563 Make GC vectors heap vectors. Clean up comments.
8564 Make visited_bbs a reference.
8565 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8566 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8569 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8572 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8573 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8575 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8577 Revert r251800 and r251799.
8579 2017-09-11 Martin Jambor <mjambor@suse.cz>
8582 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8583 arguments in advance.
8584 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8585 use it to find predecessor edges.
8586 (naive_outof_ssa): Collect vector of predecessors.
8588 2017-09-08 Jason Merrill <jason@redhat.com>
8590 PR c++/70029 - ICE with ref-qualifier and -flto
8591 * langhooks.h (struct lang_hooks_for_types): Add
8592 copy_lang_qualifiers.
8593 * attribs.c (build_type_attribute_qual_variant): Use it.
8594 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8596 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8597 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8599 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8602 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8603 (*mulsi3_sp64): New instruction.
8604 (mulsi3): New expander.
8606 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8608 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8610 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8612 * sancov.c: Include memmodel.h.
8614 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8617 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8620 2017-09-07 Carl Love <cel@us.ibm.com>
8622 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8623 the sldi instruction.
8625 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8627 * sancov.c: Include tm_p.h.
8629 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8632 * output.h (switch_to_other_text_partition): New declaration.
8633 * varasm.c (switch_to_other_text_partition): New function.
8634 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8635 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8636 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8637 to the other text partition before emitting LCL label and switch back
8638 after emitting the word after it.
8640 2017-09-07 Richard Biener <rguenther@suse.de>
8642 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8643 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8644 critical edges here, after loop init.
8645 (pass_data_pre): Remove PROP_no_crit_edges flags.
8646 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8647 for valueization of call args to avoid leaking VN_TOP.
8648 (visit_use): Assert we do not visit default defs.
8649 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8650 Use error_mark_node to more easily detect leaking VN_TOP.
8651 All default-defs are varying, not VN_TOP. Mark them visited.
8652 (run_scc_vn): Make code match comment.
8654 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8656 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8657 OPTION_MASK_FLOAT128_KEYWORD.
8658 (POWERPC_MASKS): Likewise.
8659 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8660 support for the -mfloat128-type option, and make -mfloat128
8661 default on PowerPC Linux systems. Define or undefine
8662 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8663 Define __float128 to be __ieee128 if IEEE 128-bit support is
8664 enabled, or undefine it.
8665 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8666 Delete defining __FLOAT128_TYPE__.
8667 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8668 -mfloat128-type option and make -mfloat128 default on PowerPC
8670 (TARGET_FLOAT128_TYPE): Likewise.
8671 (-mfloat128-type): Likewise.
8672 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8673 Delete the -mfloat128-type option and make -mfloat128 default on
8674 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8675 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8676 128-bit floating point is enabled. Change tests from using
8677 -mfloat128-type to -mfloat128.
8678 (rs6000_mangle_type): Use the correct mangling for the __float128
8679 type even if normal long double is restricted to 64-bits.
8680 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8682 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8683 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8684 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8686 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8687 documentation for -mfloat128.
8689 2017-09-06 Olivier Hainque <hainque@adacore.com>
8691 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8693 2017-09-06 Wish Wu <wishwu007@gmail.com>
8694 Jakub Jelinek <jakub@redhat.com>
8696 * asan.c (initialize_sanitizer_builtins): Add
8697 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8698 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8699 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8700 BT_FN_VOID_UINT64_PTR variables.
8701 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8702 (BT_FN_VOID_UINT16_UINT16): Likewise.
8703 (BT_FN_VOID_UINT32_UINT32): Likewise.
8704 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8705 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8706 (BT_FN_VOID_UINT64_PTR): Likewise.
8707 * common.opt (flag_sanitize_coverage): New variable.
8708 (fsanitize-coverage=trace-pc): Remove.
8709 (fsanitize-coverage=): Add.
8710 * flag-types.h (enum sanitize_coverage_code): New enum.
8711 * fold-const.c (fold_range_test): Disable non-short-circuit
8712 optimization if flag_sanitize_coverage.
8713 (fold_truth_andor): Likewise.
8714 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8715 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8716 (coverage_sanitizer_opts): New array.
8717 (get_closest_sanitizer_option): Add OPTS argument, handle also
8718 OPT_fsanitize_coverage_.
8719 (parse_sanitizer_options): Adjusted to also handle
8720 OPT_fsanitize_coverage_.
8721 (common_handle_option): Add OPT_fsanitize_coverage_.
8722 * sancov.c (instrument_comparison, instrument_switch): New function.
8723 (sancov_pass): Add trace-cmp support.
8724 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8725 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8726 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8727 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8728 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8729 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8730 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8731 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8732 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8733 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8735 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8737 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8738 error. Only quit immediately if parsing is complete.
8739 (BEGIN): Initialize fatal_err and parse_done.
8740 (begin fpu, end fpu): Check number of arguments.
8741 (begin arch, end arch): Likewise.
8742 (begin cpu, end cpu): Likewise.
8743 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8744 (optalias): Likewise.
8746 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8748 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8749 * config/arm/arm-isa.h: Delete. Move definitions to ...
8750 * arm-cpus.in: ... here. Use new feature and fgroup values.
8751 * config/arm/arm.c (arm_option_override): Use lower case for feature
8753 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8754 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8755 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8757 (print_isa_bits_for): New function.
8758 (gen_isa): New function.
8759 (gen_comm_data): Use print_isa_bits_for.
8760 (define feature): New keyword.
8761 (define fgroup): New keyword.
8762 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8763 (arm-isa.h): Add rule to generate file.
8764 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8765 case for feature bit names.
8767 2017-09-06 Richard Biener <rguenther@suse.de>
8769 * tree-ssa-pre.c (NECESSARY): Remove.
8770 (create_expression_by_pieces): Do not touch pass-local flags.
8771 (insert_into_preds_of_block): Likewise.
8772 (do_pre_regular_insertion): Likewise.
8773 (eliminate_insert): Likewise.
8774 (eliminate_dom_walker::before_dom_children): Likewise.
8775 (fini_eliminate): Do not look at inserted_exprs.
8776 (mark_operand_necessary): Remove.
8777 (remove_dead_inserted_code): Replace with simple work-list
8778 algorithm based on inserted_exprs and SSA uses.
8779 (pass_pre::execute): Re-order fini_eliminate and
8780 remove_dead_inserted_code.
8782 2017-09-06 Olivier Hainque <hainque@adacore.com>
8784 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8785 for VxWorks 7. Adjust surrounding comments.
8787 2017-09-06 Richard Biener <rguenther@suse.de>
8789 * gimple-ssa-strength-reduction.c
8790 (find_candidates_dom_walker::before_dom_children): Also allow
8793 2017-09-06 Richard Biener <rguenther@suse.de>
8795 PR tree-optimization/82108
8796 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8797 for gap in the non-permutation SLP case.
8799 2017-09-06 Martin Jambor <mjambor@suse.cz>
8801 PR tree-optimization/82078
8802 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8803 add_access_to_work_queue...
8804 (build_accesses_from_assign): ...here.
8805 (propagate_all_subaccesses): Make sure racc is the group
8806 representative, if there is one.
8808 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8811 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8814 2017-09-06 Richard Biener <rguenther@suse.de>
8816 * gimple-ssa-strength-reduction.c
8817 (find_candidates_dom_walker::before_doom_children): Use a
8818 type and not a mode check.
8820 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8823 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8825 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8827 2017-09-05 Jeff Law <law@redhat.com>
8829 PR tree-optimization/64910
8830 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8831 cases where we have 3 or more operands.
8833 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8836 * omp-low.c (lower_omp_for): Recompute tree invariant if
8837 gimple_omp_for_initial/final is ADDR_EXPR.
8840 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8841 into gimple val before gimplification fo the COND_EXPR.
8843 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8845 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8846 REGION_COPY argument.
8847 (thread_through_all_blocks): Remove unused argument to
8848 duplicate_thread_path.
8850 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8851 Alan Hayward <alan.hayward@arm.com>
8852 David Sherwood <david.sherwood@arm.com>
8854 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8855 Take a scalar_mode rather than a machine_mode.
8856 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8857 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8858 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8859 (aarch64_gen_adjusted_ldpstp): Likewise.
8860 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8862 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8863 Alan Hayward <alan.hayward@arm.com>
8864 David Sherwood <david.sherwood@arm.com>
8866 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8867 Take a scalar_int_mode instead of a machine_mode.
8868 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8869 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8870 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8871 (aarch64_simd_attr_length_rglist): Delete.
8872 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8873 a scalar_int_mode instead of a machine_mode.
8874 (aarch64_add_offset): Likewise.
8875 (aarch64_internal_mov_immediate): Likewise
8876 (aarch64_add_constant_internal): Likewise.
8877 (aarch64_add_constant): Likewise.
8878 (aarch64_movw_imm): Likewise.
8879 (aarch64_rtx_arith_op_extract_p): Likewise.
8880 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8881 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8882 Remove assert that the mode isn't a vector.
8883 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8884 (aarch64_expand_mov_immediate): Update calls after above changes.
8885 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8886 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8887 (aarch64_move_imm): Likewise.
8888 (aarch64_can_const_movi_rtx_p): Likewise.
8889 (aarch64_strip_extend): Likewise.
8890 (aarch64_extr_rtx_p): Likewise.
8891 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8892 a CONST_INT when the mode parameter is VOIDmode.
8893 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8895 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8897 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8898 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8899 (bitwise_type_for_mode): Update accordingly.
8901 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8903 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8904 * stor-layout.c (mode_for_size_tree): Likewise.
8905 (mode_for_array): Update accordingly.
8906 (layout_decl): Likewise.
8907 (compute_record_mode): Likewise. Only set the mode once.
8909 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8911 * target.def (get_mask_mode): Change return type to opt_mode.
8913 * doc/tm.texi: Regenerate.
8914 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8915 * targhooks.c (default_get_mask_mode): Likewise.
8916 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8917 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8918 targetm.get_mask_mode.
8919 * tree.c (build_truth_vector_type): Likewise.
8921 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8923 * machmode.h (mode_for_vector): Return an opt_mode.
8924 * stor-layout.c (mode_for_vector): Likewise.
8925 (mode_for_int_vector): Update accordingly.
8926 (layout_type): Likewise.
8927 * config/i386/i386.c (emit_memmov): Likewise.
8928 (ix86_expand_set_or_movmem): Likewise.
8929 (ix86_expand_vector_init): Likewise.
8930 (ix86_get_mask_mode): Likewise.
8931 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8933 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8934 * expmed.c (extract_bit_field_1): Likewise.
8935 * expr.c (expand_expr_real_2): Likewise.
8936 * optabs-query.c (can_vec_perm_p): Likewise.
8937 (can_vec_mask_load_store_p): Likewise.
8938 * optabs.c (expand_vec_perm): Likewise.
8939 * targhooks.c (default_get_mask_mode): Likewise.
8940 * tree-vect-stmts.c (vectorizable_store): Likewise.
8941 (vectorizable_load): Likewise.
8942 (get_vectype_for_scalar_type_and_size): Likewise.
8944 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8946 * machmode.h (mode_for_int_vector): New function.
8947 * stor-layout.c (mode_for_int_vector): Likewise.
8948 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8949 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8950 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8951 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8952 (s390_expand_vcond): Likewise.
8954 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8956 * machmode.h (opt_machine_mode): New type.
8957 (opt_mode<T>): Allow construction from anything that can be
8959 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8960 (mode_for_size): Return an opt_machine_mode.
8961 * stor-layout.c (mode_for_size): Likewise.
8962 (mode_for_size_tree): Update call accordingly.
8963 (bitwise_mode_for_mode): Likewise.
8964 (make_fract_type): Likewise.
8965 (make_accum_type): Likewise.
8966 * caller-save.c (replace_reg_with_saved_mem): Update call
8968 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8969 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8970 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8971 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8972 * expmed.c (extract_bit_field_1): Likewise.
8973 * reload.c (get_secondary_mem): Likewise.
8974 * varasm.c (assemble_integer): Likewise.
8975 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8978 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8980 * machmode.h (decimal_float_mode_for_size): New function.
8981 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8982 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8983 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8984 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8985 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8986 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8988 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8990 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8991 (get_builtin_sync_mode): Likewise.
8992 (expand_ifn_atomic_compare_exchange): Likewise.
8993 (expand_builtin_atomic_clear): Likewise.
8994 (expand_builtin_atomic_test_and_set): Likewise.
8995 (fold_builtin_atomic_always_lock_free): Likewise.
8996 * calls.c (compute_argument_addresses): Likewise.
8997 (emit_library_call_value_1): Likewise.
8998 (store_one_arg): Likewise.
8999 * combine.c (combine_instructions): Likewise.
9000 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9001 * config/arm/arm.c (arm_function_value): Likewise.
9002 (aapcs_allocate_return_reg): Likewise.
9003 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9004 * config/i386/i386.c (construct_container): Likewise.
9005 (ix86_gimplify_va_arg): Likewise.
9006 (ix86_expand_sse_cmp): Likewise.
9007 (emit_memmov): Likewise.
9008 (emit_memset): Likewise.
9009 (expand_small_movmem_or_setmem): Likewise.
9010 (ix86_expand_pextr): Likewise.
9011 (ix86_expand_pinsr): Likewise.
9012 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9013 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9015 * config/mips/mips.c (mips_function_value_1) Likewise.
9016 (mips_block_move_straight): Likewise.
9017 (mips_expand_ins_as_unaligned_store): Likewise.
9018 * config/powerpcspe/powerpcspe.c
9019 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9020 (rs6000_darwin64_record_arg_flush): Likewise.
9021 * config/rs6000/rs6000.c
9022 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9023 (rs6000_darwin64_record_arg_flush): Likewise.
9024 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9025 (sparc_function_value_1): Likewise.
9026 * config/spu/spu.c (adjust_operand): Likewise.
9027 (spu_emit_branch_or_set): Likewise.
9028 (arith_immediate_p): Likewise.
9029 * emit-rtl.c (gen_lowpart_common): Likewise.
9030 * expr.c (expand_expr_real_1): Likewise.
9031 * function.c (assign_parm_setup_block): Likewise.
9032 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9033 * reload1.c (alter_reg): Likewise.
9034 * stor-layout.c (mode_for_vector): Likewise.
9035 (layout_type): Likewise.
9037 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9039 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9040 (spu_convert_move): Likewise.
9041 * lower-subreg.c (resolve_simple_move): Likewise.
9043 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9046 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9047 define_insn to a define_expand.
9048 (altivec_vsum2sws_direct): New define_insn.
9049 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9051 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9053 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9056 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9061 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9062 realign stack if stack alignment needed is less than incoming
9065 2017-09-05 Marek Polacek <polacek@redhat.com>
9068 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9071 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9073 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9075 2017-09-05 Richard Biener <rguenther@suse.de>
9077 PR tree-optimization/82084
9078 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9080 2017-09-05 Richard Biener <rguenther@suse.de>
9082 PR tree-optimization/82102
9083 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9085 2017-09-05 Martin Liska <mliska@suse.cz>
9087 PR tree-optimization/82032
9088 * tree-cfg.c (generate_range_test): New function.
9089 * tree-cfg.h (generate_range_test): Declared here.
9090 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9091 (cleanup_control_expr_graph): Use it.
9092 * tree-switch-conversion.c (try_switch_expansion): Remove
9094 (emit_case_nodes): Use generate_range_test.
9096 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9099 * config/i386/i386.md (*<btsc><mode>_mask): Add
9100 TARGET_USE_BT to insn constraint.
9101 (*btr<mode>_mask): Ditto.
9103 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9105 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9106 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9108 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9111 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9112 TARGET_NEON and TARGET_IWMMXT.
9113 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9114 TARGET_NEON and TARGET_IWMMXT.
9115 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9117 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9119 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9120 (ix86_rewrite_tls_address): Ditto.
9121 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9122 (ix86_rewrite_tls_address_1): Ditto.
9123 (ix86_rewrite_tls_address): Ditto.
9124 * config/i386/predicates.md (tls_address_pattern): New predicate.
9125 * config/i386/i386.md (TLS address splitter): New splitter.
9127 2017-09-04 Richard Biener <rguenther@suse.de>
9129 PR tree-optimization/82084
9130 * fold-const.h (can_native_encode_string_p): Declare.
9131 * fold-const.c (can_native_encode_string_p): Factor out from ...
9132 (native_encode_string): ... here.
9133 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9134 vectorizing stores from constants we later cannot handle.
9136 2017-09-04 Marek Polacek <polacek@redhat.com>
9139 * doc/invoke.texi: Update -Wtautological-compare documentation.
9141 2017-09-04 Jeff Law <law@redhat.com>
9143 PR tree-optimization/64910
9144 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9145 swap the first and last operand if the last is a constant.
9147 2017-09-04 Marek Polacek <polacek@redhat.com>
9150 * convert.c (do_narrow): When sanitizing signed integer overflows,
9151 bail out for signed types.
9152 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9154 2017-09-04 Richard Biener <rguenther@suse.de>
9156 PR tree-optimization/82060
9157 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9158 Move devirtualization after stmt folding and before EH/AB/noreturn
9159 cleanup to get the stmt refs canonicalized. Use a bool instead
9160 of gimple_modified_p since that doesn't work for NOPs. Schedule
9161 NOPs generated by folding for removal.
9163 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9164 Alan Hayward <alan.hayward@arm.com>
9165 David Sherwood <david.sherwood@arm.com>
9167 * coretypes.h (pad_direction): New enum.
9168 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9169 (FUNCTION_ARG_PADDING): Likewise.
9170 * target.def (function_arg_padding): New hook.
9171 * targhooks.h (default_function_arg_padding): Declare.
9172 * targhooks.c (default_function_arg_padding): New function.
9173 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9174 (TARGET_FUNCTION_ARG_PADDING): ...this.
9175 * doc/tm.texi: Regenerate.
9176 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9177 instead of direction.
9178 (compute_argument_addresses): Likewise.
9179 (load_register_parameters): Likewise.
9180 (emit_library_call_value_1): Likewise.
9181 (store_one_arg): Use targetm.calls.function_arg_padding instead
9182 of FUNCTION_ARG_PADDING.
9183 (must_pass_in_stack_var_size_or_pad): Likewise.
9184 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9185 (emit_group_store): Likewise.
9186 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9187 instead of FUNCTION_ARG_PADDING.
9188 (emit_push_insn): Likewise, and propagate enum change throughout
9190 * function.h (direction): Delete.
9191 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9193 * function.c (assign_parm_find_stack_rtl): Likewise.
9194 (assign_parm_setup_block_p): Likewise.
9195 (assign_parm_setup_block): Likewise.
9196 (gimplify_parameters): Likewise.
9197 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9198 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9200 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9201 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9202 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9203 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9204 (aarch64_function_arg_padding): ...this new function.
9205 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9206 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9207 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9208 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9209 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9210 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9211 (arm_pad_arg_upward): Replace with...
9212 (arm_function_arg_padding): ...this new function.
9213 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9215 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9216 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9217 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9218 (ia64_hpux_function_arg_padding): Replace with...
9219 (ia64_function_arg_padding): ...this new function. Use pad_direction
9220 instead of direction. Check for TARGET_HPUX.
9221 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9222 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9223 (iq2000_function_arg_padding): New function.
9224 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9225 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9226 (mips_function_arg_padding): ...this new function.
9227 (mips_pad_reg_upward): Update accordingly.
9228 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9229 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9230 targetm.calls.function_arg_padding.
9231 (FUNCTION_ARG_PADDING): Delete.
9232 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9233 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9234 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9235 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9236 (nios2_block_reg_padding): Return pad_direction instead of direction.
9237 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9238 instead of direction.
9239 (nios2_function_arg_padding): Likewise. Make static.
9240 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9241 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9242 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9243 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9244 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9245 (pa_function_arg_padding): Make static. Return pad_direction instead
9247 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9248 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9249 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9250 instead of direction. Use targetm.calls.function_arg_padding.
9251 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9252 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9253 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9254 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9255 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9257 (function_arg_padding): Rename to...
9258 (rs6000_function_arg_padding): ...this. Make static. Return
9259 pad_direction instead of direction.
9260 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9261 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9262 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9263 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9264 instead of direction. Use targetm.calls.function_arg_padding.
9265 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9266 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9267 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9268 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9269 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9270 (function_arg_padding): Rename to...
9271 (rs6000_function_arg_padding): ...this. Make static. Return
9272 pad_direction instead of direction.
9273 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9274 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9275 * config/s390/s390.c (s390_function_arg_padding): New function.
9276 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9277 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9278 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9279 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9280 (function_arg_padding): Rename to...
9281 (sparc_function_arg_padding): ...this. Make static. Return
9282 pad_direction instead of direction.
9283 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9284 * config/spu/spu.c (spu_function_arg_padding): New function.
9285 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9286 * system.h (FUNCTION_ARG_PADDING): Poison.
9288 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9289 Alan Hayward <alan.hayward@arm.com>
9290 David Sherwood <david.sherwood@arm.com>
9292 * target.def (modes_tieable_p): New hook.
9293 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9294 (TARGET_MODES_TIEABLE_P): ...this.
9295 * doc/tm.texi.in: Regenerate.
9296 * hooks.h (hook_bool_mode_mode_true): Declare.
9297 * hooks.c (hook_bool_mode_mode_true): New function.
9298 * combine.c (subst): Use targetm.modes_tieable_p instead of
9300 * dse.c (find_shift_sequence): Likewise.
9301 * expmed.c (extract_low_bits): Likewise.
9302 * lower-subreg.c: Include target.h.
9303 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9305 * rtlanal.c (rtx_cost): Likewise.
9306 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9307 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9308 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9309 (TARGET_MODES_TIEABLE_P): Redefine.
9310 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9311 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9312 (TARGET_MODES_TIEABLE_P): Redefine.
9313 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9314 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9315 (arc_modes_tieable_p): New function.
9316 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9317 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9318 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9319 (arm_modes_tieable_p): Make static.
9320 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9321 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9322 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9323 (TARGET_MODES_TIEABLE_P): Redefine.
9324 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9325 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9326 (TARGET_MODES_TIEABLE_P): Redefine.
9327 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9328 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9329 (cr16_modes_tieable_p): New function.
9330 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9331 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9332 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9333 (TRULY_NOOP_TRUNCATION): Update comment.
9334 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9335 (TRULY_NOOP_TRUNCATION): Update comment.
9336 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9337 (frv_modes_tieable_p): New function.
9338 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9339 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9340 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9341 (TARGET_MODES_TIEABLE_P): Redefine.
9342 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9343 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9344 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9345 (TARGET_MODES_TIEABLE_P): Redefine.
9346 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9347 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9348 (ia64_modes_tieable_p): New function.
9349 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9350 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9351 (iq2000_modes_tieable_p): New function.
9352 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9353 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9354 (lm32_modes_tieable_p): New function.
9355 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9356 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9357 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9358 (TARGET_MODES_TIEABLE_P): Redefine.
9359 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9360 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9361 (m32r_modes_tieable_p): New function.
9362 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9363 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9364 (m68k_modes_tieable_p): New function.
9365 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9366 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9367 (mcore_modes_tieable_p): New function.
9368 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9369 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9371 (TARGET_MODES_TIEABLE_P): Redefine.
9372 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9373 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9374 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9375 (TARGET_MODES_TIEABLE_P): Redefine.
9376 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9377 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9378 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9379 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9380 (mn10300_modes_tieable_p): ...this and make static.
9381 (TARGET_MODES_TIEABLE_P): Redefine.
9382 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9383 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9384 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9385 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9386 (msp430_modes_tieable_p): Make static.
9387 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9388 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9389 (TARGET_MODES_TIEABLE_P): Redefine.
9390 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9391 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9392 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9393 (TARGET_MODES_TIEABLE_P): Redefine.
9394 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9395 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9396 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9397 (TARGET_MODES_TIEABLE_P): Redefine.
9398 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9399 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9400 (pdp11_modes_tieable_p): New function.
9401 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9402 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9403 (rs6000_modes_tieable_p): New function.
9404 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9405 * config/powerpcspe/powerpcspe.md: Update comment.
9406 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9407 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9408 (TARGET_MODES_TIEABLE_P): Redefine.
9409 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9410 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9411 (rl78_modes_tieable_p): New function.
9412 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9413 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9414 (rs6000_modes_tieable_p): New function.
9415 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9416 * config/rs6000/rs6000.md: Update comment.
9417 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9418 * config/rx/rx.c (rx_modes_tieable_p): New function.
9419 (TARGET_MODES_TIEABLE_P): Redefine.
9420 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9421 * config/s390/s390.c (s390_modes_tieable_p): New function.
9422 (TARGET_MODES_TIEABLE_P): Redefine.
9423 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9424 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9425 (sh_modes_tieable_p): New function.
9426 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9427 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9428 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9429 (sparc_modes_tieable_p): Make static.
9430 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9431 * config/spu/spu.c (spu_modes_tieable_p): New function.
9432 (TARGET_MODES_TIEABLE_P): Redefine.
9433 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9434 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9435 (TARGET_MODES_TIEABLE_P): Redefine.
9436 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9437 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9438 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9439 * config/v850/v850.c (v850_modes_tieable_p): New function.
9440 (TARGET_MODES_TIEABLE_P): Redefine.
9441 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9442 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9443 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9444 (visium_modes_tieable_p): New function.
9445 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9446 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9447 (xtensa_modes_tieable_p): New function.
9448 * system.h (MODES_TIEABLE_P): Poison.
9450 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9451 Alan Hayward <alan.hayward@arm.com>
9452 David Sherwood <david.sherwood@arm.com>
9454 * target.def (hard_regno_mode_ok): New hook.
9455 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9456 (TARGET_HARD_REGNO_MODE_OK): ...this.
9457 * doc/tm.texi.in: Regenerate.
9458 * hooks.h (hook_bool_uint_mode_true): Declare.
9459 * hooks.c (hook_bool_uint_mode_true): New function.
9460 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9462 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9463 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9464 instead of HARD_REGNO_MODE_OK.
9465 * caller-save.c: Include target.h.
9466 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9468 * combine.c (can_combine_p): Likewise.
9469 (combinable_i3pat): Likewise.
9470 (can_change_dest_mode): Likewise.
9471 * expr.c (init_expr_target): Likewise.
9472 (convert_move): Likewise.
9473 (convert_modes): Likewise.
9474 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9475 (setup_prohibited_mode_move_regs): Likewise.
9476 * ira.h (target_ira): Likewise.
9477 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9478 * lra-constraints.c (process_alt_operands): Likewise.
9479 (split_reg): Likewise.
9480 * recog.c (peep2_find_free_register): Likewise.
9481 * ree.c (combine_reaching_defs): Likewise.
9482 * regcprop.c (maybe_mode_change): Likewise.
9483 * reginfo.c (init_reg_sets_1): Likewise.
9484 (choose_hard_reg_mode): Likewise.
9485 (simplifiable_subregs): Likewise.
9486 * regrename.c (check_new_reg_p): Likewise.
9487 * reload.c (find_valid_class): Likewise.
9488 (find_valid_class_1): Likewise.
9489 (reload_inner_reg_of_subreg): Likewise.
9490 (push_reload): Likewise.
9491 (combine_reloads): Likewise.
9492 (find_dummy_reload): Likewise.
9493 (find_reloads): Likewise.
9494 * reload1.c (find_reg): Likewise.
9495 (set_reload_reg): Likewise.
9496 (allocate_reload_reg): Likewise.
9497 (choose_reload_regs): Likewise.
9498 (reload_adjust_reg_for_temp): Likewise.
9499 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9500 (simplify_subreg_regno): Likewise.
9501 * sel-sched.c (init_regs_for_mode): Likewise.
9502 * varasm.c (make_decl_rtl): Likewise.
9503 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9504 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9506 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9507 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9508 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9509 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9510 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9511 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9512 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9513 (arc_mode_class): Delete.
9514 (HARD_REGNO_MODE_OK): Delete.
9515 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9516 (arc_hard_regno_mode_ok): Rename old array to...
9517 (arc_hard_regno_mode_ok_modes): ...this.
9518 (arc_conditional_register_usage): Update accordingly.
9519 (arc_mode_class): Make static.
9520 (arc_hard_regno_mode_ok): New function.
9521 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9522 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9523 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9524 (arm_hard_regno_mode_ok): Make static.
9525 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9527 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9528 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9529 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9531 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9532 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9533 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9534 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9535 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9536 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9537 * config/bfin/predicates.md (valid_reg_operand): Use
9538 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9539 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9540 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9541 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9542 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9543 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9544 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9545 (cr16_hard_regno_mode_ok): Make static and return a bool.
9546 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9547 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9548 (cris_hard_regno_mode_ok): New function.
9549 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9550 (epiphany_mode_class): Delete.
9551 (HARD_REGNO_MODE_OK): Delete.
9552 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9553 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9554 (hard_regno_mode_ok): Rename to...
9555 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9556 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9557 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9559 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9560 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9561 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9562 (frv_hard_regno_mode_ok): Make static and return a bool.
9563 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9565 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9566 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9567 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9568 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9570 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9571 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9572 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9573 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9575 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9576 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9577 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9578 (ia64_hard_regno_mode_ok): New function.
9579 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9580 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9581 (iq2000_hard_regno_mode_ok): New function.
9582 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9583 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9584 (lm32_hard_regno_mode_ok): New function.
9585 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9586 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9587 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9588 instead of HARD_REGNO_MODE_OK.
9589 (m32c_hard_regno_ok): Rename to...
9590 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9591 (m32c_cannot_change_mode_class): Update accordingly.
9592 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9593 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9594 (m32r_mode_class): Delete.
9595 (HARD_REGNO_MODE_OK): Delete.
9596 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9597 (m32r_hard_regno_mode_ok): Rename to...
9598 (m32r_hard_regno_modes): ...this.
9599 (m32r_mode_class): Make static.
9600 (m32r_hard_regno_mode_ok): New function.
9601 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9602 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9603 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9604 (m68k_hard_regno_mode_ok): Make static.
9605 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9606 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9607 (mcore_hard_regno_mode_ok): New function.
9608 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9609 (HARD_REGNO_MODE_OK): Delete.
9610 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9612 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9613 (microblaze_hard_regno_mode_ok): New function.
9614 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9615 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9616 (mips_hard_regno_mode_ok): Delete.
9617 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9618 (mips_hard_regno_mode_ok_p): ...this and make static.
9619 (mips_hard_regno_mode_ok_p): Rename to...
9620 (mips_hard_regno_mode_ok_uncached): ...this.
9621 (mips_hard_regno_mode_ok): New function.
9622 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9623 of HARD_REGNO_MODE_OK.
9624 (mips_option_override): Update after above name changes.
9625 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9626 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9627 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9628 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9629 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9630 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9631 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9632 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9633 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9634 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9635 (msp430_hard_regno_mode_ok): Make static and return a bool.
9636 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9637 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9638 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9640 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9641 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9642 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9643 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9644 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9645 (PA_HARD_REGNO_MODE_OK): ...this
9646 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9647 (PA_HARD_REGNO_MODE_OK): ...this.
9648 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9649 (pa_hard_regno_mode_ok): New function.
9650 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9651 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9652 (pdp11_hard_regno_mode_ok): New function.
9653 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9654 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9656 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9658 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9659 (rs6000_hard_regno_mode_ok): Rename to...
9660 (rs6000_hard_regno_mode_ok_uncached): ...this.
9661 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9662 (rs6000_hard_regno_mode_ok): New function.
9663 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9664 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9665 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9666 (riscv_hard_regno_mode_ok): ...this and make static.
9667 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9668 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9669 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9670 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9671 (rl78_hard_regno_mode_ok): Make static and return bool.
9672 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9673 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9675 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9676 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9677 (rs6000_hard_regno_mode_ok): Rename to...
9678 (rs6000_hard_regno_mode_ok_uncached): ...this.
9679 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9680 (rs6000_hard_regno_mode_ok): New function.
9681 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9682 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9683 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9684 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9685 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9686 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9687 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9688 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9689 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9690 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9691 (sh_hard_regno_mode_ok): Make static.
9692 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9693 instead of HARD_REGNO_MODE_OK.
9694 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9695 (sparc_mode_class): Delete.
9696 (HARD_REGNO_MODE_OK): Delete.
9697 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9698 (hard_regno_mode_classes): Make static.
9699 (sparc_mode_class): Likewise.
9700 (sparc_hard_regno_mode_ok): New function.
9701 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9702 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9703 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9705 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9706 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9707 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9708 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9709 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9710 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9711 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9712 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9713 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9714 (visium_hard_regno_mode_ok): New function.
9715 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9716 instead of HARD_REGNO_MODE_OK.
9717 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9718 (HARD_REGNO_MODE_OK): Delete.
9719 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9720 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9721 (xtensa_option_override): Update accordingly.
9722 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9723 (xtensa_hard_regno_mode_ok): New function.
9724 * system.h (HARD_REGNO_MODE_OK): Poison.
9726 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9727 Alan Hayward <alan.hayward@arm.com>
9728 David Sherwood <david.sherwood@arm.com>
9730 * target.def (hard_regno_call_part_clobbered): New hook.
9731 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9732 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9733 * doc/tm.texi: Regenerate.
9734 * hooks.h (hook_bool_uint_mode_false): Declare.
9735 * hooks.c (hook_bool_uint_mode_false): New function.
9736 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9737 * cselib.c (cselib_process_insn): Use
9738 targetm.hard_regno_call_part_clobbered instead of
9739 HARD_REGNO_CALL_PART_CLOBBERED.
9740 * ira-conflicts.c (ira_build_conflicts): Likewise.
9741 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9742 * lra-constraints.c (need_for_call_save_p): Likewise.
9743 * lra-lives.c: Include target.h.
9744 (check_pseudos_live_through_calls): Use
9745 targetm.hard_regno_call_part_clobbered instead of
9746 HARD_REGNO_CALL_PART_CLOBBERED.
9747 * regcprop.c: Include target.h.
9748 (copyprop_hardreg_forward_1): Use
9749 targetm.hard_regno_call_part_clobbered instead of
9750 HARD_REGNO_CALL_PART_CLOBBERED.
9751 * reginfo.c (choose_hard_reg_mode): Likewise.
9752 * regrename.c (check_new_reg_p): Likewise.
9753 * reload.c (find_equiv_reg): Likewise.
9754 * reload1.c (emit_reload_insns): Likewise.
9755 * sched-deps.c (deps_analyze_insn): Likewise.
9756 * sel-sched.c (init_regs_for_mode): Likewise.
9757 (mark_unavailable_hard_regs): Likewise.
9758 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9759 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9760 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9762 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9763 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9764 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9766 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9768 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9769 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9770 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9772 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9773 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9774 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9776 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9777 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9779 * config/powerpcspe/powerpcspe.c
9780 (rs6000_hard_regno_call_part_clobbered): New function.
9781 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9782 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9783 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9785 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9786 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9787 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9789 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9790 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9791 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9793 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9794 Alan Hayward <alan.hayward@arm.com>
9795 David Sherwood <david.sherwood@arm.com>
9797 * rtl.h (subreg_memory_offset): Declare.
9798 * emit-rtl.c (subreg_memory_offset): New function.
9799 * expmed.c (store_bit_field_1): Use it.
9800 * expr.c (undefined_operand_subword_p): Likewise.
9801 * simplify-rtx.c (simplify_subreg): Likewise.
9803 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9805 PR rtl-optimization/57448
9808 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9809 using atomic_load pattern.
9810 (expand_atomic_store): Likewise.
9812 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9815 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9816 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9817 handling. Use replace_call_with_value with NULL instead of
9818 gsi_replace, unlink_stmt_vdef and release_defs.
9820 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9823 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9825 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9828 * rtl.h (emit_library_call_value_1): Declare.
9829 (emit_library_call): Replace declaration with a series of overloads.
9830 Remove the parameter count argument.
9831 (emit_library_call_value): Likewise.
9832 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9833 with an "rtx_mode_t *".
9834 (emit_library_call_value): Delete.
9835 (emit_library_call): Likewise.
9836 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9837 (asan_emit_allocas_unpoison): Likewise.
9838 * builtins.c (expand_builtin_powi): Likewise.
9839 (expand_asan_emit_allocas_unpoison): Likewise.
9840 * cfgexpand.c (expand_main_function): Likewise.
9841 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9842 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9843 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9844 * config/arm/arm.c (arm_trampoline_init): Likewise.
9845 (arm_call_tls_get_addr): Likewise.
9846 (arm_expand_divmod_libfunc): Likewise.
9847 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9848 (smulsi3_highpart): Likewise.
9849 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9850 (c6x_expand_compare): Likewise.
9851 (c6x_expand_movmem): Likewise.
9852 * config/frv/frv.c (frv_trampoline_init): Likewise.
9853 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9854 (ix86_expand_divmod_libfunc): Likewise.
9855 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9856 (ia64_expand_compare): Likewise.
9857 (ia64_profile_hook): Likewise.
9858 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9859 (nonlocal_goto): Likewise.
9860 (restore_stack_nonlocal): Likewise.
9861 * config/m32r/m32r.c (block_move_call): Likewise.
9862 (m32r_trampoline_init): Likewise.
9863 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9864 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9865 (m68k_call_m68k_read_tp): Likewise.
9866 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9867 (microblaze_expand_divide): Likewise.
9868 * config/mips/mips.h (mips_args): Likewise.
9869 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9870 (MIPS_ICACHE_SYNC): Likewise.
9871 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9872 (nios2_trampoline_init): Likewise.
9873 * config/pa/pa.c (hppa_tls_call): Likewise.
9874 (pa_trampoline_init): Likewise.
9875 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9876 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9877 (expand_strn_compare): Likewise.
9878 (rs6000_generate_compare): Likewise.
9879 (rs6000_expand_float128_convert): Likewise.
9880 (output_profile_hook): Likewise.
9881 (rs6000_trampoline_init): Likewise.
9882 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9883 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9884 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9885 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9886 (rs6000_generate_compare): Likewise.
9887 (rs6000_expand_float128_convert): Likewise.
9888 (output_profile_hook): Likewise.
9889 (rs6000_trampoline_init): Likewise.
9890 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9891 * config/sh/sh.c (sh_trampoline_init): Likewise.
9892 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9893 (sparc_emit_float_lib_cmp): Likewise.
9894 (sparc32_initialize_trampoline): Likewise.
9895 (sparc64_initialize_trampoline): Likewise.
9896 (sparc_profile_hook): Likewise.
9897 * config/spu/spu.c (ea_load_store): Likewise.
9898 * config/spu/spu.md (floatunssidf2): Likewise.
9899 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9900 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9901 * config/visium/visium.c (expand_block_move_4): Likewise.
9902 (expand_block_move_2): Likewise.
9903 (expand_block_move_1): Likewise.
9904 (expand_block_set_4): Likewise.
9905 (expand_block_set_2): Likewise.
9906 (expand_block_set_1): Likewise.
9907 (visium_trampoline_init): Likewise.
9908 (visium_profile_hook): Likewise.
9909 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9910 (xtensa_setup_frame_addresses): Likewise.
9911 (xtensa_trampoline_init): Likewise.
9912 * except.c (sjlj_emit_function_enter): Likewise.
9913 (sjlj_emit_function_exit): Likewise.
9914 * explow.c (allocate_dynamic_stack_space): Likewise.
9915 (probe_stack_range): Likewise.
9916 * expr.c (convert_mode_scalar): Likewise.
9917 * optabs.c (expand_binop): Likewise.
9918 (expand_twoval_binop_libfunc): Likewise.
9919 (expand_unop): Likewise.
9920 (prepare_cmp_insn): Likewise.
9921 (prepare_float_lib_cmp): Likewise.
9922 (expand_float): Likewise.
9923 (expand_fix): Likewise.
9924 (expand_fixed_convert): Likewise.
9925 (maybe_emit_sync_lock_test_and_set): Likewise.
9926 (expand_atomic_compare_and_swap): Likewise.
9927 (expand_mem_thread_fence): Likewise.
9928 (expand_atomic_fetch_op): Likewise.
9930 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9932 * doc/generic.texi (OpenACC): Adjust URL.
9933 * doc/invoke.texi (C Dialect Options): Ditto.
9935 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9937 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9938 predicate for operand 1. Add (m,<S>) constraint.
9939 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9940 Prevent memory operand 1 with register operand 2.
9942 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9944 PR rtl-optimization/82024
9945 * combine.c (try_combine): If the combination result is a PARALLEL,
9946 and we only need to retain the SET in there that would be placed
9947 at I2, check that we can place that at I3 instead, before doing so.
9949 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9952 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9954 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9955 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9958 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9959 Jeff Law <law@redhat.com>
9961 * varasm.c (bss_initializer_p): Do not put constants into .bss
9962 (categorize_decl_for_section): Handle bss_initializer_p returning
9963 false when DECL_INITIAL is NULL.
9965 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9968 * config/s390/s390.c (s390_can_inline_p): New function.
9970 2017-09-01 Jeff Law <law@redhat.com>
9972 PR tree-optimization/82052
9973 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9974 Always initialize the returned slot after a hash table miss
9975 when INSERT is true.
9977 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9979 * config/s390/s390.md (mem_signal_fence): Remove.
9980 * doc/md.texi (mem_signal_fence): Remove.
9981 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9983 * target-insns.def (mem_signal_fence): Remove.
9985 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9988 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9991 * asan.c (create_odr_indicator): Strip name encoding from assembler
9992 name before appending it after __odr_asan_.
9994 2017-09-01 Martin Liska <mliska@suse.cz>
9996 PR tree-optimization/82059
9997 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9998 frequency only when an edge is redirected.
10000 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10002 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10003 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10004 (arc_conditional_register_usage): Remove ARC600 lp_count
10006 (arc_file_start): Emit Tag_ARC_CPU_variation.
10007 (arc_can_use_doloop_p): New conditions to use ZOLs.
10008 (hwloop_fail): New function.
10009 (hwloop_optimize): Likewise.
10010 (hwloop_pattern_reg): Likewise.
10011 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10012 (arc_reorg_loops): New function, calls reorg_loops.
10013 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10014 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10016 (arc_loop_hazard): Remove function, functionality moved into
10018 (arc_hazard): Remove arc_loop_hazard call.
10019 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10020 into hwloop_optimize.
10021 (arc_label_align): Remove ZOL handling.
10022 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10023 * config/arc/arc.md (doloop_begin): Remove pattern.
10024 (doloop_begin_i): Likewise.
10025 (doloop_end_i): Likewise.
10026 (doloop_fallback): Likewise.
10027 (doloop_fallback_m): Likewise.
10028 (doloop_end): Reimplement expand.
10029 (arc_lp): New pattern for LP instruction.
10030 (loop_end): New pattern.
10031 (loop_fail): Likewise.
10032 (decrement_and_branch_until_zero): Likewise.
10033 * config/arc/arc.opt (mlpc-width): New option.
10034 * doc/invoke.texi (mlpc-width): Document option.
10036 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10038 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10039 (arc_ccfsm_advance): Fix checking for delay slots.
10040 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10042 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10044 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10046 (movhi_insn): Update store instruction constraint which are saving
10047 6-bit short immediates.
10048 (movsi_insn): Consider also short scaled load operations.
10049 (zero_extendhisi2_i): Use Usd constraint instead of T.
10050 (extendhisi2_i): Add q constraint.
10051 (arc_clzsi2): Add type and length attributes.
10052 (arc_ctzsi2): Likewise.
10053 * config/arc/constraints.md (Usc): Update constraint, the
10054 assembler can parse two relocations for a single instruction.
10056 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10058 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10059 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10061 2017-08-31 Olivier Hainque <hainque@adacore.com>
10063 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10064 match as powerpc-wrs-vxworks*.
10066 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10068 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10069 register constraint for by-element operand.
10070 (aarch64_mls_elt_merge<mode>): Likewise.
10072 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10074 * config/arc/arc.c (arc_can_follow_jump): Check for short
10077 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10079 * config.gcc: Use g.opt for arc.
10080 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10081 functionality moved to ...
10082 (legitimate_scaled_address_p): New function, ...here.
10083 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10084 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10085 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10087 (arc_override_options): Handle G option.
10088 (arc_output_pic_addr_const): Correct function definition.
10089 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10090 (arc_decl_anon_ns_mem_p): Delete.
10091 (arc_in_small_data_p): Overhaul this function to take into
10092 consideration the value given via G option.
10093 (arc_rewrite_small_data_1): Renamed and corrected old
10094 arc_rewrite_small_data function.
10095 (arc_rewrite_small_data): New function.
10096 (small_data_pattern): Don't use pic_offset_table_rtx.
10097 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10098 * config/arc/simdext.md (movmisalignv2hi): Use
10099 prepare_move_operands function.
10101 (movmisalign*): Likewise.
10102 * doc/invoke.texi (ARC options): Document -G option.
10104 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10106 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10108 * config/arc/arc.c (arc_print_operand): Output scalled address for
10109 sdata whenever is possible.
10110 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10111 load/stores are available.
10112 (compact_sda_memory_operand): Check for the alignment required by
10113 code density instructions.
10114 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10116 * config/arc/constraints.md (Usd): Update constraint.
10117 (Us0): New constraint.
10118 (Usc): Update constraint.
10120 2017-08-31 Richard Biener <rguenther@suse.de>
10122 PR middle-end/82054
10123 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10124 function only once.
10126 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10128 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10129 Resize type_signature.
10131 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10132 Alan Hayward <alan.hayward@arm.com>
10133 David Sherwood <david.sherwood@arm.com>
10135 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10136 subregs whose inner modes can be stored in GPRs.
10137 (aarch64_classify_index): Likewise.
10139 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10140 Alan Hayward <alan.hayward@arm.com>
10141 David Sherwood <david.sherwood@arm.com>
10143 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10144 (V_INT_EQUIV): ...this.
10145 (v_cmp_result): Rename to...
10146 (v_int_equiv): ...this.
10147 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10148 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10149 (copysign<mode>3): Likewise.
10150 (aarch64_simd_bsl<mode>_internal): Likewise.
10151 (aarch64_simd_bsl<mode>): Likewise.
10152 (vec_cmp<mode><mode>): Likewise.
10153 (vcond<mode><mode>): Likewise.
10154 (vcond<v_cmp_mixed><mode>): Likewise.
10155 (vcondu<mode><v_cmp_mixed>): Likewise.
10156 (aarch64_cm<optab><mode>): Likewise.
10157 (aarch64_cmtst<mode>): Likewise.
10158 (aarch64_fac<optab><mode>): Likewise.
10159 (vec_perm_const<mode>): Likewise.
10160 (vcond_mask_<mode><v_cmp_result>): Rename to...
10161 (vcond_mask_<mode><v_int_equiv>): ...this.
10162 (vec_cmp<mode><v_cmp_result>): Rename to...
10163 (vec_cmp<mode><v_int_equiv>): ...this.
10165 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10166 Alan Hayward <alan.hayward@arm.com>
10167 David Sherwood <david.sherwood@arm.com>
10169 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10171 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10172 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10173 (UNSPEC_LD4_DREG): New unspecs.
10174 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10175 (aarch64_ld2<mode>_dreg_be): Replace with...
10176 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10178 (aarch64_ld3<mode>_dreg_le)
10179 (aarch64_ld3<mode>_dreg_be): Replace with...
10180 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10182 (aarch64_ld4<mode>_dreg_le)
10183 (aarch64_ld4<mode>_dreg_be): Replace with...
10184 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10187 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10189 PR tree-optimization/81987
10190 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10191 insert an initializer in a location not dominated by the stride
10194 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10196 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10197 on the entire header of the finally block in the fallthru case.
10199 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10201 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10203 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10205 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10206 rs6000_emit_move_from_cr and call renamed function.
10207 (rs6000_emit_prologue): Call renamed functions.
10208 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10209 movesi_from_cr, remove volatile CRs.
10211 2017-08-30 Jon Beniston <jon@beniston.com>
10212 Richard Biener <rguenther@suse.de>
10214 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10215 of VECTOR_MODE_P check.
10216 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10217 element vector types.
10219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10221 * df.h (df_read_modify_subreg_p): Remove in favor of...
10222 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10223 const_rtx instead of an rtx.
10224 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10225 * df-problems.c (df_word_lr_mark_ref): Likewise.
10226 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10227 (mark_pseudo_reg_dead): Likewise.
10228 (mark_ref_dead): Likewise.
10229 * reginfo.c (init_subregs_of_mode): Likewise.
10230 * sched-deps.c (sched_analyze_1): Likewise.
10231 * df-scan.c (df_def_record_1): Likewise.
10232 (df_uses_record): Likewise.
10233 (df_read_modify_subreg_p): Remove in favor of...
10234 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10235 const_rtx instead of an rtx.
10237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10238 Alan Hayward <alan.hayward@arm.com>
10239 David Sherwood <david.sherwood@arm.com>
10241 * rtl.h (partial_subreg_p): New function.
10242 * caller-save.c (save_call_clobbered_regs): Use it.
10243 * calls.c (expand_call): Likewise.
10244 * combine.c (combinable_i3pat): Likewise.
10245 (simplify_set): Likewise.
10246 (make_extraction): Likewise.
10247 (make_compound_operation_int): Likewise.
10248 (gen_lowpart_or_truncate): Likewise.
10249 (force_to_mode): Likewise.
10250 (make_field_assignment): Likewise.
10251 (reg_truncated_to_mode): Likewise.
10252 (record_truncated_value): Likewise.
10253 (move_deaths): Likewise.
10254 * cse.c (record_jump_cond): Likewise.
10255 (cse_insn): Likewise.
10256 * cselib.c (cselib_lookup_1): Likewise.
10257 * expmed.c (extract_bit_field_using_extv): Likewise.
10258 * function.c (assign_parm_setup_reg): Likewise.
10259 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10260 * ira-build.c (create_insn_allocnos): Likewise.
10261 * lra-coalesce.c (merge_pseudos): Likewise.
10262 * lra-constraints.c (match_reload): Likewise.
10263 (simplify_operand_subreg): Likewise.
10264 (curr_insn_transform): Likewise.
10265 * lra-lives.c (process_bb_lives): Likewise.
10266 * lra.c (new_insn_reg): Likewise.
10267 (lra_substitute_pseudo): Likewise.
10268 * regcprop.c (mode_change_ok): Likewise.
10269 (maybe_mode_change): Likewise.
10270 (copyprop_hardreg_forward_1): Likewise.
10271 * reload.c (push_reload): Likewise.
10272 (find_reloads): Likewise.
10273 (find_reloads_subreg_address): Likewise.
10274 * reload1.c (alter_reg): Likewise.
10275 (eliminate_regs_1): Likewise.
10276 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10278 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10280 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10281 back to if statements, including unpack.
10283 2017-08-30 Martin Liska <mliska@suse.cz>
10285 PR inline-asm/82001
10286 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10288 (func_checker::compare_asm_inputs_outputs): ... this function.
10289 (func_checker::compare_gimple_asm): Use the function to compare
10290 also ASM constrains.
10291 * ipa-icf-gimple.h: Rename the function.
10293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10294 Alan Hayward <alan.hayward@arm.com>
10295 David Sherwood <david.sherwood@arm.com>
10297 * coretypes.h (complex_mode): New type.
10298 * gdbhooks.py (build_pretty_printer): Handle it.
10299 * machmode.h (complex_mode): New class.
10300 (complex_mode::includes_p): New function.
10301 (is_complex_int_mode): Likewise.
10302 (is_complex_float_mode): Likewise.
10303 * genmodes.c (get_mode_class): Handle complex mode classes.
10304 * function.c (expand_function_end): Use is_complex_int_mode.
10306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10307 Alan Hayward <alan.hayward@arm.com>
10308 David Sherwood <david.sherwood@arm.com>
10310 * coretypes.h (scalar_mode_pod): New typedef.
10311 * gdbhooks.py (build_pretty_printer): Handle it.
10312 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10313 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10314 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10315 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10316 as_a <scalar_mode>.
10318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10319 Alan Hayward <alan.hayward@arm.com>
10320 David Sherwood <david.sherwood@arm.com>
10322 * machmode.h (mode_for_vector): Take a scalar_mode instead
10324 * stor-layout.c (mode_for_vector): Likewise.
10325 * explow.c (promote_mode): Use as_a <scalar_mode>.
10326 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10329 Alan Hayward <alan.hayward@arm.com>
10330 David Sherwood <david.sherwood@arm.com>
10332 * target.def (preferred_simd_mode): Take a scalar_mode
10333 instead of a machine_mode.
10334 * targhooks.h (default_preferred_simd_mode): Likewise.
10335 * targhooks.c (default_preferred_simd_mode): Likewise.
10336 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10337 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10338 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10339 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10340 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10341 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10342 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10343 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10345 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10346 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10347 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10348 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10349 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10350 * doc/tm.texi: Regenerate.
10351 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10355 Alan Hayward <alan.hayward@arm.com>
10356 David Sherwood <david.sherwood@arm.com>
10358 * target.def (scalar_mode_supported_p): Take a scalar_mode
10359 instead of a machine_mode.
10360 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10361 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10362 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10363 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10364 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10365 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10366 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10367 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10368 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10369 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10370 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10371 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10372 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10373 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10375 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10376 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10377 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10378 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10379 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10381 * doc/tm.texi: Regenerate.
10383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10384 Alan Hayward <alan.hayward@arm.com>
10385 David Sherwood <david.sherwood@arm.com>
10387 * coretypes.h (opt_scalar_mode): New typedef.
10388 * gdbhooks.py (build_pretty_printers): Handle it.
10389 * machmode.h (mode_iterator::get_2xwider): Add overload for
10391 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10393 * expr.c (convert_mode_scalar): Likewise.
10394 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10395 * optabs.c (expand_float): Likewise.
10396 (expand_fix): Likewise.
10397 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10400 Alan Hayward <alan.hayward@arm.com>
10401 David Sherwood <david.sherwood@arm.com>
10403 * optabs.c (expand_float): Explicitly check for scalars before
10404 using a branching expansion.
10405 (expand_fix): Likewise.
10407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10408 Alan Hayward <alan.hayward@arm.com>
10409 David Sherwood <david.sherwood@arm.com>
10411 * expr.c (convert_mode): Split scalar handling out into...
10412 (convert_mode_scalar): ...this new function. Treat the modes
10415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10416 Alan Hayward <alan.hayward@arm.com>
10417 David Sherwood <david.sherwood@arm.com>
10419 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10421 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10424 Alan Hayward <alan.hayward@arm.com>
10425 David Sherwood <david.sherwood@arm.com>
10427 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10428 rather than a machine_mode.
10429 (fixed_from_string): Likewise.
10430 (fixed_convert): Likewise.
10431 (fixed_convert_from_int): Likewise.
10432 (fixed_convert_from_real): Likewise.
10433 (real_convert_from_fixed): Likewise.
10434 * fixed-value.c (fixed_from_double_int): Likewise.
10435 (fixed_from_string): Likewise.
10436 (fixed_convert): Likewise.
10437 (fixed_convert_from_int): Likewise.
10438 (fixed_convert_from_real): Likewise.
10439 (real_convert_from_fixed): Likewise.
10440 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10443 Alan Hayward <alan.hayward@arm.com>
10444 David Sherwood <david.sherwood@arm.com>
10446 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10447 of separate mode class checks. Do not allow vector modes here.
10448 (immed_wide_int_const): Use as_a <scalar_mode>.
10449 * explow.c (trunc_int_for_mode): Likewise.
10450 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10451 (wi::shwi): Likewise.
10452 (wi::min_value): Likewise.
10453 (wi::max_value): Likewise.
10454 * dwarf2out.c (loc_descriptor): Likewise.
10455 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10456 for CONST_WIDE_INT.
10458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10459 Alan Hayward <alan.hayward@arm.com>
10460 David Sherwood <david.sherwood@arm.com>
10462 * tree.h (SCALAR_TYPE_MODE): New macro.
10463 * expr.c (expand_expr_addr_expr_1): Use it.
10464 (expand_expr_real_2): Likewise.
10465 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10466 (fold_convert_const_fixed_from_int): Likewise.
10467 (fold_convert_const_fixed_from_real): Likewise.
10468 (native_encode_fixed): Likewise
10469 (native_encode_complex): Likewise
10470 (native_encode_vector): Likewise.
10471 (native_interpret_fixed): Likewise.
10472 (native_interpret_real): Likewise.
10473 (native_interpret_complex): Likewise.
10474 (native_interpret_vector): Likewise.
10475 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10476 (simd_clone_adjust_argument_types): Likewise.
10477 (simd_clone_init_simd_arrays): Likewise.
10478 (simd_clone_adjust): Likewise.
10479 * stor-layout.c (layout_type): Likewise.
10480 * tree.c (build_minus_one_cst): Likewise.
10481 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10482 * tree-inline.c (estimate_move_cost): Likewise.
10483 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10484 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10485 (vectorizable_reduction): Likewise.
10486 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10487 (vect_recog_mixed_size_cond_pattern): Likewise.
10488 (check_bool_pattern): Likewise.
10489 (adjust_bool_pattern): Likewise.
10490 (search_type_for_mask_1): Likewise.
10491 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10492 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10493 (vectorizable_load): Likewise.
10494 (vectorizable_store): Likewise.
10495 * ubsan.c (ubsan_encode_value): Likewise.
10496 * varasm.c (output_constant): Likewise.
10498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10499 Alan Hayward <alan.hayward@arm.com>
10500 David Sherwood <david.sherwood@arm.com>
10502 * coretypes.h (scalar_mode): New class.
10503 * machmode.h (scalar_mode): Likewise.
10504 (scalar_mode::includes_p): New function.
10505 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10506 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10507 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10508 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10509 * expmed.c (store_bit_field_1): Likewise.
10510 (extract_bit_field_1): Likewise.
10511 * expr.c (write_complex_part): Likewise.
10512 (read_complex_part): Likewise.
10513 (emit_move_complex_push): Likewise.
10514 (expand_expr_real_2): Likewise.
10515 * function.c (assign_parm_setup_reg): Likewise.
10516 (assign_parms_unsplit_complex): Likewise.
10517 * optabs.c (expand_binop): Likewise.
10518 * rtlanal.c (subreg_get_info): Likewise.
10519 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10520 * varasm.c (output_constant_pool_2): Likewise.
10522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10523 Alan Hayward <alan.hayward@arm.com>
10524 David Sherwood <david.sherwood@arm.com>
10526 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10528 (expmed_mult_highpart_optab): Likewise.
10529 (expmed_mult_highpart): Likewise.
10531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10532 Alan Hayward <alan.hayward@arm.com>
10533 David Sherwood <david.sherwood@arm.com>
10535 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10536 instead of a machine_mode.
10537 (builtin_memset_read_str): Likewise.
10538 * builtins.c (c_readstr): Likewise.
10539 (builtin_memcpy_read_str): Likewise.
10540 (builtin_strncpy_read_str): Likewise.
10541 (builtin_memset_read_str): Likewise.
10542 (builtin_memset_gen_str): Likewise.
10543 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10544 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10545 instead of a machine_mode.
10546 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10548 (make_extraction): Likewise.
10549 (try_widen_shift_mode): Take and return scalar_int_modes instead
10551 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10552 a scalar_int_mode instead of a machine_mode.
10553 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10554 (avr_addr_space_pointer_mode): Likewise.
10555 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10556 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10557 (msp430_unwind_word_mode): Likewise.
10558 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10559 (spu_addr_space_pointer_mode): Likewise.
10560 (spu_addr_space_address_mode): Likewise.
10561 (spu_libgcc_cmp_return_mode): Likewise.
10562 (spu_libgcc_shift_count_mode): Likewise.
10563 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10564 (rl78_addr_space_pointer_mode): Likewise.
10565 (fl78_unwind_word_mode): Likewise.
10566 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10568 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10569 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10570 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10571 (mips_valid_pointer_mode): Likewise.
10572 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10573 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10574 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10576 (ft32_addr_space_address_mode): Likewise.
10577 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10578 scalar_int_mode instead of a machine_mode.
10579 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10581 (m32c_addr_space_address_mode): Likewise.
10582 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10583 (rs6000_eh_return_filter_mode): Likewise.
10584 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10585 (rs6000_eh_return_filter_mode): Likewise.
10586 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10587 (s390_libgcc_shift_count_mode): Likewise.
10588 (s390_unwind_word_mode): Likewise.
10589 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10591 * target.def (mode_rep_extended): Likewise.
10592 (valid_pointer_mode): Likewise.
10593 (addr_space.valid_pointer_mode): Likewise.
10594 (eh_return_filter_mode): Return a scalar_int_mode rather than
10596 (libgcc_cmp_return_mode): Likewise.
10597 (libgcc_shift_count_mode): Likewise.
10598 (unwind_word_mode): Likewise.
10599 (addr_space.pointer_mode): Likewise.
10600 (addr_space.address_mode): Likewise.
10601 * doc/tm.texi: Regenerate.
10602 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10604 (do_jump): Use scalar_int_mode for local variables.
10605 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10606 rather than a machine_mode.
10607 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10608 (scompare_loc_descriptor_wide): Likewise.
10609 (scompare_loc_descriptor_narrow): Likewise.
10610 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10612 * except.c (sjlj_emit_dispatch_table): Likewise.
10613 (expand_builtin_eh_copy_values): Likewise.
10614 * explow.c (convert_memory_address_addr_space_1): Likewise.
10615 Take a scalar_int_mode rather than a machine_mode.
10616 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10617 than a machine_mode.
10618 (memory_address_addr_space): Use scalar_int_mode for local variables.
10619 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10620 rather than a machine_mode.
10621 * expmed.c (mask_rtx): Likewise.
10622 (init_expmed_one_conv): Likewise.
10623 (expand_mult_highpart_adjust): Likewise.
10624 (extract_high_half): Likewise.
10625 (expmed_mult_highpart_optab): Likewise.
10626 (expmed_mult_highpart): Likewise.
10627 (expand_smod_pow2): Likewise.
10628 (expand_sdiv_pow2): Likewise.
10629 (emit_store_flag_int): Likewise.
10630 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10632 (extract_low_bits): Likewise.
10633 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10635 * expr.c (pieces_addr::adjust): Likewise.
10636 (can_store_by_pieces): Likewise.
10637 (store_by_pieces): Likewise.
10638 (clear_by_pieces_1): Likewise.
10639 (expand_expr_addr_expr_1): Likewise.
10640 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10641 (expand_expr_real_1): Likewise.
10642 (try_casesi): Likewise.
10643 * final.c (shorten_branches): Likewise.
10644 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10645 type of "mode" to machine_mode.
10646 * internal-fn.c (expand_arith_overflow_result_store): Take a
10647 scalar_int_mode rather than a machine_mode.
10648 (expand_mul_overflow): Use scalar_int_mode for local variables.
10649 * loop-doloop.c (doloop_modify): Likewise.
10650 (doloop_optimize): Likewise.
10651 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10652 than a machine_mode.
10653 (expand_doubleword_shift_condmove): Likewise.
10654 (expand_doubleword_shift): Likewise.
10655 (expand_doubleword_clz): Likewise.
10656 (expand_doubleword_popcount): Likewise.
10657 (expand_doubleword_parity): Likewise.
10658 (expand_absneg_bit): Use scalar_int_mode for local variables.
10659 (prepare_float_lib_cmp): Likewise.
10660 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10661 rather than a machine_mode.
10662 (convert_memory_address_addr_space): Likewise.
10663 (get_mode_bounds): Likewise.
10664 (get_address_mode): Return a scalar_int_mode rather than a
10666 * rtlanal.c (get_address_mode): Likewise.
10667 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10668 than a machine_mode.
10669 * targhooks.c (default_mode_rep_extended): Likewise.
10670 (default_valid_pointer_mode): Likewise.
10671 (default_addr_space_valid_pointer_mode): Likewise.
10672 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10673 than a machine_mode.
10674 (default_libgcc_cmp_return_mode): Likewise.
10675 (default_libgcc_shift_count_mode): Likewise.
10676 (default_unwind_word_mode): Likewise.
10677 (default_addr_space_pointer_mode): Likewise.
10678 (default_addr_space_address_mode): Likewise.
10679 * targhooks.h (default_eh_return_filter_mode): Likewise.
10680 (default_libgcc_cmp_return_mode): Likewise.
10681 (default_libgcc_shift_count_mode): Likewise.
10682 (default_unwind_word_mode): Likewise.
10683 (default_addr_space_pointer_mode): Likewise.
10684 (default_addr_space_address_mode): Likewise.
10685 (default_mode_rep_extended): Take a scalar_int_mode rather than
10687 (default_valid_pointer_mode): Likewise.
10688 (default_addr_space_valid_pointer_mode): Likewise.
10689 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10691 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10692 rather than a machine_mode.
10693 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10694 for local variables.
10695 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10696 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10697 than a machine_mode.
10699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10700 Alan Hayward <alan.hayward@arm.com>
10701 David Sherwood <david.sherwood@arm.com>
10703 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10704 the mode argument to scalar_int_mode.
10705 (do_jump_by_parts_zero_rtx): Likewise.
10706 (do_jump_by_parts_equality_rtx): Likewise.
10707 (do_jump_by_parts_greater): Take a mode argument.
10708 (do_jump_by_parts_equality): Likewise.
10709 (do_jump_1): Update calls accordingly.
10711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10712 Alan Hayward <alan.hayward@arm.com>
10713 David Sherwood <david.sherwood@arm.com>
10715 * is-a.h (safe_dyn_cast): New function.
10716 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10717 (jump_table_for_label): Likewise.
10718 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10719 instead of an rtx_insn *.
10720 (shorten_branches): Use dyn_cast instead of LABEL_P and
10721 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10722 rtx_jump_table_data::get_data_mode.
10723 (final_scan_insn): Likewise.
10725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10726 Alan Hayward <alan.hayward@arm.com>
10727 David Sherwood <david.sherwood@arm.com>
10729 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10730 trying to combine a full-register integer set with a subreg
10733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10734 Alan Hayward <alan.hayward@arm.com>
10735 David Sherwood <david.sherwood@arm.com>
10737 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10738 that is always either address_mode or pointer_mode.
10740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10741 Alan Hayward <alan.hayward@arm.com>
10742 David Sherwood <david.sherwood@arm.com>
10744 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10745 when the two are known to be equal.
10747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10748 Alan Hayward <alan.hayward@arm.com>
10749 David Sherwood <david.sherwood@arm.com>
10751 * simplify-rtx.c (simplify_const_unary_operation): Use
10752 is_a <scalar_int_mode> instead of checking for a nonzero
10753 precision. Forcibly convert op_mode to a scalar_int_mode
10754 in that case. More clearly differentiate the operand and
10755 result modes and use the former when deciding what the value
10756 of a count-bits operation should be. Use is_int_mode instead
10757 of checking for a MODE_INT. Remove redundant check for whether
10758 this mode has a zero precision.
10760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10761 Alan Hayward <alan.hayward@arm.com>
10762 David Sherwood <david.sherwood@arm.com>
10764 * optabs.c (widen_leading): Change the type of the mode argument
10765 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10766 (widen_bswap): Likewise.
10767 (expand_parity): Likewise.
10768 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10769 (expand_ffs): Likewise.
10770 (epand_unop): Check for scalar integer modes before calling the
10773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10774 Alan Hayward <alan.hayward@arm.com>
10775 David Sherwood <david.sherwood@arm.com>
10777 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10779 (expand_expr_real_1): Update call accordingly.
10781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10782 Alan Hayward <alan.hayward@arm.com>
10783 David Sherwood <david.sherwood@arm.com>
10785 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10786 value_mode arguments. Use scalar_int_mode internally.
10787 (store_bit_field_1): Rename the new integer mode from imode
10788 to op0_mode and use it instead of GET_MODE (op0). Update calls
10789 to store_split_bit_field, store_bit_field_using_insv and
10790 store_fixed_bit_field.
10791 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10792 Use scalar_int_mode internally. Use a bit count rather than a mode
10793 when calculating the largest bit size for get_best_mode.
10794 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10795 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10796 Remove assertion that OP0 has a scalar integer mode.
10797 (store_split_bit_field): Add op0_mode and value_mode arguments.
10798 Update calls to extract_fixed_bit_field.
10799 (extract_bit_field_using_extv): Add an op0_mode argument.
10800 Use scalar_int_mode internally.
10801 (extract_bit_field_1): Rename the new integer mode from imode to
10802 op0_mode and use it instead of GET_MODE (op0). Update calls to
10803 extract_split_bit_field, extract_bit_field_using_extv and
10804 extract_fixed_bit_field.
10805 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10806 to extract_split_bit_field and extract_fixed_bit_field_1.
10807 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10808 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10809 on the target mode.
10810 (extract_split_bit_field): Add an op0_mode argument. Update call
10811 to extract_fixed_bit_field.
10813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10814 Alan Hayward <alan.hayward@arm.com>
10815 David Sherwood <david.sherwood@arm.com>
10817 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10818 * explow.c (hard_function_value): Likewise.
10819 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10820 convert_to_mode call outside the loop.
10821 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10822 for the mode iterator. Require the mode specified by max_pieces
10824 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10826 (copy_blkmode_to_reg): Likewise.
10827 (set_storage_via_setmem): Likewise.
10828 * optabs.c (prepare_cmp_insn): Likewise.
10829 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10830 * stor-layout.c (finish_bitfield_representative): Likewise.
10832 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10833 Alan Hayward <alan.hayward@arm.com>
10834 David Sherwood <david.sherwood@arm.com>
10836 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10837 * expr.c (convert_move): Use them.
10838 (convert_modes): Likewise.
10839 (store_expr_with_bounds): Likewise.
10841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10842 Alan Hayward <alan.hayward@arm.com>
10843 David Sherwood <david.sherwood@arm.com>
10845 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10846 parameter for the mode of "x". Remove the "known_x", "known_mode"
10847 and "known_ret" arguments. Change the type of the mode argument
10848 to scalar_int_mode.
10849 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10850 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10851 (reg_num_sign_bit_copies_for_combine): Likewise.
10852 * rtlanal.c (nonzero_bits1): Likewise.
10853 (num_sign_bit_copies1): Likewise.
10854 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10855 (reg_num_sign_bit_copies_general): Likewise.
10856 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10857 (reg_nonzero_bits_general): Likewise.
10859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10860 Alan Hayward <alan.hayward@arm.com>
10861 David Sherwood <david.sherwood@arm.com>
10863 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10864 than in subroutines. Return 1 for non-integer modes.
10865 (cached_num_sign_bit_copies): Change the type of the mode parameter
10866 to scalar_int_mode.
10867 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10868 classes. Handle CONST_INT_P first and then check whether X also
10869 has a scalar integer mode. Check the same thing for inner registers
10870 of a SUBREG and for values that are being extended or truncated.
10872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10873 Alan Hayward <alan.hayward@arm.com>
10874 David Sherwood <david.sherwood@arm.com>
10876 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10877 in subroutines. Return the mode mask for non-integer modes.
10878 (cached_nonzero_bits): Change the type of the mode parameter
10879 to scalar_int_mode.
10880 (nonzero_bits1): Likewise. Remove early exit for other mode
10881 classes. Handle CONST_INT_P first and then check whether X
10882 also has a scalar integer mode.
10884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10885 Alan Hayward <alan.hayward@arm.com>
10886 David Sherwood <david.sherwood@arm.com>
10888 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10889 Return a scalar_int_mode and assert that the size is greater than
10891 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10892 (op_by_pieces_d::op_by_pieces_d): Likewise.
10893 (op_by_pieces_d::run): Likewise.
10894 (can_store_by_pieces): Likewise.
10896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10897 Alan Hayward <alan.hayward@arm.com>
10898 David Sherwood <david.sherwood@arm.com>
10900 * combine.c (extract_left_shift): Add a mode argument and update
10902 (make_compound_operation_int): Change the type of the mode parameter
10903 to scalar_int_mode and update the call to extract_left_shift.
10905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10906 Alan Hayward <alan.hayward@arm.com>
10907 David Sherwood <david.sherwood@arm.com>
10909 * combine.c (simplify_and_const_int): Change the type of the mode
10910 parameter to scalar_int_mode.
10911 (simplify_and_const_int_1): Likewise. Update recursive call.
10913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10914 Alan Hayward <alan.hayward@arm.com>
10915 David Sherwood <david.sherwood@arm.com>
10917 * combine.c (simplify_compare_const): Check that the mode is a
10918 scalar_int_mode (rather than VOIDmode) before testing its
10920 (simplify_comparison): Move COMPARISON_P handling out of the
10921 loop and restrict the latter part of the loop to scalar_int_modes.
10922 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10923 and when considering SUBREG_REGs. Use is_int_mode instead of
10924 checking GET_MODE_CLASS against MODE_INT.
10926 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10927 Alan Hayward <alan.hayward@arm.com>
10928 David Sherwood <david.sherwood@arm.com>
10930 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10931 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10932 shift_unit_mode and for modes involved in scalar shifts.
10934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10935 Alan Hayward <alan.hayward@arm.com>
10936 David Sherwood <david.sherwood@arm.com>
10938 * combine.c (force_int_to_mode): New function, split out from...
10939 (force_to_mode): ...here. Keep xmode up-to-date and use it
10940 instead of GET_MODE (x).
10942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10943 Alan Hayward <alan.hayward@arm.com>
10944 David Sherwood <david.sherwood@arm.com>
10946 * optabs-query.h (extraction_insn::struct_mode): Change type to
10947 opt_scalar_int_mode and update comment.
10948 (extraction_insn::field_mode): Change type to scalar_int_mode.
10949 (extraction_insn::pos_mode): Likewise.
10950 * combine.c (make_extraction): Update accordingly.
10951 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10952 (get_optab_extraction_insn): Likewise.
10953 * recog.c (simplify_while_replacing): Likewise.
10954 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10955 parameter to opt_scalar_int_mode.
10957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10958 Alan Hayward <alan.hayward@arm.com>
10959 David Sherwood <david.sherwood@arm.com>
10961 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10962 to a scalar_int_mode instead of a machine_mode.
10963 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10964 (get_best_mode): Return a boolean and use a pointer argument to store
10965 the selected mode. Replace the limit mode parameter with a bit limit.
10966 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10967 for the values returned by bit_field_mode_iterator::next_mode.
10968 (store_bit_field): Update call to get_best_mode.
10969 (store_fixed_bit_field): Likewise.
10970 (extract_fixed_bit_field): Likewise.
10971 * expr.c (optimize_bitfield_assignment_op): Likewise.
10972 * fold-const.c (optimize_bit_field_compare): Likewise.
10973 (fold_truth_andor_1): Likewise.
10974 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10975 Update for new type of m_mode.
10976 (get_best_mode): As above.
10978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10979 Alan Hayward <alan.hayward@arm.com>
10980 David Sherwood <david.sherwood@arm.com>
10982 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10983 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10984 (store_bit_field): Check is_a <scalar_int_mode> before calling
10985 strict_volatile_bitfield_p.
10986 (extract_bit_field): Likewise.
10988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10989 Alan Hayward <alan.hayward@arm.com>
10990 David Sherwood <david.sherwood@arm.com>
10992 * target.def (cstore_mode): Return a scalar_int_mode.
10993 * doc/tm.texi: Regenerate.
10994 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10995 * targhooks.h (default_cstore_mode): Likewise.
10996 * targhooks.c (default_cstore_mode): Likewise, using a forced
10998 * expmed.c (emit_cstore): Expect the target of the cstore to be
11001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11002 Alan Hayward <alan.hayward@arm.com>
11003 David Sherwood <david.sherwood@arm.com>
11005 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11007 (niter_desc): Likewise mode.
11008 (iv_analyze): Add a mode parameter.
11010 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11011 and change its type to scalar_int_mode.
11012 * loop-iv.c: Update commentary at head of file.
11013 (iv_constant): Pass the mode paraeter before the rtx it describes
11014 and change its type to scalar_int_mode. Remove VOIDmode handling.
11015 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11016 (iv_extend): Likewise.
11017 (shorten_into_mode): Likewise.
11018 (iv_add): Use scalar_int_mode.
11019 (iv_mult): Likewise.
11020 (iv_shift): Likewise.
11021 (canonicalize_iv_subregs): Likewise.
11022 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11023 it describes and change its mode to scalar_int_mode. Also change
11024 the type of the returned inner_mode to scalar_int_mode.
11025 (get_biv_step): Likewise, turning outer_mode from a pointer
11026 into a direct parameter. Update call to get_biv_step_1.
11027 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11028 iv_constant and get_biv_step.
11029 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11030 and change its type to scalar_int_mode. Don't initialise iv->mode
11031 to VOIDmode and remove later checks for its still being VOIDmode.
11032 Update calls to iv_analyze_op and iv_analyze_expr. Check
11033 is_a <scalar_int_mode> when changing the mode under consideration.
11034 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11035 Update call to iv_analyze_expr.
11036 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11037 inner register is not also a scalar_int_mode. Update call to
11039 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11040 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11041 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11042 separate mode class checks. Update calls to iv_analyze. Remove
11043 fix-up of VOIDmodes after iv_analyze_biv.
11044 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11045 don't have a scalar_int_mode. Update call to biv_p.
11047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11048 Alan Hayward <alan.hayward@arm.com>
11049 David Sherwood <david.sherwood@arm.com>
11051 * cfgexpand.c (convert_debug_memory_address): Use
11052 as_a <scalar_int_mode>.
11053 * combine.c (expand_compound_operation): Likewise.
11054 (make_extraction): Likewise.
11055 (change_zero_ext): Likewise.
11056 (simplify_comparison): Likewise.
11057 * cse.c (cse_insn): Likewise.
11058 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11059 (mem_loc_descriptor): Likewise.
11060 (loc_descriptor): Likewise.
11061 * expmed.c (init_expmed_one_mode): Likewise.
11062 (synth_mult): Likewise.
11063 (emit_store_flag_1): Likewise.
11064 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11065 of a comparison with size.
11066 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11067 (reduce_to_bit_field_precision): Likewise.
11068 * function.c (expand_function_end): Likewise.
11069 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11070 * loop-doloop.c (doloop_modify): Likewise.
11071 * optabs.c (expand_binop): Likewise.
11072 (expand_unop): Likewise.
11073 (expand_copysign_absneg): Likewise.
11074 (prepare_cmp_insn): Likewise.
11075 (maybe_legitimize_operand): Likewise.
11076 * recog.c (const_scalar_int_operand): Likewise.
11077 * rtlanal.c (get_address_mode): Likewise.
11078 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11079 (simplify_cond_clz_ctz): Likewise.
11080 * tree-nested.c (get_nl_goto_field): Likewise.
11081 * tree.c (build_vector_type_for_mode): Likewise.
11082 * var-tracking.c (use_narrower_mode): Likewise.
11084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11085 Alan Hayward <alan.hayward@arm.com>
11086 David Sherwood <david.sherwood@arm.com>
11088 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11089 * builtins.c (expand_builtin_signbit): Use it.
11090 * cfgexpand.c (expand_debug_expr): Likewise.
11091 * dojump.c (do_jump): Likewise.
11092 (do_compare_and_jump): Likewise.
11093 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11094 * expmed.c (make_tree): Likewise.
11095 * expr.c (expand_expr_real_2): Likewise.
11096 (expand_expr_real_1): Likewise.
11097 (try_casesi): Likewise.
11098 * fold-const-call.c (fold_const_call_ss): Likewise.
11099 * fold-const.c (unextend): Likewise.
11100 (extract_muldiv_1): Likewise.
11101 (fold_single_bit_test): Likewise.
11102 (native_encode_int): Likewise.
11103 (native_encode_string): Likewise.
11104 (native_interpret_int): Likewise.
11105 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11106 * internal-fn.c (expand_addsub_overflow): Likewise.
11107 (expand_neg_overflow): Likewise.
11108 (expand_mul_overflow): Likewise.
11109 (expand_arith_overflow): Likewise.
11110 * match.pd: Likewise.
11111 * stor-layout.c (layout_type): Likewise.
11112 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11113 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11114 * tree-ssanames.c (get_range_info): Likewise.
11115 * tree-switch-conversion.c (array_value_type) Likewise.
11116 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11117 (vect_recog_divmod_pattern): Likewise.
11118 (vect_recog_mixed_size_cond_pattern): Likewise.
11119 * tree-vrp.c (extract_range_basic): Likewise.
11120 (simplify_float_conversion_using_ranges): Likewise.
11121 * tree.c (int_fits_type_p): Likewise.
11122 * ubsan.c (instrument_bool_enum_load): Likewise.
11123 * varasm.c (mergeable_string_section): Likewise.
11124 (narrowing_initializer_constant_valid_p): Likewise.
11125 (output_constant): Likewise.
11127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11128 Alan Hayward <alan.hayward@arm.com>
11129 David Sherwood <david.sherwood@arm.com>
11131 * machmode.h (NARROWEST_INT_MODE): New macro.
11132 * expr.c (alignment_for_piecewise_move): Use it instead of
11133 GET_CLASS_NARROWEST_MODE (MODE_INT).
11134 (push_block): Likewise.
11135 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11137 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11140 Alan Hayward <alan.hayward@arm.com>
11141 David Sherwood <david.sherwood@arm.com>
11143 * postreload.c (move2add_valid_value_p): Change the type of the
11144 mode parameter to scalar_int_mode.
11145 (move2add_use_add2_insn): Add a mode parameter and use it instead
11147 (move2add_use_add3_insn): Likewise.
11148 (reload_cse_move2add): Update accordingly.
11150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11151 Alan Hayward <alan.hayward@arm.com>
11152 David Sherwood <david.sherwood@arm.com>
11154 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11156 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11157 * optabs.c (expand_unop): Likewise.
11159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11160 Alan Hayward <alan.hayward@arm.com>
11161 David Sherwood <david.sherwood@arm.com>
11163 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11164 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11165 (popcount_loc_descriptor): Likewise.
11166 (bswap_loc_descriptor): Likewise.
11167 (rotate_loc_descriptor): Likewise.
11168 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11169 calling the functions above.
11171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11172 Alan Hayward <alan.hayward@arm.com>
11173 David Sherwood <david.sherwood@arm.com>
11175 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11177 (try_combine): Likewise.
11178 (simplify_if_then_else): Likewise.
11179 * cse.c (cse_insn): Likewise.
11180 * dwarf2out.c (mem_loc_descriptor): Likewise.
11181 * emit-rtl.c (gen_lowpart_common): Likewise.
11182 * simplify-rtx.c (simplify_truncation): Likewise.
11183 (simplify_binary_operation_1): Likewise.
11184 (simplify_const_relational_operation): Likewise.
11185 (simplify_ternary_operation): Likewise.
11186 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11189 Alan Hayward <alan.hayward@arm.com>
11190 David Sherwood <david.sherwood@arm.com>
11192 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11193 * reload.c (push_reload): Likewise.
11194 (find_reloads): Likewise.
11196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11197 Alan Hayward <alan.hayward@arm.com>
11198 David Sherwood <david.sherwood@arm.com>
11200 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11201 (make_compound_operation_int): Likewise.
11202 (change_zero_ext): Likewise.
11203 * expr.c (convert_move): Likewise.
11204 (convert_modes): Likewise.
11205 * fwprop.c (forward_propagate_subreg): Likewise.
11206 * loop-iv.c (get_biv_step_1): Likewise.
11207 * optabs.c (widen_operand): Likewise.
11208 * postreload.c (move2add_valid_value_p): Likewise.
11209 * recog.c (simplify_while_replacing): Likewise.
11210 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11211 (simplify_binary_operation_1): Likewise. Remove redundant
11212 mode equality check.
11214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11215 Alan Hayward <alan.hayward@arm.com>
11216 David Sherwood <david.sherwood@arm.com>
11218 * combine.c (combine_simplify_rtx): Add checks for
11219 is_a <scalar_int_mode>.
11220 (simplify_if_then_else): Likewise.
11221 (make_field_assignment): Likewise.
11222 (simplify_comparison): Likewise.
11223 * ifcvt.c (noce_try_bitop): Likewise.
11224 * loop-invariant.c (canonicalize_address_mult): Likewise.
11225 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11228 Alan Hayward <alan.hayward@arm.com>
11229 David Sherwood <david.sherwood@arm.com>
11231 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11232 is_a <scalar_int_mode> instead of != BLKmode.
11234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11235 Alan Hayward <alan.hayward@arm.com>
11236 David Sherwood <david.sherwood@arm.com>
11238 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11239 instead of != VOIDmode.
11240 * combine.c (if_then_else_cond): Likewise.
11241 (change_zero_ext): Likewise.
11242 * dwarf2out.c (mem_loc_descriptor): Likewise.
11243 (loc_descriptor): Likewise.
11244 * rtlanal.c (canonicalize_condition): Likewise.
11245 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11248 Alan Hayward <alan.hayward@arm.com>
11249 David Sherwood <david.sherwood@arm.com>
11251 * simplify-rtx.c (simplify_binary_operation_1): Use
11252 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11255 Alan Hayward <alan.hayward@arm.com>
11256 David Sherwood <david.sherwood@arm.com>
11258 * wide-int.h (int_traits<unsigned char>) New class.
11259 (int_traits<unsigned short>) Likewise.
11260 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11261 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11263 * combine.c (set_nonzero_bits_and_sign_copies): Use
11264 is_a <scalar_int_mode>.
11265 (find_split_point): Likewise.
11266 (combine_simplify_rtx): Likewise.
11267 (simplify_logical): Likewise.
11268 (expand_compound_operation): Likewise.
11269 (expand_field_assignment): Likewise.
11270 (make_compound_operation): Likewise.
11271 (extended_count): Likewise.
11272 (change_zero_ext): Likewise.
11273 (simplify_comparison): Likewise.
11274 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11275 (ucompare_loc_descriptor): Likewise.
11276 (minmax_loc_descriptor): Likewise.
11277 (mem_loc_descriptor): Likewise.
11278 (loc_descriptor): Likewise.
11279 * expmed.c (init_expmed_one_mode): Likewise.
11280 * lra-constraints.c (lra_constraint_offset): Likewise.
11281 * optabs.c (prepare_libcall_arg): Likewise.
11282 * postreload.c (move2add_note_store): Likewise.
11283 * reload.c (operands_match_p): Likewise.
11284 * rtl.h (load_extend_op): Likewise.
11285 * rtlhooks.c (gen_lowpart_general): Likewise.
11286 * simplify-rtx.c (simplify_truncation): Likewise.
11287 (simplify_unary_operation_1): Likewise.
11288 (simplify_binary_operation_1): Likewise.
11289 (simplify_const_binary_operation): Likewise.
11290 (simplify_const_relational_operation): Likewise.
11291 (simplify_subreg): Likewise.
11292 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11293 * var-tracking.c (adjust_mems): Likewise.
11294 (prepare_call_arguments): Likewise.
11296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11297 Alan Hayward <alan.hayward@arm.com>
11298 David Sherwood <david.sherwood@arm.com>
11300 * machmode.h (is_int_mode): New fuction.
11301 * combine.c (find_split_point): Use it.
11302 (combine_simplify_rtx): Likewise.
11303 (simplify_if_then_else): Likewise.
11304 (simplify_set): Likewise.
11305 (simplify_shift_const_1): Likewise.
11306 (simplify_comparison): Likewise.
11307 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11308 * cse.c (notreg_cost): Likewise.
11309 (cse_insn): Likewise.
11310 * cselib.c (cselib_lookup_1): Likewise.
11311 * dojump.c (do_jump_1): Likewise.
11312 (do_compare_rtx_and_jump): Likewise.
11313 * dse.c (get_call_args): Likewise.
11314 * dwarf2out.c (rtl_for_decl_init): Likewise.
11315 (native_encode_initializer): Likewise.
11316 * expmed.c (emit_store_flag_1): Likewise.
11317 (emit_store_flag): Likewise.
11318 * expr.c (convert_modes): Likewise.
11319 (store_field): Likewise.
11320 (expand_expr_real_1): Likewise.
11321 * fold-const.c (fold_read_from_constant_string): Likewise.
11322 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11323 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11324 * optabs.c (expand_binop): Likewise.
11325 (expand_unop): Likewise.
11326 (expand_abs_nojump): Likewise.
11327 (expand_one_cmpl_abs_nojump): Likewise.
11328 * simplify-rtx.c (mode_signbit_p): Likewise.
11329 (val_signbit_p): Likewise.
11330 (val_signbit_known_set_p): Likewise.
11331 (val_signbit_known_clear_p): Likewise.
11332 (simplify_relational_operation_1): Likewise.
11333 * tree.c (vector_type_mode): Likewise.
11335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11336 Alan Hayward <alan.hayward@arm.com>
11337 David Sherwood <david.sherwood@arm.com>
11339 * machmode.h (smallest_mode_for_size): Fix formatting.
11340 (smallest_int_mode_for_size): New function.
11341 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11342 instead of smallest_mode_for_size.
11343 * combine.c (make_extraction): Likewise.
11344 * config/arc/arc.c (arc_expand_movmem): Likewise.
11345 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11346 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11347 * config/s390/s390.c (s390_expand_insv): Likewise.
11348 * config/sparc/sparc.c (assign_int_registers): Likewise.
11349 * config/spu/spu.c (spu_function_value): Likewise.
11350 (spu_function_arg): Likewise.
11351 * coverage.c (get_gcov_type): Likewise.
11352 (get_gcov_unsigned_t): Likewise.
11353 * dse.c (find_shift_sequence): Likewise.
11354 * expmed.c (store_bit_field_1): Likewise.
11355 * expr.c (convert_move): Likewise.
11356 (store_field): Likewise.
11357 * internal-fn.c (expand_arith_overflow): Likewise.
11358 * optabs-query.c (get_best_extraction_insn): Likewise.
11359 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11360 * stor-layout.c (layout_type): Likewise.
11361 (initialize_sizetypes): Likewise.
11362 * targhooks.c (default_get_mask_mode): Likewise.
11363 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11366 Alan Hayward <alan.hayward@arm.com>
11367 David Sherwood <david.sherwood@arm.com>
11369 * machmode.h (opt_mode::else_blk): New function.
11370 (int_mode_for_mode): Declare.
11371 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11372 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11374 * cfgexpand.c (expand_debug_expr): Likewise.
11375 * combine.c (gen_lowpart_or_truncate): Likewise.
11376 (gen_lowpart_for_combine): Likewise.
11377 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11378 * config/avr/avr.c (avr_to_int_mode): Likewise.
11379 (avr_out_plus_1): Likewise.
11380 (avr_out_plus): Likewise.
11381 (avr_out_round): Likewise.
11382 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11383 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11384 (s390_expand_vcond): Likewise.
11385 * config/spu/spu.c (spu_split_immediate): Likewise.
11386 (spu_expand_mov): Likewise.
11387 * dse.c (get_stored_val): Likewise.
11388 * expmed.c (store_bit_field_1): Likewise.
11389 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11391 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11392 (extract_low_bits): Likewise.
11393 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11394 handling rather than repeating the check.
11395 (emit_group_store): Likewise.
11396 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11397 * optabs.c (expand_absneg_bit): Likewise.
11398 (expand_copysign_absneg): Likewise.
11399 (expand_copysign_bit): Likewise.
11400 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11401 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11402 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11403 * var-tracking.c (prepare_call_arguments): Likewise.
11404 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11405 int_mode_for_mode instead of mode_for_size.
11406 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11409 Alan Hayward <alan.hayward@arm.com>
11410 David Sherwood <david.sherwood@arm.com>
11412 * machmode.h (int_mode_for_size): New function.
11413 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11414 instead of mode_for_size.
11415 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11417 * combine.c (expand_field_assignment): Use int_mode_for_size
11418 instead of mode_for_size.
11419 (make_extraction): Likewise.
11420 (simplify_shift_const_1): Likewise.
11421 (simplify_comparison): Likewise.
11422 * dojump.c (do_jump): Likewise.
11423 * dwarf2out.c (mem_loc_descriptor): Likewise.
11424 * emit-rtl.c (init_derived_machine_modes): Likewise.
11425 * expmed.c (flip_storage_order): Likewise.
11426 (convert_extracted_bit_field): Likewise.
11427 * expr.c (copy_blkmode_from_reg): Likewise.
11428 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11429 * internal-fn.c (expand_mul_overflow): Likewise.
11430 * lower-subreg.c (simple_move): Likewise.
11431 * optabs-libfuncs.c (init_optabs): Likewise.
11432 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11433 * tree.c (vector_type_mode): Likewise.
11434 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11435 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11436 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11437 * tree-vect-stmts.c (vectorizable_load): Likewise.
11438 (vectorizable_store): Likewise.
11440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11441 Alan Hayward <alan.hayward@arm.com>
11442 David Sherwood <david.sherwood@arm.com>
11444 * coretypes.h (pod_mode): New type.
11445 (scalar_int_mode_pod): New typedef.
11446 * machmode.h (pod_mode): New class.
11447 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11448 * genmodes.c (emit_mode_int_n): Update accordingly.
11449 * lower-subreg.h (target_lower_subreg): Change type to
11450 scalar_int_mode_pod.
11451 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11452 scalar_int_mode_pod.
11454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11455 Alan Hayward <alan.hayward@arm.com>
11456 David Sherwood <david.sherwood@arm.com>
11458 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11459 machine_mode to scalar_int_mode.
11460 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11461 (rs6000_option_override_internal): Remove cast to int.
11462 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11463 machine_mode to scalar_int_mode.
11464 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11465 (rs6000_option_override_internal): Remove cast to int.
11466 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11467 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11469 * config/s390/s390.c (s390_expand_builtin): Likewise.
11470 * coretypes.h (scalar_int_mode): New type.
11471 (opt_scalar_int_mode): New typedef.
11472 * machmode.h (scalar_int_mode): New class.
11473 (scalar_int_mode::includes_p): New function.
11474 (byte_mode): Change type to scalar_int_mode.
11475 (word_mode): Likewise.
11476 (ptr_mode): Likewise.
11477 * emit-rtl.c (byte_mode): Likewise.
11478 (word_mode): Likewise.
11479 (ptr_mode): Likewise.
11480 (init_derived_machine_modes): Update accordingly.
11481 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11482 and MODE_PARTIAL_INT.
11483 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11484 opt_scalar_int_mode.
11486 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11487 Alan Hayward <alan.hayward@arm.com>
11488 David Sherwood <david.sherwood@arm.com>
11490 * target.def (libgcc_floating_mode_supported_p): Take a
11492 * doc/tm.texi: Regenerate.
11493 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11495 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11496 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11500 Alan Hayward <alan.hayward@arm.com>
11501 David Sherwood <david.sherwood@arm.com>
11503 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11504 * doc/tm.texi: Regenerate.
11505 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11506 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11507 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11508 * targhooks.h (default_floatn_mode): Likewise.
11509 * targhooks.c (default_floatn_mode): Likewise.
11510 * tree.c (build_common_tree_nodes): Update accordingly.
11512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11513 Alan Hayward <alan.hayward@arm.com>
11514 David Sherwood <david.sherwood@arm.com>
11516 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11517 (mode_iterator::iterate_p): Likewise.
11518 (mode_iterator::get_wider): Likewise.
11519 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11522 Alan Hayward <alan.hayward@arm.com>
11523 David Sherwood <david.sherwood@arm.com>
11525 * coretypes.h (opt_scalar_float_mode): New typedef.
11526 * machmode.h (float_mode_for_size): New function.
11527 * emit-rtl.c (double_mode): Delete.
11528 (init_emit_once): Use float_mode_for_size.
11529 * stor-layout.c (layout_type): Likewise.
11530 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11533 Alan Hayward <alan.hayward@arm.com>
11534 David Sherwood <david.sherwood@arm.com>
11536 * output.h (assemble_real): Take a scalar_float_mode.
11537 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11538 * config/arm/arm.md (consttable_4): Likewise.
11539 (consttable_8): Likewise.
11540 (consttable_16): Likewise.
11541 * config/mips/mips.md (consttable_float): Likewise.
11542 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11543 * varasm.c (assemble_real): Take a scalar_float_mode.
11544 (output_constant_pool_2): Update accordingly.
11545 (output_constant): Likewise.
11547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11548 Alan Hayward <alan.hayward@arm.com>
11549 David Sherwood <david.sherwood@arm.com>
11551 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11552 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11553 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11554 (native_encode_real): Likewise.
11555 (native_interpret_real): Likewise.
11556 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11557 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11560 Alan Hayward <alan.hayward@arm.com>
11561 David Sherwood <david.sherwood@arm.com>
11563 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11564 <scalar_float_mode>. Simplify.
11565 (gen_extend_conv_libfunc): Likewise.
11567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11568 Alan Hayward <alan.hayward@arm.com>
11569 David Sherwood <david.sherwood@arm.com>
11571 * coretypes.h (scalar_float_mode): New type.
11572 * machmode.h (mode_traits::from_int): Use machine_mode if
11573 USE_ENUM_MODES is defined.
11574 (is_a): New function.
11576 (dyn_cast): Likewise.
11577 (scalar_float_mode): New class.
11578 (scalar_float_mode::includes_p): New function.
11579 (is_float_mode): Likewise.
11580 * gdbhooks.py (MachineModePrinter): New class.
11581 (build_pretty_printer): Use it for scalar_float_mode.
11582 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11583 (format_helper::format_helper): Turn into a template.
11584 * genmodes.c (get_mode_class): New function.
11585 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11586 or machine_mode if none.
11587 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11588 as_a <scalar_float_mode>.
11589 * dwarf2out.c (mem_loc_descriptor): Likewise.
11590 (insert_float): Likewise.
11591 (add_const_value_attribute): Likewise.
11592 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11593 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11594 (expand_unop): Update accordingly.
11595 (expand_abs_nojump): Likewise.
11596 (expand_copysign_absneg): Take a scalar_float_mode.
11597 (expand_copysign_bit): Likewise.
11598 (expand_copysign): Update accordingly.
11600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11601 Alan Hayward <alan.hayward@arm.com>
11602 David Sherwood <david.sherwood@arm.com>
11604 * coretypes.h (opt_mode): New class.
11605 * machmode.h (opt_mode): Likewise.
11606 (opt_mode::else_void): New function.
11607 (opt_mode::require): Likewise.
11608 (opt_mode::exists): Likewise.
11609 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11610 (GET_MODE_2XWIDER_MODE): Likewise.
11611 (mode_iterator::get_wider): Update accordingly.
11612 (mode_iterator::get_2xwider): Likewise.
11613 (mode_iterator::get_known_wider): Likewise, turning into a template.
11614 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11615 forcing a wider mode to exist.
11616 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11617 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11618 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11619 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11620 * lower-subreg.c (init_lower_subreg): Likewise.
11621 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11622 on the final iteration.
11623 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11624 a wider mode exists before asking for a move pattern.
11625 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11626 forcing a wider mode to exist.
11627 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11628 returning false if no such mode exists.
11629 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11630 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11631 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11632 Avoid checking for a MODE_INT if we already know the mode is not a
11634 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11635 forcing a wider mode to exist.
11636 (expmed_mult_highpart_optab): Likewise.
11637 (expmed_mult_highpart): Likewise.
11638 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11640 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11641 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11642 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11643 * internal-fn.c (expand_mul_overflow): Update use of
11644 GET_MODE_2XWIDER_MODE.
11645 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11646 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11647 GET_MODE_WIDER_MODE.
11648 (convert_plusminus_to_widen): Likewise.
11649 * tree-switch-conversion.c (array_value_type): Likewise.
11650 * var-tracking.c (emit_note_insn_var_location): Likewise.
11651 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11652 Return false inside rather than outside the loop if no wider mode
11654 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11655 and GET_MODE_2XWIDER_MODE
11656 (can_compare_p): Use else_void.
11657 * gdbhooks.py (OptMachineModePrinter): New class.
11658 (build_pretty_printer): Use it for opt_mode.
11660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11661 Alan Hayward <alan.hayward@arm.com>
11662 David Sherwood <david.sherwood@arm.com>
11664 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11665 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11668 Alan Hayward <alan.hayward@arm.com>
11669 David Sherwood <david.sherwood@arm.com>
11671 * machmode.h (mode_traits): New structure.
11672 (get_narrowest_mode): New function.
11673 (mode_iterator::start): Likewise.
11674 (mode_iterator::iterate_p): Likewise.
11675 (mode_iterator::get_wider): Likewise.
11676 (mode_iterator::get_known_wider): Likewise.
11677 (mode_iterator::get_2xwider): Likewise.
11678 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11679 (FOR_EACH_MODE): Likewise.
11680 (FOR_EACH_MODE_FROM): Likewise.
11681 (FOR_EACH_MODE_UNTIL): Likewise.
11682 (FOR_EACH_WIDER_MODE): Likewise.
11683 (FOR_EACH_2XWIDER_MODE): Likewise.
11684 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11685 * combine.c (simplify_comparison): Likewise
11686 * config/i386/i386.c (type_natural_mode): Likewise.
11687 * cse.c (cse_insn): Likewise.
11688 * dse.c (find_shift_sequence): Likewise.
11689 * emit-rtl.c (init_derived_machine_modes): Likewise.
11690 (init_emit_once): Likewise.
11691 * explow.c (hard_function_value): Likewise.
11692 * expmed.c (extract_fixed_bit_field_1): Likewise.
11693 (extract_bit_field_1): Likewise.
11694 (expand_divmod): Likewise.
11695 (emit_store_flag_1): Likewise.
11696 * expr.c (init_expr_target): Likewise.
11697 (convert_move): Likewise.
11698 (alignment_for_piecewise_move): Likewise.
11699 (widest_int_mode_for_size): Likewise.
11700 (emit_block_move_via_movmem): Likewise.
11701 (copy_blkmode_to_reg): Likewise.
11702 (set_storage_via_setmem): Likewise.
11703 (compress_float_constant): Likewise.
11704 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11705 * optabs-query.c (get_best_extraction_insn): Likewise.
11706 * optabs.c (expand_binop): Likewise.
11707 (expand_twoval_unop): Likewise.
11708 (expand_twoval_binop): Likewise.
11709 (widen_leading): Likewise.
11710 (widen_bswap): Likewise.
11711 (expand_parity): Likewise.
11712 (expand_unop): Likewise.
11713 (prepare_cmp_insn): Likewise.
11714 (prepare_float_lib_cmp): Likewise.
11715 (expand_float): Likewise.
11716 (expand_fix): Likewise.
11717 (expand_sfix_optab): Likewise.
11718 * postreload.c (move2add_use_add2_insn): Likewise.
11719 * reg-stack.c (reg_to_stack): Likewise.
11720 * reginfo.c (choose_hard_reg_mode): Likewise.
11721 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11722 * stor-layout.c (mode_for_size): Likewise.
11723 (smallest_mode_for_size): Likewise.
11724 (mode_for_vector): Likewise.
11725 (finish_bitfield_representative): Likewise.
11726 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11727 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11728 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11729 * var-tracking.c (prepare_call_arguments): Likewise.
11731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11732 Alan Hayward <alan.hayward@arm.com>
11733 David Sherwood <david.sherwood@arm.com>
11735 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11736 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11737 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11738 * machmode.h (mode_size): Move earlier in file.
11739 (mode_precision): Likewise.
11740 (mode_inner): Likewise.
11741 (mode_nunits): Likewise.
11742 (mode_unit_size): Likewise.
11743 (unit_unit_precision): Likewise.
11744 (mode_wider): Likewise.
11745 (mode_2xwider): Likewise.
11746 (machine_mode): New class.
11747 (mode_to_bytes): New function.
11748 (mode_to_bits): Likewise.
11749 (mode_to_precision): Likewise.
11750 (mode_to_inner): Likewise.
11751 (mode_to_unit_size): Likewise.
11752 (mode_to_unit_precision): Likewise.
11753 (mode_to_nunits): Likewise.
11754 (GET_MODE_SIZE): Use mode_to_bytes.
11755 (GET_MODE_BITSIZE): Use mode_to_bits.
11756 (GET_MODE_PRECISION): Use mode_to_precision.
11757 (GET_MODE_INNER): Use mode_to_inner.
11758 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11759 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11760 (GET_MODE_NUNITS): Use mode_to_nunits.
11761 * system.h (ALWAYS_INLINE): New macro.
11762 * config/powerpcspe/powerpcspe-c.c
11763 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11764 int for arg1_mode and arg2_mode.
11766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11767 Alan Hayward <alan.hayward@arm.com>
11768 David Sherwood <david.sherwood@arm.com>
11770 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11771 Prefix mode names with E_ in case statements.
11772 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11773 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11774 (aarch64_split_simd_move): Likewise.
11775 (aarch64_gen_storewb_pair): Likewise.
11776 (aarch64_gen_loadwb_pair): Likewise.
11777 (aarch64_gen_store_pair): Likewise.
11778 (aarch64_gen_load_pair): Likewise.
11779 (aarch64_get_condition_code_1): Likewise.
11780 (aarch64_constant_pool_reload_icode): Likewise.
11781 (get_rsqrte_type): Likewise.
11782 (get_rsqrts_type): Likewise.
11783 (get_recpe_type): Likewise.
11784 (get_recps_type): Likewise.
11785 (aarch64_gimplify_va_arg_expr): Likewise.
11786 (aarch64_simd_container_mode): Likewise.
11787 (aarch64_emit_load_exclusive): Likewise.
11788 (aarch64_emit_store_exclusive): Likewise.
11789 (aarch64_expand_compare_and_swap): Likewise.
11790 (aarch64_gen_atomic_cas): Likewise.
11791 (aarch64_emit_bic): Likewise.
11792 (aarch64_emit_atomic_swap): Likewise.
11793 (aarch64_emit_atomic_load_op): Likewise.
11794 (aarch64_evpc_trn): Likewise.
11795 (aarch64_evpc_uzp): Likewise.
11796 (aarch64_evpc_zip): Likewise.
11797 (aarch64_evpc_ext): Likewise.
11798 (aarch64_evpc_rev): Likewise.
11799 (aarch64_evpc_dup): Likewise.
11800 (aarch64_gen_ccmp_first): Likewise.
11801 (aarch64_gen_ccmp_next): Likewise.
11802 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11803 (alpha_emit_xfloating_libcall): Likewise.
11804 (emit_insxl): Likewise.
11805 (alpha_arg_type): Likewise.
11806 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11807 (arc_preferred_simd_mode): Likewise.
11808 (arc_secondary_reload): Likewise.
11809 (get_arc_condition_code): Likewise.
11810 (arc_print_operand): Likewise.
11811 (arc_legitimate_constant_p): Likewise.
11812 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11813 * config/arc/arc.md (casesi_load): Likewise.
11814 (casesi_compact_jump): Likewise.
11815 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11816 (cc_use_register): Likewise.
11817 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11818 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11819 (arm_init_iwmmxt_builtins): Likewise.
11820 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11821 (neon_expand_vector_init): Likewise.
11822 (arm_attr_length_move_neon): Likewise.
11823 (maybe_get_arm_condition_code): Likewise.
11824 (arm_emit_vector_const): Likewise.
11825 (arm_preferred_simd_mode): Likewise.
11826 (arm_output_iwmmxt_tinsr): Likewise.
11827 (thumb1_output_casesi): Likewise.
11828 (thumb2_output_casesi): Likewise.
11829 (arm_emit_load_exclusive): Likewise.
11830 (arm_emit_store_exclusive): Likewise.
11831 (arm_expand_compare_and_swap): Likewise.
11832 (arm_evpc_neon_vuzp): Likewise.
11833 (arm_evpc_neon_vzip): Likewise.
11834 (arm_evpc_neon_vrev): Likewise.
11835 (arm_evpc_neon_vtrn): Likewise.
11836 (arm_evpc_neon_vext): Likewise.
11837 (arm_validize_comparison): Likewise.
11838 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11839 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11840 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11841 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11842 (c6x_preferred_simd_mode): Likewise.
11843 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11844 (epiphany_rtx_costs): Likewise.
11845 * config/epiphany/predicates.md (proper_comparison_operator):
11847 * config/frv/frv.c (condexec_memory_operand): Likewise.
11848 (frv_emit_move): Likewise.
11849 (output_move_single): Likewise.
11850 (output_condmove_single): Likewise.
11851 (frv_hard_regno_mode_ok): Likewise.
11852 (frv_matching_accg_mode): Likewise.
11853 * config/h8300/h8300.c (split_adds_subs): Likewise.
11854 (h8300_rtx_costs): Likewise.
11855 (h8300_print_operand): Likewise.
11856 (compute_mov_length): Likewise.
11857 (output_logical_op): Likewise.
11858 (compute_logical_op_length): Likewise.
11859 (compute_logical_op_cc): Likewise.
11860 (h8300_shift_needs_scratch_p): Likewise.
11861 (output_a_shift): Likewise.
11862 (compute_a_shift_length): Likewise.
11863 (compute_a_shift_cc): Likewise.
11864 (expand_a_rotate): Likewise.
11865 (output_a_rotate): Likewise.
11866 * config/i386/i386.c (classify_argument): Likewise.
11867 (function_arg_advance_32): Likewise.
11868 (function_arg_32): Likewise.
11869 (function_arg_64): Likewise.
11870 (function_value_64): Likewise.
11871 (ix86_gimplify_va_arg): Likewise.
11872 (ix86_legitimate_constant_p): Likewise.
11873 (put_condition_code): Likewise.
11874 (split_double_mode): Likewise.
11875 (ix86_avx256_split_vector_move_misalign): Likewise.
11876 (ix86_expand_vector_logical_operator): Likewise.
11877 (ix86_split_idivmod): Likewise.
11878 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11879 (ix86_build_const_vector): Likewise.
11880 (ix86_build_signbit_mask): Likewise.
11881 (ix86_match_ccmode): Likewise.
11882 (ix86_cc_modes_compatible): Likewise.
11883 (ix86_expand_branch): Likewise.
11884 (ix86_expand_sse_cmp): Likewise.
11885 (ix86_expand_sse_movcc): Likewise.
11886 (ix86_expand_int_sse_cmp): Likewise.
11887 (ix86_expand_vec_perm_vpermi2): Likewise.
11888 (ix86_expand_vec_perm): Likewise.
11889 (ix86_expand_sse_unpack): Likewise.
11890 (ix86_expand_int_addcc): Likewise.
11891 (ix86_split_to_parts): Likewise.
11892 (ix86_vectorize_builtin_gather): Likewise.
11893 (ix86_vectorize_builtin_scatter): Likewise.
11894 (avx_vpermilp_parallel): Likewise.
11895 (inline_memory_move_cost): Likewise.
11896 (ix86_tieable_integer_mode_p): Likewise.
11897 (x86_maybe_negate_const_int): Likewise.
11898 (ix86_expand_vector_init_duplicate): Likewise.
11899 (ix86_expand_vector_init_one_nonzero): Likewise.
11900 (ix86_expand_vector_init_one_var): Likewise.
11901 (ix86_expand_vector_init_concat): Likewise.
11902 (ix86_expand_vector_init_interleave): Likewise.
11903 (ix86_expand_vector_init_general): Likewise.
11904 (ix86_expand_vector_set): Likewise.
11905 (ix86_expand_vector_extract): Likewise.
11906 (emit_reduc_half): Likewise.
11907 (ix86_emit_i387_round): Likewise.
11908 (ix86_mangle_type): Likewise.
11909 (ix86_expand_round_sse4): Likewise.
11910 (expand_vec_perm_blend): Likewise.
11911 (canonicalize_vector_int_perm): Likewise.
11912 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11913 (expand_vec_perm_1): Likewise.
11914 (expand_vec_perm_interleave3): Likewise.
11915 (expand_vec_perm_even_odd_pack): Likewise.
11916 (expand_vec_perm_even_odd_1): Likewise.
11917 (expand_vec_perm_broadcast_1): Likewise.
11918 (ix86_vectorize_vec_perm_const_ok): Likewise.
11919 (ix86_expand_vecop_qihi): Likewise.
11920 (ix86_expand_mul_widen_hilo): Likewise.
11921 (ix86_expand_sse2_abs): Likewise.
11922 (ix86_expand_pextr): Likewise.
11923 (ix86_expand_pinsr): Likewise.
11924 (ix86_preferred_simd_mode): Likewise.
11925 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11926 * config/i386/sse.md (*andnot<mode>3): Likewise.
11927 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11928 (*<code><mode>3): Likewise.
11929 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11930 (ia64_expand_atomic_op): Likewise.
11931 (ia64_arg_type): Likewise.
11932 (ia64_mode_to_int): Likewise.
11933 (ia64_scalar_mode_supported_p): Likewise.
11934 (ia64_vector_mode_supported_p): Likewise.
11935 (expand_vec_perm_broadcast): Likewise.
11936 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11937 (iq2000_function_arg_advance): Likewise.
11938 (iq2000_function_arg): Likewise.
11939 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11940 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11941 (m68k_libcall_value): Likewise.
11942 (m68k_function_value): Likewise.
11943 (sched_attr_op_type): Likewise.
11944 * config/mcore/mcore.c (mcore_output_move): Likewise.
11945 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11947 (microblaze_function_arg): Likewise.
11948 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11949 (mips_print_operand): Likewise.
11950 (mips_mode_ok_for_mov_fmt_p): Likewise.
11951 (mips_vector_mode_supported_p): Likewise.
11952 (mips_preferred_simd_mode): Likewise.
11953 (mips_expand_vpc_loongson_even_odd): Likewise.
11954 (mips_expand_vec_unpack): Likewise.
11955 (mips_expand_vi_broadcast): Likewise.
11956 (mips_expand_vector_init): Likewise.
11957 (mips_expand_vec_reduc): Likewise.
11958 (mips_expand_msa_cmp): Likewise.
11959 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11960 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11961 (cc_flags_for_mode): Likewise.
11962 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11963 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11964 (nds32_output_casesi_pc_relative): Likewise.
11965 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11966 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11967 (nvptx_gen_unpack): Likewise.
11968 (nvptx_gen_pack): Likewise.
11969 (nvptx_gen_shuffle): Likewise.
11970 (nvptx_gen_wcast): Likewise.
11971 (nvptx_preferred_simd_mode): Likewise.
11972 * config/pa/pa.c (pa_secondary_reload): Likewise.
11973 * config/pa/predicates.md (base14_operand): Likewise.
11974 * config/powerpcspe/powerpcspe-c.c
11975 (altivec_resolve_overloaded_builtin): Likewise.
11976 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11978 (rs6000_preferred_simd_mode): Likewise.
11979 (output_vec_const_move): Likewise.
11980 (rs6000_expand_vector_extract): Likewise.
11981 (rs6000_split_vec_extract_var): Likewise.
11982 (reg_offset_addressing_ok_p): Likewise.
11983 (rs6000_legitimate_offset_address_p): Likewise.
11984 (rs6000_legitimize_address): Likewise.
11985 (rs6000_emit_set_const): Likewise.
11986 (rs6000_const_vec): Likewise.
11987 (rs6000_emit_move): Likewise.
11988 (spe_build_register_parallel): Likewise.
11989 (rs6000_darwin64_record_arg_recurse): Likewise.
11990 (swap_selector_for_mode): Likewise.
11991 (spe_init_builtins): Likewise.
11992 (paired_init_builtins): Likewise.
11993 (altivec_init_builtins): Likewise.
11994 (do_load_for_compare): Likewise.
11995 (rs6000_generate_compare): Likewise.
11996 (rs6000_expand_float128_convert): Likewise.
11997 (emit_load_locked): Likewise.
11998 (emit_store_conditional): Likewise.
11999 (rs6000_output_function_epilogue): Likewise.
12000 (rs6000_handle_altivec_attribute): Likewise.
12001 (rs6000_function_value): Likewise.
12002 (emit_fusion_gpr_load): Likewise.
12003 (emit_fusion_p9_load): Likewise.
12004 (emit_fusion_p9_store): Likewise.
12005 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12006 (fusion_gpr_mem_load): Likewise.
12007 (fusion_addis_mem_combo_load): Likewise.
12008 (fusion_addis_mem_combo_store): Likewise.
12009 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12010 (fusion_gpr_mem_load): Likewise.
12011 (fusion_addis_mem_combo_load): Likewise.
12012 (fusion_addis_mem_combo_store): Likewise.
12013 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12015 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12016 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12017 (rs6000_preferred_simd_mode): Likewise.
12018 (output_vec_const_move): Likewise.
12019 (rs6000_expand_vector_extract): Likewise.
12020 (rs6000_split_vec_extract_var): Likewise.
12021 (reg_offset_addressing_ok_p): Likewise.
12022 (rs6000_legitimate_offset_address_p): Likewise.
12023 (rs6000_legitimize_address): Likewise.
12024 (rs6000_emit_set_const): Likewise.
12025 (rs6000_const_vec): Likewise.
12026 (rs6000_emit_move): Likewise.
12027 (rs6000_darwin64_record_arg_recurse): Likewise.
12028 (swap_selector_for_mode): Likewise.
12029 (paired_init_builtins): Likewise.
12030 (altivec_init_builtins): Likewise.
12031 (rs6000_expand_float128_convert): Likewise.
12032 (emit_load_locked): Likewise.
12033 (emit_store_conditional): Likewise.
12034 (rs6000_output_function_epilogue): Likewise.
12035 (rs6000_handle_altivec_attribute): Likewise.
12036 (rs6000_function_value): Likewise.
12037 (emit_fusion_gpr_load): Likewise.
12038 (emit_fusion_p9_load): Likewise.
12039 (emit_fusion_p9_store): Likewise.
12040 * config/rx/rx.c (rx_gen_move_template): Likewise.
12041 (flags_from_mode): Likewise.
12042 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12043 (s390_slb_comparison): Likewise.
12044 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12045 (s390_vector_mode_supported_p): Likewise.
12046 (s390_cc_modes_compatible): Likewise.
12047 (s390_match_ccmode_set): Likewise.
12048 (s390_canonicalize_comparison): Likewise.
12049 (s390_emit_compare_and_swap): Likewise.
12050 (s390_branch_condition_mask): Likewise.
12051 (s390_rtx_costs): Likewise.
12052 (s390_secondary_reload): Likewise.
12053 (__SECONDARY_RELOAD_CASE): Likewise.
12054 (s390_expand_cs): Likewise.
12055 (s390_preferred_simd_mode): Likewise.
12056 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12057 * config/sh/sh.c (sh_print_operand): Likewise.
12058 (dump_table): Likewise.
12059 (sh_secondary_reload): Likewise.
12060 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12061 * config/sh/sh.md (casesi_worker_1): Likewise.
12062 (casesi_worker_2): Likewise.
12063 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12064 (fcc_comparison_operator): Likewise.
12065 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12066 (emit_soft_tfmode_cvt): Likewise.
12067 (sparc_preferred_simd_mode): Likewise.
12068 (output_cbranch): Likewise.
12069 (sparc_print_operand): Likewise.
12070 (sparc_expand_vec_perm_bmask): Likewise.
12071 (vector_init_bshuffle): Likewise.
12072 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12073 (spu_vector_mode_supported_p): Likewise.
12074 (spu_expand_insv): Likewise.
12075 (spu_emit_branch_or_set): Likewise.
12076 (spu_handle_vector_attribute): Likewise.
12077 (spu_builtin_splats): Likewise.
12078 (spu_builtin_extract): Likewise.
12079 (spu_builtin_promote): Likewise.
12080 (spu_expand_sign_extend): Likewise.
12081 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12082 (tilegx_simd_int): Likewise.
12083 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12084 (tilepro_simd_int): Likewise.
12085 * config/v850/v850.c (const_double_split): Likewise.
12086 (v850_print_operand): Likewise.
12087 (ep_memory_offset): Likewise.
12088 * config/vax/vax.c (vax_rtx_costs): Likewise.
12089 (vax_output_int_move): Likewise.
12090 (vax_output_int_add): Likewise.
12091 (vax_output_int_subtract): Likewise.
12092 * config/visium/predicates.md (visium_branch_operator): Likewise.
12093 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12094 (visium_print_operand_address): Likewise.
12095 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12096 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12097 (xtensa_expand_conditional_branch): Likewise.
12098 (xtensa_copy_incoming_a7): Likewise.
12099 (xtensa_output_literal): Likewise.
12100 * dfp.c (decimal_real_maxval): Likewise.
12101 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12104 Alan Hayward <alan.hayward@arm.com>
12105 David Sherwood <david.sherwood@arm.com>
12107 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12108 (mode_nunits_inline): Likewise.
12109 (mode_inner_inline): Likewise.
12110 (mode_unit_size_inline): Likewise.
12111 (mode_unit_precision_inline): Likewise.
12112 (emit_insn_modes_h): Likewise. Also emit a #define of the
12114 (emit_mode_wider): Add an E_ prefix to mode names.
12115 (emit_mode_complex): Likewise.
12116 (emit_mode_inner): Likewise.
12117 (emit_mode_adjustments): Likewise.
12118 (emit_mode_int_n): Likewise.
12119 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12120 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12121 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12122 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12123 (CRC32_BUILTIN, ENTRY): Likewise.
12124 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12125 (aarch64_pop_regs): Likewise.
12126 (aarch64_process_components): Likewise.
12127 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12128 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12129 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12130 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12131 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12132 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12134 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12135 (pcmpistr): Likewise.
12136 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12137 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12138 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12140 * config/rl78/rl78.c (mduc_regs): Likewise.
12141 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12142 (htm_expand_builtin): Likewise.
12143 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12144 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12145 * config/xtensa/xtensa.c (print_operand): Likewise.
12146 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12147 (NUM_MODE_VECTOR_INT): Likewise.
12148 * genoutput.c (null_operand): Likewise.
12149 (output_operand_data): Likewise.
12150 * genrecog.c (print_parameter_value): Likewise.
12151 * lra.c (debug_operand_data): Likewise.
12153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12155 * dwarf2out.c (scompare_loc_descriptor_wide)
12156 (scompare_loc_descriptor_narrow): New functions, split out from...
12157 (scompare_loc_descriptor): ...here.
12158 * expmed.c (emit_store_flag_int): New function, split out from...
12159 (emit_store_flag): ...here.
12161 2017-08-30 Richard Biener <rguenther@suse.de>
12163 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12164 (dwarf2out_early_finish): Move setting of AT_pubnames from
12165 early debug output to early finish.
12167 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12169 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12170 and -mdata-region to the assembler.
12172 2017-08-30 Richard Biener <rguenther@suse.de>
12174 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12176 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12177 (dwarf2out_early_global_decl): Only generate a DIE for the
12178 abstract origin if it doesn't already exist or is a declaration DIE.
12179 (resolve_addr): Do not add the linkage name twice when
12180 generating a stub DIE for the DW_TAG_GNU_call_site target.
12182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12184 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12185 Use machine_mode rather than int for arg1_mode.
12187 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12190 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12191 that the second argument of the built-in functions to unpack
12192 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12193 switch statement instead a lot of if statements.
12194 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12195 Allow 64-bit values to be in Altivec registers as well as
12196 traditional floating point registers.
12197 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12199 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12201 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12202 MAX_REGS_PER_ADDRESS == 1.
12204 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12206 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12207 * config/i386/i386.c (ix86_option_override_internal): Simplify
12208 setting of opts->x_flag_entry.
12210 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12211 Jakub Jelinek <jakub@redhat.com>
12212 Richard Biener <rguenther@suse.de>
12214 PR tree-optimization/81503
12215 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12216 folded constant fits in the target type; reorder tests for clarity.
12218 2017-08-29 Martin Liska <mliska@suse.cz>
12220 * passes.def: Include pass_lower_switch.
12221 * stmt.c (dump_case_nodes): Remove and move to
12222 tree-switch-conversion.
12223 (case_values_threshold): Likewise.
12224 (expand_switch_as_decision_tree_p): Likewise.
12225 (emit_case_decision_tree): Likewise.
12226 (expand_case): Likewise.
12227 (balance_case_nodes): Likewise.
12228 (node_has_low_bound): Likewise.
12229 (node_has_high_bound): Likewise.
12230 (node_is_bounded): Likewise.
12231 (emit_case_nodes): Likewise.
12232 (struct simple_case_node): New struct.
12233 (add_case_node): Remove.
12234 (emit_case_dispatch_table): Use vector instead of case_list.
12235 (reset_out_edges_aux): Remove.
12236 (compute_cases_per_edge): Likewise.
12237 (expand_case): Build list of simple_case_node.
12238 (expand_sjlj_dispatch_table): Use it.
12239 * tree-switch-conversion.c (struct case_node): Moved from
12240 stmt.c and adjusted.
12241 (emit_case_nodes): Likewise.
12242 (node_has_low_bound): Likewise.
12243 (node_has_high_bound): Likewise.
12244 (node_is_bounded): Likewise.
12245 (case_values_threshold): Likewise.
12246 (reset_out_edges_aux): Likewise.
12247 (compute_cases_per_edge): Likewise.
12248 (add_case_node): Likewise.
12249 (dump_case_nodes): Likewise.
12250 (balance_case_nodes): Likewise.
12251 (expand_switch_as_decision_tree_p): Likewise.
12252 (emit_jump): Likewise.
12253 (emit_case_decision_tree): Likewise.
12254 (try_switch_expansion): Likewise.
12255 (do_jump_if_equal): Likewise.
12256 (emit_cmp_and_jump_insns): Likewise.
12257 (fix_phi_operands_for_edge): New function.
12258 (record_phi_operand_mapping): Likewise.
12259 (class pass_lower_switch): New pass.
12260 (pass_lower_switch::execute): New function.
12261 (make_pass_lower_switch): Likewise.
12262 (conditional_probability):
12263 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12264 * tree-pass.h: Add make_pass_lower_switch.
12266 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12269 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12272 2017-08-29 Richard Biener <rguenther@suse.de>
12274 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12275 we do not add a DW_AT_inline attribute twice.
12276 (gen_subprogram_die): Remove code setting DW_AT_inline on
12277 DECL_ABSTRACT_P nodes.
12279 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12281 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12282 calls to internal functions.
12283 (gimplify_modify_expr): Likewise.
12284 * tree-call-cdce.c (use_internal_fn): Likewise.
12285 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12286 (convert_to_divmod): Set the nothrow flag.
12287 * tree-if-conv.c (predicate_mem_writes): Likewise.
12288 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12289 (vectorizable_call): Likewise.
12290 (vectorizable_store): Likewise.
12291 (vectorizable_load): Likewise.
12292 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12293 (vect_recog_mask_conversion_pattern): Likewise.
12295 2017-08-29 Martin Liska <mliska@suse.cz>
12298 * gcc.c (driver_handle_option): Add new argument.
12299 * opts-common.c (handle_option): Pass
12300 target_option_override_hook.
12301 * opts-global.c (lang_handle_option): Add new option.
12302 (set_default_handlers): Add new argument.
12303 (decode_options): Likewise.
12304 * opts.c (target_handle_option): Likewise.
12305 (common_handle_option): Call target_option_override_hook.
12306 * opts.h (struct cl_option_handler_func): Add hook for
12307 target option override.
12308 (struct cl_option_handlers): Likewise.
12309 (set_default_handlers): Add new argument.
12310 (decode_options): Likewise.
12311 (common_handle_option): Likewise.
12312 (target_handle_option): Likewise.
12313 * toplev.c (toplev::main): Pass targetm.target_option.override
12316 2017-08-29 Richard Biener <rguenther@suse.de>
12317 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12319 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12320 life to the active subtree.
12322 2017-08-28 Jeff Law <law@redhat.com>
12324 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12325 derive_equivalences.
12326 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12327 Code moved into....
12328 (edge_info::derive_equivalences): New private member function
12330 * tree-ssa-dom.c (class edge_info): Changed from a struct
12331 to a class. Add ctor/dtor, methods and data members.
12332 (edge_info::edge_info): Renamed from allocate_edge_info.
12333 Initialize additional members.
12334 (edge_info::~edge_info): New.
12335 (free_dom_edge_info): Delete the edge info.
12336 (record_edge_info): Use new class & associated member functions.
12337 Tighten forms for testing for edge equivalences.
12338 (record_temporary_equivalences): Iterate over the simple
12339 equivalences rather than assuming there's only one per edge.
12340 (cprop_into_successor_phis): Iterate over the simple
12341 equivalences rather than assuming there's only one per edge.
12342 (optimize_stmt): Use operand_equal_p rather than pointer
12343 equality for mini-DSE code.
12345 2017-08-28 Nathan Sidwell <nathan@acm.org>
12347 * gcc.c (execute): Fold SIGPIPE handling into switch
12348 statement. Adjust internal error message.
12350 2017-08-28 Richard Biener <rguenther@suse.de>
12353 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12354 Do nothing for removed DIEs.
12356 2017-08-28 Richard Biener <rguenther@suse.de>
12358 PR tree-optimization/81977
12359 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12362 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12365 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12366 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12367 using targetm.gen_mem_thread_fence.
12369 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12372 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12373 predicate to register_operand. Reorder operands.
12374 (*btr<mode>): Ditto.
12375 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12376 (*btr<mode>_mask): Ditto.
12378 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12380 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12381 * config/rs6000/xmmintrin.h: New file.
12382 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12384 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12387 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12388 parameter and_insn and return it.
12389 (recombine_lvx_pattern): Insert a copy to ensure availability of
12390 the base register of the copied masking operation at the point of
12391 the instruction replacement.
12392 (recombine_stvx_pattern): Likewise.
12394 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12396 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12397 undocumented switches.
12398 (-mpower9-dform-vector): Likewise.
12399 (-mpower9-dform): Likewise.
12400 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12401 comments to delete references to -mpower9-dform* switches.
12402 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12403 Delete reference to -mpower9-dform* switches, test for
12404 -mpower9-vector instead.
12405 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12406 (OTHER_P9_VECTOR_MASKS): Likewise.
12407 (POWERPC_MASKS): Likewise.
12408 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12409 tests against -mpower9-dform* to -mpower9-vector. Delete code
12410 that checked for -mpower9-dform* consistancy with other options.
12411 Add test for -mpower9-misc to enable other power9 switches.
12412 (rs6000_init_hard_regno_mode_ok): Likewise.
12413 (rs6000_option_override_internal): Likewise.
12414 (rs6000_emit_prologue): Likewise.
12415 (rs6000_emit_epilogue): Likewise.
12416 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12417 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12418 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12420 (emit_fusion_p9_store): Likewise.
12421 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12422 resetting these macros if the assembler does not support ISA 3.0
12424 (TARGET_P9_DFORM_VECTOR): Likewise.
12425 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12426 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12428 2017-08-25 Alan Modra <amodra@gmail.com>
12430 PR rtl-optimization/81747
12431 * cse.c (cse_extended_basic_block): Don't attempt to record
12432 equivalences for degenerate conditional branches that branch
12433 to their fall-through.
12435 2017-08-24 Martin Sebor <msebor@redhat.com>
12437 PR middle-end/81908
12438 * gimple-fold.c (size_must_be_zero_p): New function.
12439 (gimple_fold_builtin_memory_op): Call it.
12441 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12443 * config/rs6000/mm_malloc.h: New file.
12445 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12447 PR tree-optimization/81913
12448 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12449 analysis when either IVs in condition can wrap.
12451 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12453 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12454 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12456 2017-08-24 Richard Biener <rguenther@suse.de>
12459 * targhooks.c (default_target_can_inline_p): Properly
12460 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12461 is present and always compare.
12462 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12463 infer -mfpmath=sse from TARGET_SSE_P.
12464 (ix86_can_inline_p): Properly use target_option_default_node when
12465 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12467 2017-08-24 Richard Biener <rguenther@suse.de>
12470 * dwarf2out.c (output_die): Handle flag_generate_offload like
12472 (output_comp_unit): Likewise.
12473 (gen_array_type_die): Likewise.
12474 (dwarf2out_early_finish): Likewise.
12475 (note_variable_value_in_expr): Likewise.
12476 (dwarf2out_finish): Likewise. Adjust assert.
12477 * cgraphunit.c (symbol_table::compile): Move setting of
12478 flag_generate_offload earlier ...
12479 (symbol_table::finalize_compilation_unit): ... here, before
12480 early debug finalization.
12482 2017-08-24 Richard Biener <rguenther@suse.de>
12484 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12485 and ipa-fnsummary.h.
12486 (ix86_can_inline_p): When ix86_fpmath flags do not match
12487 check whether the callee uses FP math at all.
12489 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12491 PR middle-end/81931
12492 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12493 instead of TYPE_PRECISION.
12495 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12496 Alan Hayward <alan.hayward@arm.com>
12497 David Sherwood <david.sherwood@arm.com>
12499 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12501 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12503 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12505 (ix86_frame::stack_realign_allocate): New field.
12506 (struct machine_frame_state): Modify comments.
12507 (machine_frame_state::sp_realigned_fp_end): New field.
12508 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12509 layout calculation.
12510 (sp_valid_at): Add assertion to assure no attempt to access invalid
12511 offset of a realigned stack.
12512 (fp_valid_at): Likewise.
12513 (choose_baseaddr): Modify comments.
12514 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12515 ix86_expand_prologue.
12516 (ix86_expand_prologue): Modify stack realignment and allocation.
12517 (ix86_expand_epilogue): Modify comments.
12518 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12519 avx2_runtime, avx512f, and avx512f_runtime.
12521 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12523 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12524 (mstackrealign): Do not init to -1.
12525 * config/i386/i386.c (ix86_option_override_internal):
12526 Check opts_set, not opts when setting default value of
12527 opts->x_ix86_force_align_arg_pointer.
12529 2017-08-23 Richard Biener <rguenther@suse.de>
12531 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12532 lang_hooks.decl_printable_name.
12533 * print-rtl-function.c (print_rtx_function): Likewise.
12534 * tree-pretty-print.c (dump_function_header): Likewise.
12536 2017-08-23 Richard Biener <rguenther@suse.de>
12539 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12540 -g0 at compile-time.
12542 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12544 PR middle-end/19706
12545 * doc/sourcebuild.texi (Other hardware attributes):
12548 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12550 PR middle-end/19706
12551 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12552 Fix single-use check.
12554 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12556 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12558 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12560 * doc/install.texi: Modify to add more details on running selected
12563 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12565 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12566 is combined with -mabi=ms.
12567 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12570 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12572 PR tree-optimization/81488
12573 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12574 and cached_basis fields.
12575 (MAX_SPREAD): New constant.
12576 (alloc_cand_and_find_basis): Initialize new fields.
12577 (clear_visited): New function.
12578 (create_phi_basis_1): Rename from create_phi_basis, set visited
12579 and cached_basis fields.
12580 (create_phi_basis): New wrapper function.
12581 (phi_add_costs_1): Rename from phi_add_costs, add spread
12582 parameter, set visited field, short-circuit when limits reached.
12583 (phi_add_costs): New wrapper function.
12584 (record_phi_increments_1): Rename from record_phi_increments, set
12586 (record_phi_increments): New wrapper function.
12587 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12588 (phi_incr_cost): New wrapper function.
12589 (all_phi_incrs_profitable_1): Rename from
12590 all_phi_incrs_profitable, set visited field.
12591 (all_phi_incrs_profitable): New wrapper function.
12593 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12594 Alan Hayward <alan.hayward@arm.com>
12595 David Sherwood <david.sherwood@arm.com>
12597 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12598 that takes the outer and inner modes.
12599 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12600 comparison as the canonical test for a paradoxical subreg.
12601 * combine.c (simplify_set): Use paradoxical_subreg_p.
12602 (make_extraction): Likewise.
12603 (force_to_mode): Likewise.
12604 (rtx_equal_for_field_assignment_p): Likewise.
12605 (gen_lowpart_for_combine): Likewise.
12606 (simplify_comparison): Likewise.
12607 * cse.c (equiv_constant): Likewise.
12608 * expmed.c (store_bit_field_1): Likewise.
12609 * final.c (alter_subreg): Likewise.
12610 * fwprop.c (propagate_rtx): Likewise.
12611 (forward_propagate_subreg): Likewise.
12612 * ira-conflicts.c (ira_build_conflicts): Likewise.
12613 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12614 * lra-constraints.c (curr_insn_transform): Likewise.
12615 (split_reg): Likewise.
12616 * lra-eliminations.c (move_plus_up): Likewise.
12617 (lra_eliminate_regs_1): Likewise.
12618 * recog.c (general_operand): Likewise.
12619 * ree.c (combine_reaching_defs): Likewise.
12620 * reload.c (push_reload): Likewise.
12621 (find_reloads): Likewise.
12622 * reload1.c (elimination_effects): Likewise.
12623 (compute_reload_subreg_offset): Likewise.
12624 (choose_reload_regs): Likewise.
12625 * rtlanal.c (subreg_lsb_1): Likewise.
12626 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12627 (simplify_subreg): Likewise.
12628 * var-tracking.c (track_loc_p): Likewise.
12629 * emit-rtl.c (byte_lowpart_offset): Likewise.
12630 (paradoxical_subreg_p): Delete out-of-line definition.
12632 2017-08-22 Jeff Law <law@redhat.com>
12634 PR tree-optimization/81741
12635 PR tree-optimization/71947
12636 * tree-ssa-dom.c: Include tree-inline.h.
12637 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12638 equivalences if one is more expensive to compute than the other.
12639 * tree-ssa-scopedtables.h (class const_or_copies): Make
12640 record_const_or_copy_raw method private.
12641 (class avail_exprs_stack): New method simplify_binary_operation.
12642 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12643 avail_exprs_stack::simplify_binary_operation as needed.
12644 (avail_exprs_stack::simplify_binary_operation): New function.
12646 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12648 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12649 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12650 (DOT_SYMBOLS): Likewise.
12651 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12652 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12653 (RS6000_ABI_NAME): Likewise.
12654 (TARGET_CMODEL): Likewise.
12655 (TOC_SECTION_ASM_OP): Likewise.
12656 (SET_CMODEL): New macro.
12657 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12659 2017-08-22 Richard Biener <rguenther@suse.de>
12661 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12662 to allow for free-lang-data replacements similar to verify_type_variant.
12664 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12665 Alan Hayward <alan.hayward@arm.com>
12666 David Sherwood <david.sherwood@arm.com>
12668 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12669 VOIDmode for the LABEL_REF.
12671 2017-08-22 Richard Biener <rguenther@suse.de>
12673 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12675 2017-08-22 Richard Biener <rguenther@suse.de>
12677 * common.opt (feliminate-dwarf2-dups): Ignore.
12678 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12679 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12680 same_die_p_wrap, compute_section_prefix,
12681 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12682 (comdat_symbol_id, comdat_symbol_number): Likewise.
12683 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12685 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12686 (output_die): Mark unreachable path unreachable.
12687 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12688 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12689 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12690 (dwarf2out_early_finish): Likewise.
12692 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12694 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12696 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12699 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12700 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12701 (avr_attribute_table) <io, io_low, address>: Initialize
12702 .decl_required with true.
12704 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12706 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12707 undocumented debugging options.
12708 (-mvsx-scalar-double): Likewise.
12709 (-mallow-df-permute): Likewise.
12710 (-mvectorize-builtins): Likewise.
12711 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12712 (rs6000_builtin_vectorized_function): Likewise.
12713 (rs6000_builtin_md_vectorized_function): Likewise.
12714 (rs6000_opt_vars): Likewise.
12716 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12719 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12720 (*btrq_imm): Rename from *btrq.
12721 (*btcq_imm): Rename from *btcq.
12722 (btsc): New code attribute.
12723 (*<btsc><mode>): New insn pattern.
12724 (*btr<mode>): Ditto.
12725 (*<btsc><mode>_mask): New insn_and_split pattern.
12726 (*btr<mode>_mask): Ditto.
12728 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12729 Alan Hayward <alan.hayward@arm.com>
12730 David Sherwood <david.sherwood@arm.com>
12732 * function.c (pad_below): Simplify padding calculation.
12734 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12735 Alan Hayward <alan.hayward@arm.com>
12736 David Sherwood <david.sherwood@arm.com>
12738 * target.def (function_prologue): Remove frame size argument.
12739 (function_epilogue): Likewise.
12740 * doc/tm.texi: Regenerate.
12741 * final.c (final_start_function): Update call to function_prologue.
12742 (final_end_function): Update call to function_epilogue.
12743 (default_function_pro_epilogue): Remove frame size argument.
12744 * output.h (default_function_pro_epilogue): Likewise.
12745 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12746 (arm_output_function_prologue): Likewise.
12747 * config/frv/frv.c (frv_function_prologue): Likewise.
12748 (frv_function_epilogue): Likewise.
12749 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12750 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12751 (ia64_output_function_epilogue): Likewise.
12752 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12753 (m32r_output_function_epilogue): Likewise.
12754 * config/microblaze/microblaze.c (microblaze_function_prologue)
12755 (microblaze_function_epilogue): Likewise.
12756 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12757 (mips_output_function_epilogue): Likewise.
12758 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12759 (mmix_target_asm_function_epilogue): Likewise.
12760 * config/msp430/msp430.c (msp430_start_function): Likewise.
12761 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12762 (nds32_asm_function_epilogue): Likewise.
12763 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12764 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12765 (pa_output_function_epilogue): Likewise.
12766 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12767 (rs6000_output_function_epilogue): Likewise.
12768 * config/rl78/rl78.c (rl78_start_function): Likewise.
12769 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12770 (rs6000_output_function_epilogue): Likewise.
12771 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12772 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12773 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12774 (sparc_asm_function_epilogue): Likewise.
12776 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12778 * tree.h (type_has_mode_precision_p): New function.
12779 * convert.c (convert_to_integer_1): Use it.
12780 * expr.c (expand_expr_real_2): Likewise.
12781 (expand_expr_real_1): Likewise.
12782 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12783 * match.pd: Likewise.
12784 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12785 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12786 * tree-tailcall.c (process_assignment): Likewise.
12787 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12788 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12789 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12790 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12791 (vectorizable_assignment): Likewise.
12792 (vectorizable_shift): Likewise.
12793 (vectorizable_operation): Likewise.
12794 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12796 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12798 * match.pd: Add pow (C, x) simplification.
12800 2017-08-21 Richard Biener <rguenther@suse.de>
12802 PR tree-optimization/81900
12803 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12804 for blocks with abnormal predecessors.
12805 (compute_antic): Do not set visited flag prematurely.
12807 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12810 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12812 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12814 * stor-layout.h (vector_type_mode): Move to...
12815 * tree.h (vector_type_mode): ...here.
12816 * stor-layout.c (vector_type_mode): Move to...
12817 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12819 2017-08-21 Richard Biener <rguenther@suse.de>
12821 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12822 register_external_die hooks.
12823 (debug_false_tree_charstarstar_uhwistar): Declare.
12824 (debug_nothing_tree_charstar_uhwi): Likewise.
12825 * debug.c (do_nothing_debug_hooks): Adjust.
12826 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12827 (debug_nothing_tree_charstar_uhwi): Likewise.
12828 * dbxout.c (dbx_debug_hooks): Adjust.
12829 (xcoff_debug_hooks): Likewise.
12830 * sdbout.c (sdb_debug_hooks): Likewise.
12831 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12832 * dwarf2out.c (macinfo_label_base): New global.
12833 (dwarf2out_register_external_die): New function for the
12834 register_external_die hook.
12835 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12836 (dwarf2_debug_hooks): Use them.
12837 (dwarf2_lineno_debug_hooks): Adjust.
12838 (struct die_struct): Add with_offset flag.
12839 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12840 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12841 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12842 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12843 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12844 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12845 defining section names for the early LTO debug variants.
12846 (reset_indirect_string): New helper.
12847 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12848 (print_dw_val): Add support for offsetted symbol references.
12849 (get_ultimate_context): Split out from is_cxx.
12850 (is_cxx): Use get_ultimate_context.
12851 (is_fortran): Add decl overload.
12852 (compute_comp_unit_symbol): Split out worker from
12853 compute_section_prefix.
12854 (compute_section_prefix): Call compute_comp_unit_symbol and
12855 set comdat_type_p here.
12856 (output_die): Skip DIE symbol output for the LTO added one.
12857 Handle DIE symbol references with offset.
12858 (output_comp_unit): Guard section name mangling properly.
12859 For LTO debug sections emit a symbol at the section beginning
12860 which we use to refer to its DIEs.
12861 (add_abstract_origin_attribute): For DIEs registered via
12862 dwarf2out_register_external_die directly refer to the early
12863 DIE rather than indirectly through the shadow one we created.
12864 Remove obsolete call to dwarf2out_abstract_function for
12865 non-function/block origins.
12866 (gen_array_type_die): When generating early LTO debug do
12867 not emit DW_AT_string_length.
12868 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12869 late when in LTO. As suggested place a gcc_unreachable for
12870 the DECL_ABSTRACT_P case.
12871 (gen_subprogram_die): Avoid another specification DIE
12872 for early built declarations/definitions for the late LTO case.
12873 (gen_variable_die): Add type references for late duplicated VLA dies
12875 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12876 we have the abstract instance already.
12877 (process_scope_var): Adjust decl DIE contexts in LTO which
12878 first puts them in limbo.
12879 (gen_decl_die): Do not generate type DIEs late apart from
12880 types for VLAs or for decls we do not yet have a DIE. Do not
12881 call dwarf2out_abstract_function late.
12882 (dwarf2out_early_global_decl): Make sure to create DIEs
12883 for abstract instances of a decl first.
12884 (dwarf2out_late_global_decl): Adjust comment.
12885 (output_macinfo_op): With multiple macro sections use
12886 macinfo_label_base to distinguish labels.
12887 (output_macinfo): Likewise. Update macinfo_label_base.
12888 Pass in the line info label.
12889 (note_variable_value_in_expr): When generating LTO resolve
12890 all variable values here by generating DIEs as needed.
12891 (init_sections_and_labels): Add early LTO debug flag parameter
12892 and generate different sections and names if set. Add generation
12893 counter for the labels so we can have multiple of them.
12894 (reset_dies): Helper to allow DIEs to be output multiple times.
12895 (dwarf2out_finish): When outputting DIEs to the fat part of an
12896 LTO object first reset DIEs.
12897 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12898 (modified_type_die): Check for decl_ultimate_origin being self
12900 (gen_type_die_with_usage): Likewise.
12901 (gen_typedef_die): Allow decl_ultimate_origin being self.
12902 (set_decl_abstract_flags): Remove.
12903 (set_block_abstract_flags): Likewise.
12904 (dwarf2out_abstract_function): Treat the early generated DIEs
12905 as the abstract copy and only add DW_AT_inline and
12906 DW_AT_artificial here and call set_decl_origin_self.
12907 If the DIE has an abstract origin don't do anything.
12908 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12909 if we have none yet (Go fails to build one, PR78628).
12910 (variably_modified_type_p): Prevent endless recursion for Ada
12911 cyclic pointer types.
12912 * lto-streamer-in.c: Include debug.h.
12913 (dref_queue): New global.
12914 (lto_read_tree_1): Stream in DIE references.
12915 (lto_input_tree): Register DIE references.
12916 (input_function): Stream DECL_DEBUG_ARGS.
12917 * lto-streamer-out.c: Include debug.h.
12918 (lto_write_tree_1): Output DIE references.
12919 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12920 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12921 (output_function): Stream DECL_DEBUG_ARGS.
12922 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12923 Stream DECL_ABSTRACT_ORIGIN.
12924 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12925 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12926 DECL_CONTEXT for file-scope decls.
12927 * lto-streamer.h (struct dref_entry): Declare.
12928 (dref_queue): Likewise.
12929 * cfgexpand.c (pass_expand::execute): Do not call the
12930 outlining_inline_function hook here.
12931 * lto-wrapper.c (debug_obj): New global.
12932 (tool_cleanup): Unlink it if required.
12933 (debug_objcopy): New function.
12934 (run_gcc): Handle early debug sections in the IL files by
12935 extracting them to separate files, partially linkin them and
12936 feeding the result back as result to the linker.
12937 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12938 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12939 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12940 sections into a separate segment.
12941 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12943 (darwin_asm_dwarf_section): Likewise.
12944 (darwin_asm_output_dwarf_offset): Likewise.
12945 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12947 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12948 Alan Hayward <alan.hayward@arm.com>
12949 David Sherwood <david.sherwood@arm.com>
12951 * read-md.h (md_reader::record_potential_iterator_use): Replace
12952 pointer argument with an rtx and an index.
12953 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12954 (apply_mode_iterator): Likewise.
12955 (apply_code_iterator): Likewise.
12956 (apply_int_iterator): Likewise.
12957 (apply_subst_iterator): Likewise.
12958 (record_iterator_use): Likewise.
12959 (record_attribute_use): Likewise.
12960 (md_reader::record_potential_iterator_use): Likewise. Update calls
12961 to record_iterator_use and apply_iterator.
12962 (iterator_use): Replace ptr with x and index.
12963 (attribute_use): Likewise.
12964 (apply_attribute_uses): Update calls to apply_iterator.
12965 (apply_iterators): Likewise. Update initialization of iterator_use.
12966 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12967 and record_potential_iterator_use.
12968 (rtx_reader::read_rtx_operand): Likewise.
12970 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12971 Alan Hayward <alan.hayward@arm.com>
12972 David Sherwood <david.sherwood@arm.com>
12974 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12977 2017-08-21 Richard Biener <rguenther@suse.de>
12979 PR middle-end/81884
12980 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12981 at struct end conservatively when comparing common bases.
12983 2017-08-21 Richard Biener <rguenther@suse.de>
12985 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12986 (mem_ref_in_stmt): Remove.
12987 (determine_max_movement): Use ref index to get at the reference.
12988 (invariantness_dom_walker::before_dom_children): Deal with
12989 lim data already initialized.
12990 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12992 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12994 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12995 (TARGET_ISA_ROUND): Ditto.
12996 (TARGET_ROUND): Ditto.
12997 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12998 * config/i386/i386.md: Ditto.
12999 * config/i386/sse.md: Ditto.
13000 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13001 with OPTION_MASK_ISA_SSE4_1.
13003 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13006 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13007 __builtin_ia32_lzcnt_u16.
13009 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13012 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13013 (rs6000_set_current_function): Rewrite function to use it.
13015 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13018 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13019 and TYPE_WARN_IF_NOT_ALIGN.
13020 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13021 (handle_warn_if_not_align): New.
13022 (place_union_field): Call handle_warn_if_not_align.
13023 (place_field): Call handle_warn_if_not_align.
13024 Copy TYPE_WARN_IF_NOT_ALIGN.
13025 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13026 (layout_type): Likewise.
13027 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13029 (tree_decl_common): Add warn_if_not_align.
13030 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13031 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13032 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13033 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13034 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13035 * doc/extend.texi: Document warn_if_not_aligned attribute.
13036 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13038 2017-08-17 Martin Liska <mliska@suse.cz>
13041 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13042 (get_data_dependence): Use it as pointer type.
13043 (distribute_loop): Likewise.
13045 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13047 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13048 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13049 (p8_vmrgow_<mode>_direct): New define_insn.
13050 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13051 handle endianness for vmrgew and vmrgow permute patterns.
13053 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13055 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13056 * config/rs6000/rs6000-cpus.def: Remove comment.
13057 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13058 (POWERPC_MASKS): Likewise.
13059 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13060 use of TARGET_VSX_TIMODE.
13061 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13062 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13063 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13064 (rs6000_option_override_internal): Remove dead code.
13065 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13066 (rs6000_legitimize_reload_address): Likewise.
13067 (rs6000_legitimate_address_p): Likewise.
13068 (rs6000_opt_masks): Delete "vsx-timode".
13069 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13070 from function comment.
13071 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13072 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13073 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13075 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13076 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13077 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13078 (VSX_TI): Likewise.
13080 (define_peephole2): Likewise.
13082 2017-08-17 Martin Sebor <msebor@redhat.com>
13085 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13086 past the end of an array.
13087 (test_pp_format): Add test cases.
13089 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13091 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13092 missing ECF_NOTHROW flags.
13094 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13097 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13098 operands residing in integer registers.
13099 (*vsx_le_perm_load_<mode>): Likewise.
13100 (*vsx_le_perm_store_<mode>): Likewise.
13101 (define_peephole2): Add peepholes to optimize the above.
13103 2017-08-17 Marek Polacek <polacek@redhat.com>
13105 PR middle-end/81814
13106 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13107 to mimic what shorten_compare did. Change the return type to bool.
13108 (fold_cond_expr_with_comparison): Update call to
13109 operand_equal_for_comparison_p.
13110 (fold_ternary_loc): Likewise.
13112 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13114 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13116 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13118 2017-08-17 Richard Biener <rguenther@suse.de>
13120 * tree-ssa-structalias.c (solve_graph): When propagating
13121 to successors update the graphs succ edges and avoid duplicate work.
13123 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13126 * config/i386/i386.c (ix86_option_override_internal): Save target
13127 specific options after ix86_stack_protector_guard_reg was changed.
13129 2017-08-17 Richard Biener <rguenther@suse.de>
13131 PR tree-optimization/81827
13132 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13133 (new_var_info): Initialize it conservatively.
13134 (get_call_vi): Mark register vars.
13135 (new_scalar_tmp_constraint_exp): Likewise.
13136 (handle_rhs_call): Likewise.
13137 (handle_const_call): Likewise.
13138 (create_function_info_for): Likewise.
13139 (solve_constraints): Sort varinfos to separate register from
13140 non-register vars to pack points-to solution bitmaps during
13143 2017-08-17 Marek Polacek <polacek@redhat.com>
13145 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13147 2017-08-17 Richard Biener <rguenther@suse.de>
13149 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13150 to true when overflow is undefined and we saturated the result.
13152 2017-08-17 Alan Modra <amodra@gmail.com>
13155 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13156 Don't use store multiple if only one reg needs saving.
13157 (interesting_frame_related_regno): New function.
13158 (rs6000_frame_related): Don't emit frame info for regs that
13160 (rs6000_emit_epilogue): Likewise.
13162 2017-08-16 Nathan Sidwell <nathan@acm.org>
13164 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13165 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13166 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13167 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13168 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13169 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13170 (verify_type): Adjust for TYPE_BINFO move.
13171 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13172 process TYPE_BINFO directly.
13173 (hash_tree): Likewise.
13174 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13176 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13179 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13181 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13183 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13186 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13187 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13188 Add ix86_binary_operator_ok to insn constraint.
13189 (*iordi_1_bts): Ditto.
13190 (*xordi_1_btc): Ditto.
13191 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13192 Update corresponding peephole2 pattern.
13196 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13198 PR tree-optimization/81832
13199 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13200 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13202 2017-08-16 Marek Polacek <polacek@redhat.com>
13205 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13206 regarding size_zero_node.
13208 2017-08-16 Martin Liska <mliska@suse.cz>
13211 * config.gcc: Respect previously set extra_objs in case
13214 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13216 PR tree-optimization/81835
13217 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13218 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13219 not depend on the phi.
13221 2017-08-16 Alan Modra <amodra@gmail.com>
13223 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13226 2017-08-16 Alan Modra <amodra@gmail.com>
13228 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13229 (save_reg_p): ..into this. Update all callers.
13230 (first_reg_to_save): Simplify.
13232 2017-08-16 Alan Modra <amodra@gmail.com>
13234 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13237 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13241 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13242 constant count if that count is less than 32.
13244 2017-08-15 Nathan Sidwell <nathan@acm.org>
13246 * gcc.c (execute): Emit friendlier message if inferior is killed
13247 by an external cause.
13249 2017-08-15 Richard Biener <rguenther@suse.de>
13251 PR tree-optimization/81790
13252 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13253 CONSTRUCTORs from simplifying and VN.
13255 2017-08-14 Martin Sebor <msebor@redhat.com>
13257 * builtin-attrs.def: Add comments.
13259 2017-08-14 Martin Sebor <msebor@redhat.com>
13262 * doc/extend.texi (attribute nonstring): Document new attribute.
13264 2017-08-14 Martin Sebor <msebor@redhat.com>
13267 * tree-diagnostic.c (default_tree_printer): Handle %G.
13268 * gimple-pretty-print.h (percent_G_format): Declare new function.
13269 * gimple-pretty-print.c (percent_G_format): Define.
13270 * tree-pretty-print.c (percent_K_format): Add argument.
13272 2017-08-14 Martin Sebor <msebor@redhat.com>
13274 PR translation/79998
13275 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13276 Remove a stray space.
13278 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13281 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13282 (*iordi_1_bts): Ditto.
13283 (*xordi_1_btc): Ditto.
13285 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13288 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13289 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13291 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13292 quoted strings, and make more translator-friendly.
13293 (darwin_rs6000_override_options): Likewise.
13294 (rs6000_option_override_internal): Likewise.
13295 (rs6000_return_in_memory): Fix overlong line.
13296 (init_cmulative_args): Use quoted strings, and make more
13297 translator-friendly.
13298 (rs6000_pass_by_reference): Fix overlong line.
13299 (def_builtin): Use quoted strings.
13300 (altivec_expand_predicate_builtin): Use quoted strings, and make
13301 more translator-friendly.
13302 (htm_expand_builtin): Use quoted strings.
13303 (cpu_expand_builtin): Use quoted strings, and make more
13304 translator-friendly.
13305 (altivec_expand_builtin): Likewise.
13306 (paired_expand_predicate_builtin): Likewise.
13307 (rs6000_invalid_builtin): Likewise.
13308 (builtin_function_type): Use quoted strings.
13309 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13310 more translator-friendly.
13311 (rs6000_trampoline_init): Likewise.
13312 (rs6000_handle_altivec_attribute): Likewise.
13313 (rs6000_inner_target_options): Use quoted strings.
13314 (rs6000_disable_incompatible_switches): Likewise.
13315 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13316 strings, and make more translator-friendly.
13317 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13319 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13321 PR tree-optimization/81799
13322 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13323 cond_expr to simple gimple operand.
13325 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13327 PR middle-end/46932
13328 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13330 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13334 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13335 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13336 TARGET_GASISR_PROLOGUES.
13337 * config/avr/avr.c (avr_option_override): Same.
13338 (avr_pass_pre_proep::execute): Same.
13340 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13343 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13344 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13346 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13348 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13349 define_insn_and_split. Split to a memory load from 0 in
13350 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13351 using PTR mode iterator.
13352 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13353 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13354 (*add_tp_<mode>): Redefine as define_insn_and_split.
13355 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13356 address space. Merge with *add_tp_x32 using PTR mode iterator.
13357 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13358 Split to an add with a memory load from 0 in
13359 DEFAULT_TLS_SEG_REG address space.
13361 2017-08-12 Andrew Pinski <apinski@cavium.com>
13363 * config/aarch64/aarch64-option-extensions.def (rdma):
13364 Fix feature string to what Linux prints out in /proc/cpuinfo.
13366 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13369 * dwarf2out.c (modified_type_die): For C typedef types that have
13370 an ultimate origin, process the ultimate origin instead of the
13372 (gen_typedef_die): Assert that input DECLs have no ultimate
13374 (gen_type_die_with_usage): For typedef variants that have an
13375 ultimate origin, just call gen_decl_die on the original DECL.
13376 (process_scope_var): Avoid creating DIEs for local typedefs and
13377 concrete static variables.
13379 2017-08-12 Alan Modra <amodra@gmail.com>
13383 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13384 match gnu-user.h startfile.
13385 (ENDFILE_LINUX_SPEC): Similarly.
13387 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13390 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13392 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13394 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13395 * config/aarch64/aarch64.md (mov<mode>): Change.
13396 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13397 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13398 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13400 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13402 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13403 for storage order barriers.
13405 2017-08-11 Martin Liska <mliska@suse.cz>
13408 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13409 variables of void type.
13411 2017-08-11 Martin Liska <mliska@suse.cz>
13413 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13414 TARGET_SUPPORTS_ALIASES.
13415 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13416 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13417 (optimize_weakref): Likewise.
13418 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13419 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13420 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13422 2017-08-11 Martin Liska <mliska@suse.cz>
13425 * config/i386/i386.c (make_resolver_func): Do complete
13426 refactoring of the function.
13428 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13431 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13432 * config/i386/i386.c (ix86_stack_protect_guard): Use
13433 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13434 * doc/invoke.texi (x86 Options): Document
13435 -mstack-protector-guard-symbol= option.
13437 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13439 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13440 * config/i386/i386.c (ix86_split_stack_guard): New function.
13441 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13442 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13443 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13444 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13445 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13446 (split_stack_space_check): Call ix86_split_stack_guard.
13448 2017-08-10 Martin Sebor <msebor@redhat.com>
13450 * print-tree.c (print_node): Print location using the established
13452 Replace spaces with colons.
13453 (debug_raw, debug): Ditto.
13455 2017-08-10 Martin Sebor <msebor@redhat.com>
13458 * pretty-print.c (pp_format): Correct the handling of %s precision.
13460 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13463 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13465 (ix86_finalize_stack_frame_flags): This. Also clear
13466 frame_pointer_needed if -fno-omit-frame-pointer is used without
13468 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13469 with ix86_finalize_stack_frame_flags.
13470 (ix86_expand_epilogue): Likewise.
13471 (ix86_expand_split_stack_prologue): Likewise.
13472 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13474 2017-08-10 Martin Liska <mliska@suse.cz>
13477 * c-attribs.c (handle_target_attribute):
13478 Report warning for an empty string argument of target attribute.
13480 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13483 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13485 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13486 or DECL_NONLOCAL labels.
13487 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13488 or DECL_NONLOCAL labels here.
13490 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13492 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13493 to indicate when early gimple folding has been disabled.
13494 (rs6000_gimple_fold_builtin): Add debug content.
13495 (rs6000_invalid_builtin): Fix whitespace.
13496 (rs6000_expand_builtin): Fix whitespace.
13497 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13499 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13502 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13503 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13505 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13507 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13508 (qdf24xx): Likewise.
13509 * config/aarch64/aarch64.md: Include falkor.md.
13510 * config/aarch64/falkor.md: New.
13512 2017-08-09 Marek Polacek <polacek@redhat.com>
13515 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13516 * diagnostic.c (emit_diagnostic): Add a comment.
13517 (emit_diagnostic_valist): New function.
13519 2017-08-09 Marek Polacek <polacek@redhat.com>
13522 * input.c (make_location): New overload.
13523 * input.h (make_location): Declare.
13525 2017-08-08 Alan Modra <amodra@gmail.com>
13526 H.J. Lu <hongjiu.lu@intel.com>
13529 * gcc.c (NO_PIE_SPEC): Delete.
13530 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13532 (LINK_PIE_SPEC): ..to here.
13533 (LINK_COMMAND_SPEC): Support -no-pie.
13534 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13535 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13537 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13538 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13539 (ENDFILE_CRTEND_SPEC): Similarly.
13541 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13544 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13545 (mstack-protector-guard-offset=): Ditto.
13546 * config/i386/i386.c (ix86_option_override): Handle
13547 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13549 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13550 ix86_stack_protect_guard_offset variables.
13551 (TARGET_STACK_PROTECT_GUARD): Always define.
13552 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13553 and -mstack-protector-guard-offset= options.
13555 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13557 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13558 boundary case for the last candidate.
13560 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13562 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13563 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13565 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13567 PR middle-end/19706
13568 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13569 * config/aarch64/aarch64-builtins.c
13570 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13571 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13572 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13574 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13575 Andrew Pinski <pinskia@gmail.com>
13577 PR middle-end/19706
13578 * internal-fn.def (XORSIGN): New.
13579 * optabs.def (xorsign_optab): New.
13580 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13581 (convert_expand_mult_copysign): New.
13582 (pass_optimize_widening_mul::execute): Call
13583 convert_expand_mult_copysign.
13585 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13587 PR tree-optimization/81354
13588 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13589 Insert on edges rather than explicitly creating landing pads.
13590 (analyze_candidates_and_replace): Commit edge inserts.
13592 2017-08-08 Richard Biener <rguenther@suse.de>
13594 PR middle-end/81719
13595 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13596 (expand_simple_operations): Also look through ADDR_EXPRs with
13597 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13599 2017-08-08 Richard Biener <rguenther@suse.de>
13601 PR tree-optimization/81723
13602 * tree-vect-slp.c (struct bst_traits): New hash traits.
13603 (bst_fail): New global.
13604 (vect_build_slp_tree_2): New worker, split out from ...
13605 (vect_build_slp_tree): ... this now wrapping it with using
13606 bst_fail set to cache SLP tree build fails. Properly handle
13608 (vect_analyze_slp_instance): Allocate and free bst_fail.
13610 2017-08-08 Martin Liska <mliska@suse.cz>
13613 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13614 LABEL_DECLs that can be from a different function.
13616 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13618 PR tree-optimization/81744
13619 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13620 loop's number of iterations.
13622 2017-08-08 Martin Liska <mliska@suse.cz>
13624 * asan.c: Include header files.
13625 * attribs.c (build_decl_attribute_variant): New function moved
13627 (build_type_attribute_qual_variant): Likewise.
13628 (cmp_attrib_identifiers): Likewise.
13629 (simple_cst_list_equal): Likewise.
13630 (omp_declare_simd_clauses_equal): Likewise.
13631 (attribute_value_equal): Likewise.
13632 (comp_type_attributes): Likewise.
13633 (build_type_attribute_variant): Likewise.
13634 (lookup_ident_attribute): Likewise.
13635 (remove_attribute): Likewise.
13636 (merge_attributes): Likewise.
13637 (merge_type_attributes): Likewise.
13638 (merge_decl_attributes): Likewise.
13639 (merge_dllimport_decl_attributes): Likewise.
13640 (handle_dll_attribute): Likewise.
13641 (attribute_list_equal): Likewise.
13642 (attribute_list_contained): Likewise.
13643 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13644 (lookup_attribute_by_prefix): Likewise.
13645 * bb-reorder.c: Include header files.
13646 * builtins.c: Likewise.
13647 * calls.c: Likewise.
13648 * cfgexpand.c: Likewise.
13649 * cgraph.c: Likewise.
13650 * cgraphunit.c: Likewise.
13651 * convert.c: Likewise.
13652 * dwarf2out.c: Likewise.
13653 * final.c: Likewise.
13654 * fold-const.c: Likewise.
13655 * function.c: Likewise.
13656 * gimple-expr.c: Likewise.
13657 * gimple-fold.c: Likewise.
13658 * gimple-pretty-print.c: Likewise.
13659 * gimple.c: Likewise.
13660 * gimplify.c: Likewise.
13661 * hsa-common.c: Likewise.
13662 * hsa-gen.c: Likewise.
13663 * internal-fn.c: Likewise.
13664 * ipa-chkp.c: Likewise.
13665 * ipa-cp.c: Likewise.
13666 * ipa-devirt.c: Likewise.
13667 * ipa-fnsummary.c: Likewise.
13668 * ipa-inline.c: Likewise.
13669 * ipa-visibility.c: Likewise.
13671 * lto-cgraph.c: Likewise.
13672 * omp-expand.c: Likewise.
13673 * omp-general.c: Likewise.
13674 * omp-low.c: Likewise.
13675 * omp-offload.c: Likewise.
13676 * omp-simd-clone.c: Likewise.
13677 * opts-global.c: Likewise.
13678 * passes.c: Likewise.
13679 * predict.c: Likewise.
13680 * sancov.c: Likewise.
13681 * sanopt.c: Likewise.
13682 * symtab.c: Likewise.
13683 * toplev.c: Likewise.
13684 * trans-mem.c: Likewise.
13685 * tree-chkp.c: Likewise.
13686 * tree-eh.c: Likewise.
13687 * tree-into-ssa.c: Likewise.
13688 * tree-object-size.c: Likewise.
13689 * tree-parloops.c: Likewise.
13690 * tree-profile.c: Likewise.
13691 * tree-ssa-ccp.c: Likewise.
13692 * tree-ssa-live.c: Likewise.
13693 * tree-ssa-loop.c: Likewise.
13694 * tree-ssa-sccvn.c: Likewise.
13695 * tree-ssa-structalias.c: Likewise.
13696 * tree-ssa.c: Likewise.
13697 * tree-streamer-in.c: Likewise.
13698 * tree-vectorizer.c: Likewise.
13699 * tree-vrp.c: Likewise.
13700 * tsan.c: Likewise.
13701 * ubsan.c: Likewise.
13702 * varasm.c: Likewise.
13703 * varpool.c: Likewise.
13704 * tree.c: Remove functions moved to attribs.[ch].
13705 * tree.h: Likewise.
13706 * config/aarch64/aarch64.c: Add attrs.h header file.
13707 * config/alpha/alpha.c: Likewise.
13708 * config/arc/arc.c: Likewise.
13709 * config/arm/arm.c: Likewise.
13710 * config/avr/avr.c: Likewise.
13711 * config/bfin/bfin.c: Likewise.
13712 * config/c6x/c6x.c: Likewise.
13713 * config/cr16/cr16.c: Likewise.
13714 * config/cris/cris.c: Likewise.
13715 * config/darwin.c: Likewise.
13716 * config/epiphany/epiphany.c: Likewise.
13717 * config/fr30/fr30.c: Likewise.
13718 * config/frv/frv.c: Likewise.
13719 * config/ft32/ft32.c: Likewise.
13720 * config/h8300/h8300.c: Likewise.
13721 * config/i386/winnt.c: Likewise.
13722 * config/ia64/ia64.c: Likewise.
13723 * config/iq2000/iq2000.c: Likewise.
13724 * config/lm32/lm32.c: Likewise.
13725 * config/m32c/m32c.c: Likewise.
13726 * config/m32r/m32r.c: Likewise.
13727 * config/m68k/m68k.c: Likewise.
13728 * config/mcore/mcore.c: Likewise.
13729 * config/microblaze/microblaze.c: Likewise.
13730 * config/mips/mips.c: Likewise.
13731 * config/mmix/mmix.c: Likewise.
13732 * config/mn10300/mn10300.c: Likewise.
13733 * config/moxie/moxie.c: Likewise.
13734 * config/msp430/msp430.c: Likewise.
13735 * config/nds32/nds32-isr.c: Likewise.
13736 * config/nds32/nds32.c: Likewise.
13737 * config/nios2/nios2.c: Likewise.
13738 * config/nvptx/nvptx.c: Likewise.
13739 * config/pa/pa.c: Likewise.
13740 * config/pdp11/pdp11.c: Likewise.
13741 * config/powerpcspe/powerpcspe.c: Likewise.
13742 * config/riscv/riscv.c: Likewise.
13743 * config/rl78/rl78.c: Likewise.
13744 * config/rx/rx.c: Likewise.
13745 * config/s390/s390.c: Likewise.
13746 * config/sh/sh.c: Likewise.
13747 * config/sol2.c: Likewise.
13748 * config/sparc/sparc.c: Likewise.
13749 * config/spu/spu.c: Likewise.
13750 * config/stormy16/stormy16.c: Likewise.
13751 * config/tilegx/tilegx.c: Likewise.
13752 * config/tilepro/tilepro.c: Likewise.
13753 * config/v850/v850.c: Likewise.
13754 * config/vax/vax.c: Likewise.
13755 * config/visium/visium.c: Likewise.
13756 * config/xtensa/xtensa.c: Likewise.
13758 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13761 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13762 constraints since the -mupper-regs-* switches have been
13764 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13765 into a vector from a double word element that was extracted from
13766 another vector, and eliminate extra XXPERMDI instructions.
13767 (vsx_concat_<mode>_2): Likewise.
13768 (vsx_concat_<mode>_3): Likewise.
13769 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13770 concat to allow optimizing inserts from previous extracts.
13772 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13774 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13775 memory reference to a SSP offset in TLS address space.
13776 (ix86_print_operand) <case '@'>: Remove.
13777 (ix86_print_operand_punct_valid_p): Remove '@' code.
13778 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13779 UNSPEC_SP_TLS_TEST.
13780 (stack_tls_protect_set_<mode>): Remove.
13781 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13782 (stack_tls_protect_test_<mode>): Remove.
13783 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13785 2017-08-07 Olivier Hainque <hainque@adacore.com>
13788 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13790 2017-08-07 Douglas Rupp <rupp@adacore.com>
13792 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13793 variable was referenced as multidir in command.
13795 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13798 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13801 2017-08-07 Martin Liska <mliska@suse.cz>
13803 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13804 * config/rl78/rl78.c: Add include of attribs.h.
13805 * config/sh/sh.c: Likewise.
13806 * config/v850/v850.c: Likewise.
13808 2017-08-07 Tom de Vries <tom@codesourcery.com>
13810 PR middle-end/78266
13811 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13813 2017-08-07 Martin Liska <mliska@suse.cz>
13815 * config/mips/mips.c: Include attribs.h.
13817 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13820 * doc/invoke.texi: Document change in behvaior for -Ofast for
13823 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13825 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13827 (aarch64_pop_regs): Likewise.
13828 (aarch64_gen_load_pair): Likewise.
13829 (aarch64_save_callee_saves): Likewise.
13830 (aarch64_restore_callee_saves): Likewise.
13832 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13834 * config/i386/i386.c: Revert the last change.
13836 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13839 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13841 (ix86_finalize_stack_frame_flags): This. Also clear
13842 frame_pointer_needed if -fno-omit-frame-pointer is used without
13844 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13845 with ix86_finalize_stack_frame_flags.
13846 (ix86_expand_epilogue): Likewise.
13847 (ix86_expand_split_stack_prologue): Likewise.
13849 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13852 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13853 to P_AES for Westmere.
13855 2017-08-07 Jonathan Yong <10walls@gmail.com>
13857 * config/i386/mingw.opt (fset-stack-executable): Removed.
13858 * config/i386/cygming.opt (fset-stack-executable): Moved
13860 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13862 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13864 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13866 2017-08-07 Marek Polacek <polacek@redhat.com>
13868 PR middle-end/81737
13869 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13871 2017-08-07 Martin Liska <mliska@suse.cz>
13873 * attribs.h (canonicalize_attr_name): New function.
13874 (cmp_attribs): Move from c-format.c and adjusted.
13875 (is_attribute_p): Moved from tree.h.
13876 * tree-inline.c: Add new includes.
13877 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13878 (private_is_attribute_p): Remove.
13879 (private_lookup_attribute): Likewise.
13880 (private_lookup_attribute_by_prefix): Simplify.
13881 (remove_attribute): Use is_attribute_p.
13882 * tree.h: Remove removed declarations.
13884 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13886 PR middle-end/81698
13887 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13888 instead of computing it in the function. Formatting fix.
13889 (expand_case): Don't rely on default_edge being the first edge,
13890 clear it if removing it, pass default_edge to
13891 emit_case_dispatch_table.
13892 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13895 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13897 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13898 insn in the function, emit NOP after the insn.
13900 2017-08-06 Tom de Vries <tom@codesourcery.com>
13902 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13905 2017-08-06 Tom de Vries <tom@codesourcery.com>
13907 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13910 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13912 PR tree-optimization/57371
13913 * match.pd: New pattern.
13915 2017-08-04 Marek Polacek <polacek@redhat.com>
13917 PR middle-end/81695
13918 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13919 perform the computation in offset_int.
13921 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13923 PR tree-optimization/81136
13924 * tree-vectorizer.h: Include tree-hash-traits.h.
13925 (vec_base_alignments): New typedef.
13926 (vec_info): Add a base_alignments field.
13927 (vect_record_base_alignments): Declare.
13928 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13930 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13931 (create_data_ref): Add an is_conditional_in_stmt argument.
13932 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13933 the is_conditional_in_stmt field.
13934 (data_ref_loc): Add an is_conditional_in_stmt field.
13935 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13936 (find_data_references_in_stmt): Update call to create_data_ref.
13937 (graphite_find_data_references_in_stmt): Likewise.
13938 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13939 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13940 (vect_record_base_alignment): New function.
13941 (vect_record_base_alignments): Likewise.
13942 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13943 for nested statements even if we fail to compute a misalignment.
13944 Use pooled base alignments for unconditional references.
13945 (vect_find_same_alignment_drs): Compare base addresses instead
13947 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13948 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13950 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13952 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13953 Add an explicit name for the enum. Use auto_vec for slp_instances
13954 and grouped_stores.
13955 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13957 (_bb_vec_info): Add a constructor and destructor.
13958 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13959 (destroy_loop_vec_info): Delete.
13960 (vect_destroy_datarefs): Likewise.
13961 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13962 (vec_info::vec_info): New function.
13963 (vec_info::~vec_info): Likewise.
13964 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13965 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13966 destroy_loop_vec_info.
13967 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13968 (_loop_vec_info::_loop_vec_info): ...this.
13969 (destroy_loop_vec_info): Replace with...
13970 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13971 the stmt_vec_infos. Leave handling of vec_info information to its
13972 destructor. Remove explicit vector releases.
13973 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13974 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13975 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13976 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13977 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13978 (destroy_bb_vec_info): Replace with...
13979 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13980 information to its destructor.
13981 (vect_slp_analyze_bb_1): Use new and delete instead of
13982 new_bb_vec_info and destroy_bb_vec_info.
13983 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13986 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13988 * tree-data-ref.h (subscript): Add access_fn field.
13989 (data_dependence_relation): Add could_be_independent_p.
13990 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13991 (same_access_functions): Move to tree-data-ref.c.
13992 * tree-data-ref.c (ref_contains_union_access_p): New function.
13993 (access_fn_component_p): Likewise.
13994 (access_fn_components_comparable_p): Likewise.
13995 (dr_analyze_indices): Add a reference to access_fn_component_p.
13996 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13998 (constant_access_functions): Likewise.
13999 (add_other_self_distances): Likewise.
14000 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14001 (initialize_data_dependence_relation): Use XCNEW and remove
14002 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14003 of access functions that have the same type. Allow the
14004 subsequence to end with different bases in some circumstances.
14005 Record the chosen access functions in SUB_ACCESS_FN.
14006 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14007 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14008 (subscript_dependence_tester_1): Likewise dra and drb.
14009 (build_classic_dist_vector): Update calls accordingly.
14010 (subscript_dependence_tester): Likewise.
14011 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14012 DDR_COULD_BE_INDEPENDENT_P.
14013 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14014 comp_alias_ddrs instead of may_alias_ddrs.
14015 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14017 (vect_analyze_data_ref_dependence): Use it if
14018 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14019 distance vectors if that fails.
14020 (dependence_distance_ge_vf): New function.
14021 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14022 LOOP_VINFO_MAY_ALIAS_DDRS.
14024 2017-08-04 Richard Biener <rguenther@suse.de>
14026 PR middle-end/81705
14027 * fold-const.c (fold_binary_loc): Properly restrict
14028 minus_var0 && minus_var1 case when associating undefined overflow
14031 2017-08-04 Tom de Vries <tom@codesourcery.com>
14033 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14035 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14037 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14038 Don't start diagnostic messages with a capital letter.
14039 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14041 (rs6000_invalid_builtin): Likewise.
14042 (rs6000_trampoline_init): Likewise.
14044 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14047 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14048 after setting changeable df flags.
14050 2017-08-03 Richard Biener <rguenther@suse.de>
14052 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14053 up if the use is in USE - X.
14055 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14057 * toplev.c (dumpfile.h): New include.
14058 (internal_error_reentered): New static function. Use it...
14059 (internal_error_function): ...here to handle reentered internal_error.
14061 2017-08-03 Richard Biener <rguenther@suse.de>
14063 PR middle-end/81148
14064 * fold-const.c (split_tree): Add minus_var and minus_con
14065 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14066 here but always use minus_*.
14067 (associate_trees): Assert we never associate with MINUS_EXPR
14068 and NULL first operand. Do not recurse for PLUS_EXPR operands
14069 when associating as MINUS_EXPR either.
14070 (fold_binary_loc): Track minus_var and minus_con.
14072 2017-08-03 Tom de Vries <tom@codesourcery.com>
14075 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14076 ACCEL_COMPILER, apply finish_options on
14077 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14079 2017-08-03 Tom de Vries <tom@codesourcery.com>
14082 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14083 function_entry_patch_area_size > 0.
14085 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14088 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14089 instead of 10??LU, perform unit multiplication in wide_int,
14090 don't change alloc_object_size_limit if the limit is larger
14093 PR tree-optimization/81655
14094 PR tree-optimization/81588
14095 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14096 the case when ranges[i].low and high are 1 for unsigned type with
14099 PR middle-end/81052
14100 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14101 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14103 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14105 * tree-vrp.h: Add include guard.
14107 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14110 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14111 (ud2): New insn pattern.
14112 * config/i386/i386.c (ix86_expand_epilogue):
14113 For naked functions, generate ud2 instead of trap insn.
14115 2017-08-02 Marek Polacek <polacek@redhat.com>
14118 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14120 2017-08-02 Tom de Vries <tom@codesourcery.com>
14121 Cesar Philippidis <cesar@codesourcery.com>
14123 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14124 Add missing edge probabilities.
14126 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14128 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14129 Correct endianness.
14131 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14133 PR middle-end/79499
14134 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14135 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14136 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14137 prologue_seq sequences - if any.
14139 2017-08-02 Richard Biener <rguenther@suse.de>
14141 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14142 via vectors if supported, integer extracts via punning if supported
14143 or otherwise vector extracts.
14145 2017-08-02 Richard Biener <rguenther@suse.de>
14147 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14149 (bitmap_insert_into_set): ... this.
14151 2017-08-02 Richard Biener <rguenther@suse.de>
14153 PR tree-optimization/81633
14155 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14157 PR tree-optimization/71752
14158 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14160 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14162 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14163 (machine_function::call_ms2sysv_pad_out): Remove field.
14164 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14165 (ix86_compute_frame_layout): Likewise.
14167 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14170 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14171 attribute with interrupt attribute.
14173 2017-08-01 Andrew Pinski <apinski@cavium.com>
14175 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14176 BIT_INSERT_EXPR's operand 1
14177 to see if the types precision matches.
14179 2017-08-01 Martin Liska <mliska@suse.cz>
14181 PR middle-end/70140
14182 * builtins.c (expand_builtin_memcpy_args): Remove.
14183 (expand_builtin_memcpy): Call newly added function
14184 expand_builtin_memory_copy_args.
14185 (expand_builtin_memcpy_with_bounds): Likewise.
14186 (expand_builtin_mempcpy): Remove last argument.
14187 (expand_builtin_mempcpy_with_bounds): Likewise.
14188 (expand_builtin_memory_copy_args): New function created from
14189 expand_builtin_mempcpy_args with small modifications.
14190 (expand_builtin_mempcpy_args): Remove.
14191 (expand_builtin_stpcpy): Remove unused argument.
14192 (expand_builtin): Likewise.
14193 (expand_builtin_with_bounds): Likewise.
14195 2017-08-01 Martin Liska <mliska@suse.cz>
14198 Make mempcpy more optimal (PR middle-end/70140).
14200 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14203 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14204 __builtin_vec_cmpne verify both arguments are compatible vectors
14205 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14206 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14207 move computation of aligned to after checking the argument types.
14211 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14212 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14215 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14216 Doug Rupp <rupp@adacore.com>
14217 Olivier Hainque <hainque@adacore.com>
14219 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14220 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14222 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14223 for TARGET_OS_CPP_BUILTIN.
14224 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14225 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14227 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14228 passing required abi options to the assembler for EABI configurations.
14229 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14230 of .text.hot and .text.unlikely sections for kernel modules when
14231 using ARM style exceptions.
14232 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14233 options. Add EXTRA_CC1_SPEC.
14234 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14236 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14238 (ARM_TARGET2_DWARF_FORMAT): Define.
14239 * config/arm/t-vxworks: Adjust multilib control to removal of the
14240 Diab command line options.
14242 2017-08-01 Martin Liska <mliska@suse.cz>
14244 PR gcov-profile/81561
14245 * gcov.c (unblock): Make unblocking safe as we need to preserve
14246 index correspondence of blocks and block_lists.
14248 2017-08-01 Richard Biener <rguenther@suse.de>
14250 PR tree-optimization/81181
14251 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14252 (compute_antic): ... end of iteration here.
14254 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14256 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14257 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14258 (ftree-slp-vectorize): Likewise.
14259 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14260 can no longer be set independent of flag_tree_loop_vectorize.
14261 * omp-general.c (emp_max_vf): Likewise.
14262 * opts.c (enable_fdo_optimizations): Remove references to
14263 flag_tree_vectorize, these are now implicit.
14264 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14265 and leave it for the options machinery.
14267 2017-08-01 Martin Liska <mliska@suse.cz>
14269 PR middle-end/70140
14270 * builtins.c (expand_builtin_memcpy_args): Remove.
14271 (expand_builtin_memcpy): Call newly added function
14272 expand_builtin_memory_copy_args.
14273 (expand_builtin_memcpy_with_bounds): Likewise.
14274 (expand_builtin_mempcpy): Remove last argument.
14275 (expand_builtin_mempcpy_with_bounds): Likewise.
14276 (expand_builtin_memory_copy_args): New function created from
14277 expand_builtin_mempcpy_args with small modifications.
14278 (expand_builtin_mempcpy_args): Remove.
14279 (expand_builtin_stpcpy): Remove unused argument.
14280 (expand_builtin): Likewise.
14281 (expand_builtin_with_bounds): Likewise.
14283 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14286 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14287 print "ds:" only for immediates in generic address space.
14289 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14292 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14293 (ix86_function_ok_for_sibcall): Return false for naked functions.
14295 2017-08-01 Richard Biener <rguenther@suse.de>
14297 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14298 (compute_antic): Seed worklist with exit block predecessors.
14299 * cfganal.c (dfs_find_deadend): For a cycle return the source
14300 of the edge closing it.
14302 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14304 * config/aarch64/aarch64.c
14305 (aarch64_can_const_movi_rtx_p): Move 0 check.
14307 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14309 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14310 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14312 * match.pd: Ditto in address comparison pattern.
14314 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14316 PR tree-optimization/81627
14317 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14318 closed ssa form for store-store chain.
14320 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14322 PR tree-optimization/81620
14323 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14324 for store-store chain.
14326 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14328 PR tree-optimization/81588
14329 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14330 ranges[i].in_p, invert comparison code ccode. For >/>=,
14331 swap rhs1 and rhs2 and comparison code unconditionally,
14332 for </<= don't do that. Don't swap rhs1/rhs2 again if
14333 ranges[i].in_p, instead invert comparison code ccode if
14334 opcode or oe->rank is BIT_IOR_EXPR.
14337 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14338 a direct optab to conversion optab.
14339 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14340 with GET_MODE_INNER as last argument instead of optab_handler.
14341 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14342 vector extraction if possible and optab is available.
14343 * expr.c (store_constructor): Use convert_optab_handler instead
14344 of optab_handler. Use vector initialization from smaller
14345 vectors if possible and optab is available.
14346 * tree-vect-stmts.c (vectorizable_load): Likewise.
14347 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14348 now have two modes.
14349 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14350 of vec_init from half-sized vectors with the same element mode.
14351 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14352 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14353 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14354 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14355 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14356 after mode in gen_vec_extract* calls.
14357 (vec_extract<mode>): Renamed to ...
14358 (vec_extract<mode><ssescalarmodelower>): ... this.
14359 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14360 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14361 element mode after mode in gen_vec_init* calls.
14362 (VEC_INIT_HALF_MODE): New mode iterator.
14363 (vec_init<mode>): Renamed to ...
14364 (vec_init<mode><ssescalarmodelower>): ... this.
14365 (vec_init<mode><ssehalfvecmodelower>): New expander.
14366 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14367 (vec_extractv2sfsf): ... this.
14368 (vec_initv2sf): Renamed to ...
14369 (vec_initv2sfsf): ... this.
14370 (vec_extractv2si): Renamed to ...
14371 (vec_extractv2sisi): ... this.
14372 (vec_initv2si): Renamed to ...
14373 (vec_initv2sisi): ... this.
14374 (vec_extractv4hi): Renamed to ...
14375 (vec_extractv4hihi): ... this.
14376 (vec_initv4hi): Renamed to ...
14377 (vec_initv4hihi): ... this.
14378 (vec_extractv8qi): Renamed to ...
14379 (vec_extractv8qiqi): ... this.
14380 (vec_initv8qi): Renamed to ...
14381 (vec_initv8qiqi): ... this.
14382 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14383 (vec_init<mode>): Renamed to ...
14384 (vec_init<mode><VEC_base_l>): ... this.
14385 (vec_extract<mode>): Renamed to ...
14386 (vec_extract<mode><VEC_base_l>): ... this.
14387 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14388 (vec_initv2sfsf): ... this.
14389 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14390 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14391 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14392 element mode after mode in gen_vec_init* calls.
14393 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14394 (vec_init<mode><Vel>): ... this.
14395 (vec_extract<mode>): Renamed to ...
14396 (vec_extract<mode><Vel>): ... this.
14397 * config/aarch64/iterators.md (Vel): New mode attribute.
14398 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14399 Add element mode after mode in gen_vec_extract* calls.
14400 * config/s390/vector.md (non_vec_l): New mode attribute.
14401 (vec_extract<mode>): Renamed to ...
14402 (vec_extract<mode><non_vec_l>): ... this.
14403 (vec_init<mode>): Renamed to ...
14404 (vec_init<mode><non_vec_l>): ... this.
14405 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14406 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14408 * config/arm/iterators.md (V_elem_l): New mode attribute.
14409 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14410 (vec_extract<mode><V_elem_l>): ... this.
14411 (vec_extractv2di): Renamed to ...
14412 (vec_extractv2didi): ... this.
14413 (vec_init<mode>): Renamed to ...
14414 (vec_init<mode><V_elem_l>): ... this.
14415 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14416 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14417 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14418 Add element mode after gen_vec_extract* calls.
14419 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14420 (vec_init<mode><unitmode>): ... this.
14421 (vec_extract<mode>): Renamed to ...
14422 (vec_extract<mode><unitmode>): ... this.
14423 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14424 (vec_init<mode><unitmode>): ... this.
14425 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14426 (vec_initv2sfsf): ... this.
14427 (vec_extractv2sf): Renamed to ...
14428 (vec_extractv2sfsf): ... this.
14429 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14430 Add element mode after gen_vec_extract* calls.
14431 * config/mips/mips.md (unitmode): New mode iterator.
14432 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14433 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14434 * config/spu/spu.md (inner_l): New mode attribute.
14435 (vec_init<mode>): Renamed to ...
14436 (vec_init<mode><inner_l>): ... this.
14437 (vec_extract<mode>): Renamed to ...
14438 (vec_extract<mode><inner_l>): ... this.
14439 * config/sparc/sparc.md (veltmode): New mode iterator.
14440 (vec_init<VMALL:mode>): Renamed to ...
14441 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14442 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14443 (vec_initv2sisi): ... this.
14444 (vec_initv2sf): Renamed to ...
14445 (vec_initv2sfsf): ... this.
14446 (vec_extractv2sf): Renamed to ...
14447 (vec_extractv2sfsf): ... this.
14448 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14449 (vec_init<mode>): Renamed to ...
14450 (vec_init<mode><VEC_base_l>): ... this.
14451 (vec_extract<mode>): Renamed to ...
14452 (vec_extract<mode><VEC_base_l>): ... this.
14453 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14454 (vec_initv2sfsf): ... this.
14455 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14456 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14457 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14458 gen_vec_init* calls.
14460 2017-08-01 Richard Biener <rguenther@suse.de>
14462 PR tree-optimization/81297
14463 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14464 TREE_OVERFLOW from INTEGER_CSTs.
14466 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14468 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14470 2017-07-31 Carl Love <cel@us.ibm.com>
14472 * config/rs6000/rs6000-c: Add support for built-in functions
14473 vector signed char vec_xl_be (signed long long, signed char *);
14474 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14475 vector signed int vec_xl_be (signed long long, signed int *);
14476 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14477 vector signed long long vec_xl_be (signed long long, signed long long *);
14478 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14479 vector signed short vec_xl_be (signed long long, signed short *);
14480 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14481 vector double vec_xl_be (signed long long, double *);
14482 vector float vec_xl_be (signed long long, float *);
14483 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14484 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14485 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14487 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14488 (altivec_expand_builtin): Add switch statement to call
14489 altivec_expand_xl_be for each builtin.
14490 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14491 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14492 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14493 __builtin_vsx_le_be_v16qi.
14494 * doc/extend.texi: Update the built-in documentation file for the
14495 new built-in functions.
14497 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14500 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14502 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14504 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14506 * config.gcc: Add z14.
14507 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14508 CPU model numbers for z13s and z14.
14509 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14511 * config/s390/s390-opts.h (enum processor_type): Rename
14512 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14513 * config/s390/s390.c (processor_table): Add field for CPU name to
14514 be passed to Binutils.
14515 (s390_asm_output_machine_for_arch): Use the new field in
14516 processor_table for Binutils.
14517 (s390_expand_builtin): Replace arch12 with z14.
14518 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14519 (s390_get_sched_attrmask): Likewise.
14520 (s390_get_unit_mask): Likewise.
14521 * config/s390/s390.opt: Add z14 to processor_type enum.
14523 2017-07-31 Martin Jambor <mjambor@suse.cz>
14526 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14527 regardless of optimization level.
14529 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14530 Martin Liska <mliska@suse.cz>
14532 * predict.def: Remove old comment and adjust probability.
14533 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14534 PREDICT statements.
14536 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14539 * config/i386/i386.c (ix86_function_naked): New function.
14540 (ix86_can_use_return_insn_p): Return false for naked functions.
14541 (ix86_expand_prologue): Skip prologue for naked functions.
14542 (ix86_expand_epilogue): Skip epilogue for naked functions
14543 and emit trap instruction.
14544 (ix86_warn_func_return): New function.
14545 (ix86_attribute_table): Add "naked" attribute specification.
14546 (TARGET_WARN_FUNC_RETURN): Define.
14547 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14549 2017-07-31 Martin Liska <mliska@suse.cz>
14551 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14552 (dump_gimple_bb_header): Always dump BB info.
14553 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14555 2017-07-31 Martin Liska <mliska@suse.cz>
14558 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14559 also with current_function_decl non-null equality.
14561 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14564 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14565 change type to the element type, instead add eltype variable and
14566 use it where we are interested in the element type.
14568 PR tree-optimization/81603
14569 * ipa-polymorphic-call.c
14570 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14571 offset arithmetic in offset_int, bail out if the resulting bit offset
14572 doesn't fit into shwi.
14574 2017-07-31 Martin Liska <mliska@suse.cz>
14576 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14577 (gimplify_save_expr): Fix comment.
14579 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14582 * config/i386/i386.c (ix86_function_arg): Update arguments for
14584 (ix86_compute_frame_layout): Set the initial stack offset to
14585 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14586 INCOMING_FRAME_SP_OFFSET.
14587 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14588 stack before exception handler returns.
14589 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14590 the 'ERROR_CODE' for exception handler.
14592 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14594 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14595 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14596 (ASM_OUTPUT_REG_POP): Ditto.
14597 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14598 instead of asm_fprintf to output pure string.
14600 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14602 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14603 to imported_module_or_decl hook.
14604 (debug_nothing_tree_tree_tree_bool): Remove.
14605 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14606 * debug.c (do_nothing_debug_hooks): Use
14607 debug_nothing_tree_tree_tree_bool_bool instead of
14608 debug_nothing_tree_tree_tree_bool.
14609 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14610 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14611 * sdbout.c (sdb_debug_hooks): Likewise.
14612 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14613 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14615 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14616 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14617 attribute, don't add anything.
14619 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14621 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14622 (fold_build2_stat_loc): Likewise.
14623 (fold_build3_stat_loc): Likewise.
14624 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14625 (fold_build1_loc): Remove macro.
14626 (fold_build2_loc): Likewise.
14627 (fold_build3_loc): Likewise.
14629 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14631 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14632 (gimple_build_debug_bind_source_stat): Likewise.
14633 * gimple.h (gimple_build_debug_bind): Remove macro.
14634 (gimple_build_debug_bind_source): Likewise.
14636 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14638 * bitmap.c (bitmap_alloc): Adjust.
14639 (bitmap_gc_alloc): Likewise.
14640 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14642 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14644 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14645 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14646 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14647 (bitmap_gc_alloc_stat): Likewise.
14648 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14650 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14652 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14653 * rtl.h (shallow_copy_rtx): Remove macro.
14655 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14657 * emit-rtl.c (gen_raw_REG): Adjust.
14658 * gengenrtl.c (gendef): Likewise.
14659 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14660 * rtl.h (rtx_alloc): Remove macro.
14662 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14664 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14665 (build_tree_list_stat): Likewise.
14666 * tree.h (build_tree_list): Remove macro.
14667 (build_tree_list_vec): Likewise.
14669 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14671 * tree.c (make_vector_stat): Remove _stat from name.
14672 (build_vector_stat): Likewise.
14673 * tree.h (make_vector_stat): Remove macro.
14674 (build_vector_stat): Likewise.
14676 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14678 * tree.h (build_var_debug_value): Remove prototype.
14680 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14682 * tree.c (tree_cons_stat): Remove _stat from name.
14683 * tree.h (tree_cons): Remove macro.
14685 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14687 * tree.c (build_vl_exp_stat): Remove _stat from name.
14688 * tree.h (build_vl_exp): Remove macro.
14690 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14692 * tree.c (build_decl_stat): Remove _stat from name.
14693 * tree.h (build_decl): Remove macro.
14695 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14697 * gimple.c (gimple_build_with_ops_stat): Adjust.
14698 (gimple_alloc_stat): Remove _stat from name.
14699 * gimple.h (gimple_alloc): Remove macro.
14701 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14703 * tree.c (make_tree_vec_stat): Remove _stat from name.
14704 (grow_tree_vec_stat): Likewise.
14705 * tree.h (make_tree_vec_stat): Adjust prototype.
14706 (grow_tree_vec_stat): Likewise.
14707 (make_tree_vec): Remove macro.
14708 (grow_tree_vec): Likewise.
14710 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14712 * fold-const.c (fold_build1_stat_loc): Adjust.
14713 (fold_build2_stat_loc): Likewise.
14714 (fold_build3_stat_loc): Likewise.
14715 * tree.c (build0_stat): Remove _stat from name.
14716 (build1_stat): Likewise.
14717 (build2_stat): Likewise.
14718 (build3_stat): Likewise.
14719 (build4_stat): Likewise.
14720 (build5_stat): Likewise.
14721 * tree.h (build1_loc): Remove macro, and rename _stat function
14723 (build2_loc): Likewise.
14724 (build3_loc): Likewise.
14725 (build4_loc): Likewise.
14726 (build5_loc): Likewise.
14728 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14730 * tree.c (make_int_cst_stat): Remove _stat from name.
14731 * tree.h (make_int_cst_stat): Adjust prototype.
14732 (make_int_cst): Remove macro.
14734 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14736 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14737 * tree.h (make_tree_binfo_stat): Adjust prototype.
14738 (make_tree_binfo): Remove.
14740 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14742 * tree.c (copy_node_stat): Rename to copy_node.
14743 (build_distinct_type_copy): Adjust.
14744 * tree.h (copy_node_stat): Adjust prototype.
14745 (copy_node): Remove macro.
14747 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14749 * tree.c (make_node_stat): rename to make_node.
14750 (build_tree_list_stat): Adjust.
14751 (build0_stat): Likewise.
14752 (build2_stat): Likewise.
14753 (build3_stat): Likewise.
14754 (build4_stat): Likewise.
14755 (build5_stat): Likewise.
14756 (build_decl_stat): Likewise.
14757 * tree.h (make_node_stat): Adjust prototype.
14758 (make_node): remove macro.
14760 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14762 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14763 (PPC_FEATURE2_SCV): Likewise.
14764 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14766 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14768 * config/aarch64/aarch64.c
14769 (aarch64_internal_mov_immediate): Add new special pattern.
14770 * config/aarch64/aarch64.md (*movdi_aarch64):
14771 Add reg/32bit const mov case.
14773 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14774 Richard Sandiford <richard.sandiford@linaro.org>
14776 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14777 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14778 Add integer and movi cases.
14779 (movi-split-hf-df-sf split, fp16): New.
14780 (enabled): Added TARGET_FP_F16INST.
14781 * config/aarch64/iterators.md (GPF_HF): New.
14782 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14784 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14786 * config/aarch64/aarch64.c
14787 (aarch64_simd_container_mode): Add prototype.
14788 (aarch64_expand_mov_immediate): Add HI support.
14789 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14790 (aarch64_can_const_movi_rtx_p): New.
14791 (aarch64_preferred_reload_class):
14792 Remove restrictions of using FP registers for certain SIMD operations.
14793 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14794 (aarch64_valid_floating_const): Add integer move validation.
14795 (aarch64_simd_imm_scalar_p): Remove.
14796 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14797 (aarch64_legitimate_constant_p): Expand list of supported cases.
14798 * config/aarch64/aarch64-protos.h
14799 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14800 (aarch64_reinterpret_float_as_int): New.
14801 (aarch64_simd_imm_scalar_p): Remove.
14802 * config/aarch64/constraints.md (Uvi): New.
14803 (Dd): Split into Ds and new Dd.
14804 * config/aarch64/aarch64.md (*movsi_aarch64):
14806 (*movdi_aarch64): Add SIMD mov case.
14808 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14810 * tree-predcom.c: (struct chain): Handle store-store chain in which
14811 stores for elimination only store loop invariant values.
14812 (execute_pred_commoning_chain): Ditto.
14813 (prepare_initializers_chain_store_elim): Ditto.
14814 (prepare_finalizers): Ditto.
14815 (is_inv_store_elimination_chain): New function.
14816 (initialize_root_vars_store_elim_1): New function.
14818 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14820 * tree-predcom.c: Revise general description of the pass.
14821 (enum chain_type): New enum type for store elimination.
14822 (struct chain): New field supporting store elimination.
14823 (struct component): Ditto.
14824 (dump_chain): Dump store-stores chain.
14825 (release_chain): Release resources.
14826 (split_data_refs_to_components): Compute and create component
14827 contains only stores for elimination.
14828 (get_chain_last_ref_at): New function.
14829 (make_invariant_chain): Initialization.
14830 (make_rooted_chain): Specify chain type in parameter and record it.
14831 (add_looparound_copies): Skip for store-stores chain.
14832 (determine_roots_comp): Compute type of chain and pass it to
14834 (initialize_root_vars_store_elim_2): New function.
14835 (finalize_eliminated_stores): New function.
14836 (remove_stmt): Handle store for elimination.
14837 (execute_pred_commoning_chain): Execute predictive commoning on
14838 store-store chains.
14839 (determine_unroll_factor): Skip unroll for store-stores chain.
14840 (prepare_initializers_chain_store_elim): New function.
14841 (prepare_initializers_chain): Hanlde store-store chain.
14842 (prepare_finalizers_chain, prepare_finalizers): New function.
14843 (tree_predictive_commoning_loop): Return integer value indicating
14844 if loop is unrolled or lcssa form is corrupted.
14845 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14847 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14849 * tree-predcom.c (initialize_root): Delete.
14850 (execute_pred_commoning_chain): Initialize root vars and replace
14851 reference of non-combined chain directly, rather than call above
14854 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14856 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14857 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14859 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14861 * tree-predcom.c (struct chain): New field init_seq.
14862 (release_chain): Release init_seq.
14863 (prepare_initializers_chain): Record intialization stmts in above
14865 (insert_init_seqs): New function.
14866 (tree_predictive_commoning_loop): Call insert_init_seqs.
14868 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14870 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14872 2017-07-28 Richard Biener <rguenther@suse.de>
14874 * match.pd: Remove superfluous :c.
14875 * genmatch.c (simplify::id): Add member.
14876 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14878 (current_id): New global.
14879 (dt_node::parent): Move from ...
14880 (dt_operand::parent): ... here. Add for_id member.
14881 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14882 (decision_tree::find_node): Relax order requirement when
14883 merging DT_TRUE nodes to ones inbetween the current simplify
14884 and the one we try to merge with. Add diagnostic whenever
14885 we need to enforce pattern order by not merging.
14886 (decision_tree::insert): Set current_id.
14887 (decision_tree::print_node): Dump parent node and for_id.
14888 (parser::last_id): Add member.
14889 (parser::push_simplify): Assign unique id.
14890 (parser::parser): Initialize last_id.
14892 2017-07-28 Martin Liska <mliska@suse.cz>
14895 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14896 gimple_build_debug_bind.
14898 2017-07-28 Richard Biener <rguenther@suse.de>
14900 PR tree-optimization/81502
14901 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14903 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14905 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14906 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14907 for BIT_FIELD_REF args.
14908 * fold-const.c (make_bit_field_ref): Likewise.
14909 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14911 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14914 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14915 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14916 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14917 Or it into SANITIZER_UNDEFINED.
14918 * ubsan.c: Include gimple-fold.h and varasm.h.
14919 (ubsan_expand_ptr_ifn): New function.
14920 (instrument_pointer_overflow): New function.
14921 (maybe_instrument_pointer_overflow): New function.
14922 (instrument_object_size): Formatting fix.
14923 (pass_ubsan::execute): Call instrument_pointer_overflow
14924 and maybe_instrument_pointer_overflow.
14925 * internal-fn.c (expand_UBSAN_PTR): New function.
14926 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14927 * sanitizer.def (__ubsan_handle_pointer_overflow,
14928 __ubsan_handle_pointer_overflow_abort): New builtins.
14929 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14930 * internal-fn.def (UBSAN_PTR): New internal function.
14931 * opts.c (sanitizer_opts): Add pointer-overflow.
14932 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14933 * fold-const.c (build_range_check): Compute pointer range check in
14934 integral type if pointer arithmetics would be needed. Formatting
14937 2017-07-28 Martin Liska <mliska@suse.cz>
14940 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14941 parameters that are of a variable-length.
14943 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14945 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14947 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14948 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14949 (CRT_CALL_STATIC_FUNCTION): Likewise.
14950 (ASM_DEFAULT_SPEC): New define.
14951 (ASM_SPEC32): Likewise.
14952 (ASM_SPEC64): Likewise.
14953 (ASM_SPEC_COMMON): Likewise.
14954 (ASM_SPEC): Likewise.
14955 (INVALID_64BIT): Likewise.
14956 (LINK_OS_DEFAULT_SPEC): Likewise.
14957 (LINK_OS_SPEC32): Likewise.
14958 (LINK_OS_SPEC64): Likewise.
14959 (POWERPC_LINUX): Likewise.
14960 (PTRDIFF_TYPE): Likewise.
14961 (RESTORE_FP_PREFIX): Likewise.
14962 (RESTORE_FP_SUFFIX): Likewise.
14963 (SAVE_FP_PREFIX): Likewise.
14964 (SAVE_FP_SUFFIX): Likewise.
14965 (SIZE_TYPE): Likewise.
14966 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14967 (TARGET_64BIT): Likewise.
14968 (TARGET_64BIT): Likewise.
14969 (TARGET_AIX): Likewise.
14970 (WCHAR_TYPE_SIZE): Likewise.
14971 (WCHAR_TYPE): Undefine.
14972 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14973 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14974 (CPP_OS_RTEMS_SPEC): Delete.
14975 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14976 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14978 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14980 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14982 PR tree-optimization/81578
14983 * tree-parloops.c (build_new_reduction): Bail out if
14984 reduction_code isn't one of the standard OpenMP reductions.
14985 Move the details printing after that decision.
14987 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14989 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14990 related to reload_in_progress.
14991 (splat_input_operand): Likewise.
14992 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14994 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14996 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14997 (TARGET_INSTANTIATE_DECLS): Likewise.
14998 (legitimate_indexed_address_p): Delete reload_in_progress code.
14999 (rs6000_debug_legitimate_address_p): Likewise.
15000 (rs6000_eliminate_indexed_memrefs): Likewise.
15001 (rs6000_emit_le_vsx_store): Likewise.
15002 (rs6000_emit_move_si_sf_subreg): Likewise.
15003 (rs6000_emit_move): Likewise.
15004 (register_to_reg_type): Likewise.
15005 (rs6000_pre_atomic_barrier): Likewise.
15006 (rs6000_machopic_legitimize_pic_address): Likewise.
15007 (rs6000_allocate_stack_temp): Likewise.
15008 (rs6000_address_for_fpconvert): Likewise.
15009 (rs6000_address_for_altivec): Likewise.
15010 (rs6000_secondary_memory_needed_rtx): Delete function.
15011 (rs6000_check_sdmode): Likewise.
15012 (rs6000_alloc_sdmode_stack_slot): Likewise.
15013 (rs6000_instantiate_decls): Likewise.
15014 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15015 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15016 Delete reload_in_progress.
15017 (*vec_reload_and_plus_<mptrsize>): Likewise.
15018 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15019 (vsx_div_v2di): Likewise.
15020 (vsx_udiv_v2di): Likewise.
15022 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15024 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15025 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15026 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15027 (rs6000_debug_reg_global): Delete print of LRA status.
15028 (rs6000_option_override_internal): Delete dead LRA related code.
15029 (rs6000_lra_p): Delete function.
15030 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15032 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15034 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15035 * config/riscv/rtems.h: New file.
15037 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15038 Sudakshina Das <sudi.das@arm.com>
15040 * config/aarch64/aarch64.md
15041 (define_split for and<mode>3nr_compare): Move
15042 non aarch64_logical_operand to a register.
15043 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15044 register immediate operand to a register.
15045 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15047 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15049 PR middle-end/81564
15050 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15052 2017-07-27 Richard Biener <rguenther@suse.de>
15054 PR tree-optimization/81573
15055 PR tree-optimization/81494
15056 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15057 multi defuse cycle case.
15059 2017-07-27 Richard Biener <rguenther@suse.de>
15061 PR tree-optimization/81571
15062 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15065 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15067 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15068 earlier and only if MASK_FPU is set. Adjust formatting.
15070 2017-07-27 Martin Liska <mliska@suse.cz>
15072 * opt-functions.awk: Add validation of value of Init.
15073 * optc-gen.awk: Pass new argument.
15075 2017-07-27 Martin Liska <mliska@suse.cz>
15077 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15078 Fix wrong condition.
15080 2017-07-27 Martin Liska <mliska@suse.cz>
15082 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15083 BBs and edges seen by autoFDO.
15085 2017-07-27 Richard Biener <rguenther@suse.de>
15087 PR tree-optimization/81502
15088 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15089 with incompatible but same sized type.
15090 (execute_update_addresses_taken): Likewise.
15092 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15094 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15095 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15097 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15100 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15101 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15102 Change s_operand to memory_operand.
15104 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15106 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15107 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15108 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15109 Emit instructions rather than returning an expression. Handle TFmode
15110 and KFmode by casting to TImode.
15111 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15112 (rs6000_emit_le_vsx_store): Likewise.
15113 * config/rs6000/vsx.md (VSX_TI): New iterator.
15114 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15115 (*vsx_le_undo_permute_<mode>): Likewise.
15116 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15117 emit the split sequence.
15118 (*vsx_le_perm_store_<mode>): Likewise.
15120 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15122 PR tree-optimization/81555
15123 PR tree-optimization/81556
15124 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15125 if true, force CHANGED for the recursive invocation.
15126 (reassociate_bb): Remember original length of ops array, pass
15127 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15129 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15130 attributes for noipa attribute. For naked attribute use
15131 lookup_attribute first before lookup_attribute_spec.
15132 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15134 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15135 for functions with noipa attribute.
15136 (cgraph_externally_visible_p): Return true for functions with noipa
15138 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15139 for functions with noipa attribute.
15140 * doc/extend.texi: Document noipa function attribute.
15141 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15142 also for functions with noipa attribute.
15143 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15145 2017-07-26 Andrew Pinski <apinski@cavium.com>
15147 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15148 vec_unalign_load_cost and vec_unalign_store_cost.
15150 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15152 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15153 -mvsx-small-integer option.
15154 (ISA_3_0_MASKS_IEEE): Likewise.
15155 (OTHER_VSX_VECTOR_MASKS): Likewise.
15156 (POWERPC_MASKS): Likewise.
15157 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15158 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15159 code, only testing for DImode being allowed in non-VSX floating
15161 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15162 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15164 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15165 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15166 TARGET_P8_VECTOR test.
15167 (rs6000_secondary_reload_simple_move): Likewise.
15168 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15169 since TARGET_P9_VECTOR was already tested.
15170 (rs6000_opt_masks): Remove -mvsx-small-integer.
15171 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15172 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15174 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15175 test for TARGET_VEXTRACTUB was used, and that uses
15177 (p9 extract splitter): Likewise.
15178 (vsx_extract_<mode>_di_p9): Likewise.
15179 (vsx_extract_<mode>_store_p9): Likewise.
15180 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15181 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15182 the elimination of TARGET_VSX_SMALL_INTEGER.
15183 (vsx_extract_<mode>_p8): Likewise.
15184 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15185 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15186 (vsx_set_<mode>_p9): Likewise.
15187 (vsx_set_v4sf_p9): Likewise.
15188 (vsx_set_v4sf_p9_zero): Likewise.
15189 (vsx_insert_extract_v4sf_p9): Likewise.
15190 (vsx_insert_extract_v4sf_p9_2): Likewise.
15191 * config/rs6000/rs6000.md (sign extend splitter): Change
15192 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15193 (floatsi<mode>2_lfiwax_mem): Likewise.
15194 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15195 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15196 since a test for TARGET_P9_VECTOR was used.
15197 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15198 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15199 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15200 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15201 TARGET_P8_VECTOR test.
15202 (fix_trunc<mode>si2_stfiwx): Likewise.
15203 (fix_trunc<mode>si2_internal): Likewise.
15204 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15205 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15207 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15208 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15209 TARGET_P8_VECTOR test.
15210 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15211 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15212 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15214 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15215 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15216 since a test for TARGET_P9_VECTOR was used.
15217 (splitter for loading small constants): Likewise.
15219 2017-07-26 Andrew Pinski <apinski@cavium.com>
15221 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15224 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15227 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15228 (fp_valid_at): Likewise.
15230 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15232 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15233 (qdf24xx_addrcost_table): Likewise.
15234 (cortexa57_tunings): Update to use generic_branch_cost.
15235 (cortexa72_tunings): Likewise.
15236 (cortexa73_tunings): Likewise.
15237 (qdf24xx_tunings): Likewise.
15239 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15241 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15242 (thunderx2t99_branch_cost): Likewise.
15243 (cortexa35_tunings): Update to use generic_branch_cost.
15244 (cortexa53_tunings): Likewise.
15245 (cortexa57_tunings): Likewise.
15246 (cortexa72_tunings): Likewise.
15247 (cortexa73_tunings): Likewise.
15248 (thunderx2t99_tunings): Likewise.
15250 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15252 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15253 (sparc_option_override): Honour MASK_FSMULD.
15254 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15255 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15256 * config/sparc/sparc.opt (mfsmuld): New option.
15257 * doc/invoke.texi (mfsmuld): Document option.
15259 2017-07-26 Marek Polacek <polacek@redhat.com>
15261 PR middle-end/70992
15262 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15264 2017-07-26 Richard Biener <rguenther@suse.de>
15266 * gimple-match-head.c (do_valueize): Return OP if valueize
15268 (get_def): New helper to get at the def stmt of a SSA name
15269 if valueize allows.
15270 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15271 do_valueize to get at the def stmt.
15272 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15274 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15276 PR middle-end/46932
15277 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15279 2017-07-26 Martin Liska <mliska@suse.cz>
15282 * function.c (expand_function_start): Make expansion of
15283 nonlocal_goto_save_area after parm_birth_insn.
15285 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15287 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15288 from all CPU target flags enable members.
15290 2017-07-26 Richard Biener <rguenther@suse.de>
15292 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15293 (decision_tree::gen): Make 'type' const.
15294 (write_predicate): Likewise.
15296 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15298 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15299 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15300 (rs6000_option_override_internal): Likewise.
15301 (rs6000_expand_vector_set): Likewise.
15302 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15303 (TARGET_UPPER_REGS_SF): Likewise.
15304 (TARGET_UPPER_REGS_DI): Likewise.
15305 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15306 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15307 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15308 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15309 (Splitters for DI constants in Altivec registers): Likewise.
15310 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15311 (vsx_set_v4sf_p9): Likewise.
15312 (vsx_set_v4sf_p9_zero): Likewise.
15313 (vsx_insert_extract_v4sf_p9): Likewise.
15314 (vsx_insert_extract_v4sf_p9_2): Likewise.
15316 2017-07-25 Carl Love <cel@us.ibm.com>
15318 * doc/extend.texi: Update the built-in documentation file for the
15319 existing built-in functions
15320 vector signed char vec_cnttz (vector signed char);
15321 vector unsigned char vec_cnttz (vector unsigned char);
15322 vector signed short vec_cnttz (vector signed short);
15323 vector unsigned short vec_cnttz (vector unsigned short);
15324 vector signed int vec_cnttz (vector signed int);
15325 vector unsigned int vec_cnttz (vector unsigned int);
15326 vector signed long long vec_cnttz (vector signed long long);
15327 vector unsigned long long vec_cnttz (vector unsigned long long);
15329 2017-07-25 Andrew Pinski <apinski@cavium.com>
15331 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15332 accesses where the use is for the first operand of a BIT_INSERT.
15334 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15337 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15338 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15340 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15342 * config/i386/gstabs.h: Delete.
15343 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15345 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15347 * config/i386/i386.c (ix86_decompose_address): Do not check for
15348 register RTX when looking at index_reg or base_reg.
15349 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15351 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15353 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15354 to update EH info here.
15356 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15358 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15360 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15362 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15364 2017-07-25 Torsten Duwe <duwe@suse.de>
15366 * common.opt: Introduce -fpatchable-function-entry
15367 command line option, and its variables function_entry_patch_area_size
15368 and function_entry_patch_area_start.
15369 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15370 including a two-value parser.
15371 * target.def (print_patchable_function_entry): New target hook.
15372 * targhooks.h (default_print_patchable_function_entry): New function.
15373 * targhooks.c (default_print_patchable_function_entry): Likewise.
15374 * toplev.c (process_options): Switch off IPA-RA if
15375 patchable function entries are being generated.
15376 * varasm.c (assemble_start_function): Look at the
15377 patchable-function-entry command line switch and current
15378 function attributes and maybe generate NOP instructions by
15379 calling the print_patchable_function_entry hook.
15380 * doc/extend.texi: Document patchable_function_entry attribute.
15381 * doc/invoke.texi: Document -fpatchable_function_entry
15382 command line option.
15383 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15385 * doc/tm.texi: Re-generate.
15387 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15390 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15391 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15393 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15395 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15397 2017-07-25 Richard Biener <rguenther@suse.de>
15399 PR tree-optimization/81455
15400 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15401 not walk in cycles when looking for guards.
15403 2017-07-25 Richard Biener <rguenther@suse.de>
15405 PR tree-optimization/81529
15406 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15407 when optimizing backedge uses.
15409 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15411 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15413 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15414 to dl_section_ref. On AIX, append an expression to subtract
15415 the size of the section length to dl_section_ref.
15417 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15419 * configure.ac: If any of the config.* scripts fail, exit 1.
15420 * configure: Regenerate.
15422 2017-07-25 Richard Biener <rguenther@suse.de>
15424 PR middle-end/81546
15425 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15426 of immediate uses, be more verbose on errors.
15428 2017-07-25 Richard Biener <rguenther@suse.de>
15430 PR tree-optimization/81510
15431 * tree-vect-loop.c (vect_is_simple_reduction): When the
15432 reduction stmt is not inside the loop bail out.
15434 2017-07-25 Richard Biener <rguenther@suse.de>
15436 PR tree-optimization/81303
15437 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15438 profitability check against LOOP_VINFO_NITERSM1.
15440 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15442 * domwalk.c (cmp_bb_postorder): Simplify.
15443 (sort_bbs_postorder): New function. Use it...
15444 (dom_walker::walk): ...here to optimize common cases.
15446 2017-07-25 Martin Liska <mliska@suse.cz>
15449 * ipa-visibility.c (function_and_variable_visibility): Make the
15450 redirection just on target that supports aliasing.
15451 Fix GNU coding style.
15453 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15456 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15458 * config/arm/rtems.h: Likewise.
15459 * config/bfin/rtems.h: Likewise.
15460 * config/i386/rtemself.h: Likewise.
15461 * config/lm32/rtems.h: Likewise.
15462 * config/m32c/rtems.h: Likewise.
15463 * config/m68k/rtemself.h: Likewise.
15464 * config/microblaze/rtems.h: Likewise.
15465 * config/mips/rtems.h: Likewise.
15466 * config/moxie/rtems.h: Likewise.
15467 * config/nios2/rtems.h: Likewise.
15468 * config/powerpcspe/rtems.h: Likewise.
15469 * config/rs6000/rtems.h: Likewise.
15470 * config/rtems.h: Likewise.
15471 * config/sh/rtems.h: Likewise.
15472 * config/sh/rtemself.h: Likewise.
15473 * config/sparc/rtemself.h: Likewise.
15475 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15478 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15479 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15480 * tree-ssa-structalias.c (alias_get_name): Same.
15482 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15485 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15486 instructions if no du chain is found.
15488 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15490 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15492 2017-07-25 Richard Biener <rguenther@suse.de>
15494 PR middle-end/81505
15495 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15498 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15500 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15501 upper-regs options.
15502 (ISA_2_7_MASKS_SERVER): Likewise.
15503 (ISA_3_0_MASKS_IEEE): Likewise.
15504 (OTHER_P8_VECTOR_MASKS): Likewise.
15505 (OTHER_VSX_VECTOR_MASKS): Likewise.
15506 (POWERPC_MASKS): Likewise.
15507 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15508 duplicate list of options.
15509 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15510 explicit -mupper-regs options.
15511 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15512 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15513 alias for -mupper-regs-df.
15514 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15515 (rs6000_init_hard_regno_mode_ok): Likewise.
15516 (rs6000_option_override_internal): Likewise.
15517 (rs6000_opt_masks): Likewise.
15518 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15519 options in terms of whether -mvsx or -mpower8-vector was used.
15520 (TARGET_UPPER_REGS_DI): Likewise.
15521 (TARGET_UPPER_REGS_SF): Likewise.
15522 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15523 -mupper-regs-* options.
15525 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15527 * passes.c (emergency_dump_function): Print some empty lines and a
15528 header before the RTL dump.
15530 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15532 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15534 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15537 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15538 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15540 2017-07-24 Carl Love <cel@us.ibm.com>
15542 * config/rs6000/rs6000-c.c: Add support for built-in functions
15543 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15544 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15545 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15546 vec_extract_fp_from_shortl): Add defines for the two builtins.
15547 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15548 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15550 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15551 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15552 * doc/extend.texi: Update the built-in documentation file for the
15553 new built-in function.
15555 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15558 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15560 * doc/generic.texi: Likewise.
15561 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15562 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15564 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15566 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15567 (aarch64_mls_elt_merge<mode>): Likewise.
15569 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15571 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15572 having __cxa_atexit.
15574 2017-07-23 Michael Collison <michael.collison@arm.com>
15576 * config/arm/arm.c (arm_option_override): Deprecate
15577 use of -mstructure-size-boundary.
15578 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15579 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15581 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15584 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15585 Reduce cost estimate for direct moves.
15587 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15590 * config/i386/i386.c (ix86_option_override_internal): Disable
15591 BMI, BMI2 and TBM instructions for -m16.
15593 2017-07-21 Carl Love <cel@us.ibm.com>
15595 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15596 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15597 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15598 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15599 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15600 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15601 VMULOSW): New enum "unspec" values.
15602 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15603 altivec_vmulosw): New patterns.
15604 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15605 VMULOSW): Add definitions.
15607 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15609 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15610 (qdf24xx): Likewise.
15611 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15612 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15613 (AARCH64_FL_V8_1): Renumber.
15614 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15615 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15616 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15617 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15618 rdma to feature modifiers list.
15620 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15622 PR middle-end/56727
15623 * ipa-visibility (function_and_variable_visibility): Convert
15624 recursive PLT call to direct call if appropriate.
15626 2017-07-21 Andrew Pinski <apinski@cavium.com>
15628 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15629 operand 1 to see if the types precision matches.
15630 * fold-const.c (operand_equal_p): Likewise.
15632 2017-07-21 Richard Biener <rguenther@suse.de>
15634 PR tree-optimization/81303
15635 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15636 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15637 (vect_peeling_hash_get_lowest_cost): Adjust.
15638 (vect_enhance_data_refs_alignment): Likewise. Use
15639 vect_get_peeling_costs_all_drs to compute the penalty for no
15640 peeling to match up costs.
15642 2017-07-21 Richard Biener <rguenther@suse.de>
15644 PR tree-optimization/81500
15645 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15646 we didn't identify a reduction path.
15648 2017-07-21 Tom de Vries <tom@codesourcery.com>
15649 Cesar Philippidis <cesar@codesourcery.com>
15651 PR gcov-profile/81442
15652 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15655 2017-07-21 Tom de Vries <tom@codesourcery.com>
15658 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15660 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15661 nvptx_override_options_after_change.
15663 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15665 * dwarf2out.c (output_file_names): Avoid double testing for
15666 dwarf_version >= 5.
15668 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15670 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15672 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15674 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15676 (try_crossjump_to_edge): Do not punt on partitioned functions.
15678 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15680 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15681 Put all BBs reachable only via paths crossing cold region to cold
15683 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15685 2016-07-21 Richard Biener <rguenther@suse.de>
15687 PR tree-optimization/81303
15688 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15689 into account prologue and epilogue iterations when raising
15690 min_profitable_iters to sth at least covering one vector iteration.
15692 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15694 * config/arm/arm.c (arm_test_cpu_arch_dat):
15697 2017-07-20 Nathan Sidwell <nathan@acm.org>
15699 Remove TYPE_METHODS.
15700 * tree.h (TYPE_METHODS): Delete.
15701 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15702 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15703 (dbxout_type_methods): Scan TYPE_FIELDS.
15704 (dbxout_type): Don't check TYPE_METHODS here.
15705 * function.c (use_register_for_decl): Always ignore register for
15706 class types when not optimizing.
15707 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15708 * tree.c (free_lang_data_in_type): Stitch out member functions and
15709 templates from TYPE_FIELDS.
15710 (build_distinct_type_copy, verify_type_variant,
15711 verify_type): Member fns are on TYPE_FIELDS.
15712 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15713 * tree-pretty-print.c (dump_generic_node): Likewise.
15715 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15718 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15719 V2TImode and V4TImode.
15720 (ix86_expand_vector_extract): Likewise.
15721 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15722 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15723 (ssescalarmode): Handle V4TImode and V2TImode.
15724 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15725 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15726 (VEXTRACTI128_MODE): New mode iterator.
15727 (splitter for *vec_extractv?ti first element): New.
15728 (VEC_INIT_MODE): New mode iterator.
15729 (vec_init<mode>): Consolidate 3 expanders into one using
15730 VEC_INIT_MODE mode iterator.
15732 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15734 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15735 non_spilled_static_chain_regno_p.
15737 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15739 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15741 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15743 * bb-reorder.c (connect_traces): Allow copying of blocks within
15746 2017-07-20 Richard Biener <rguenther@suse.de>
15748 * gimple.h (gimple_phi_result): Add gphi * overload.
15749 (gimple_phi_result_ptr): Likewise.
15750 (gimple_phi_arg): Likewise. Adjust index assert to only
15751 allow actual argument accesses rather than all slots available
15753 (gimple_phi_arg_def): Add gphi * overload.
15754 * tree-phinodes.c (make_phi_node): Initialize only actual
15756 (resize_phi_node): Clear memory not covered by old node,
15757 do not initialize excess argument slots.
15758 (reserve_phi_args_for_new_edge): Initialize new argument slot
15761 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15763 PR tree-optimization/81388
15765 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15767 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15768 by removing computation of may_be_zero.
15770 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15771 Tom de Vries <tom@codesourcery.com>
15773 PR middle-end/81030
15774 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15775 when gimple level profile disagrees with what RTL expander did.
15777 2017-07-20 Richard Biener <rguenther@suse.de>
15779 PR tree-optimization/61171
15780 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15781 (vect_analyze_stmt): Add slp instance parameter.
15782 (vectorizable_reduction): Likewise.
15783 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15784 (vect_is_simple_reduction): Deal with chains not detected
15785 as SLP reduction chain, specifically not properly associated
15786 chains containing a mix of plus/minus.
15787 (get_reduction_op): Remove.
15788 (get_initial_defs_for_reduction): Simplify, pass in whether
15789 this is a reduction chain, pass in the SLP node for the PHIs.
15790 (vect_create_epilog_for_reduction): Get the SLP instance as
15792 (vectorizable_reduction): Get the SLP instance as arg.
15793 During analysis remember the SLP node with the PHIs in the
15794 instance. Simplify getting at the vectorized reduction PHIs.
15795 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15796 through SLP instance.
15797 (vect_slp_analyze_operations): Likewise.
15798 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15799 (vect_transform_stmt): Likewise.
15801 2017-07-20 Tom de Vries <tom@codesourcery.com>
15803 PR tree-optimization/81489
15804 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15805 read of phi arg location to before loop that modifies phi.
15807 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15809 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15812 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15814 PR middle-end/81331
15815 * except.c (execute): Fix ordering issue.
15817 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15819 PR rtl-optimization/81423
15820 * combine.c (make_compound_operation_int): Don't try to optimize
15821 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15823 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15825 PR rtl-optimization/81423
15826 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15827 with a constant that is -1 in the truncated to mode.
15829 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15831 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15832 (determine_unlikely_bbs): ... here.
15833 * predict.h (propagate_unlikely_bbs_forward): Declare.
15834 * cfgexpand.c (pass_expand::execute): Use it.
15835 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15837 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15838 propagate_unlikely_bbs_forward.
15840 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15842 PR middle-end/81331
15843 * except.c (maybe_add_nop_after_section_switch): New function.
15846 2017-07-19 Tom de Vries <tom@codesourcery.com>
15848 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15850 2017-07-19 Tom de Vries <tom@codesourcery.com>
15852 * gimple.h (gimple_phi_arg): Make assert more strict.
15854 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15856 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15857 * config/rs6000/mmintrin.h: New file.
15858 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15860 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15862 PR tree-optimization/81346
15863 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15865 2017-07-19 Tom de Vries <tom@codesourcery.com>
15867 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15869 2017-07-19 Tom de Vries <tom@codesourcery.com>
15871 * config/nvptx/nvptx-modes.def: Add V2DImode.
15872 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15873 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15874 (nvptx_output_mov_insn): Handle lack of mov.b128.
15875 (nvptx_print_operand): Handle 'H' and 'L' codes.
15876 (nvptx_vector_mode_supported): Allow V2DImode.
15877 (nvptx_preferred_simd_mode): New function.
15878 (nvptx_data_alignment): New function.
15879 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15880 nvptx_preferred_simd_mode.
15881 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15883 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15885 2017-07-19 Tom de Vries <tom@codesourcery.com>
15887 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15888 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15889 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15890 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15891 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15892 (mov<VECIM>_insn): New define_insn.
15893 (define_expand "mov<VECIM>): New define_expand.
15895 2017-07-19 Tom de Vries <tom@codesourcery.com>
15897 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15899 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15901 PR tree-optimization/81346
15902 * fold-const.h (fold_div_compare, range_check_type): Declare.
15903 * fold-const.c (range_check_type): New function.
15904 (build_range_check): Use range_check_type.
15905 (fold_div_compare): No longer static, rewritten into
15906 a match.pd helper function.
15907 (fold_comparison): Don't call fold_div_compare here.
15908 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15909 as helper function.
15911 2017-07-19 Nathan Sidwell <nathan@acm.org>
15913 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15914 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15915 * tree.c (find_decls_types_r, verify_type): Use
15916 TYPE_{MIN,MAX}_VALUE_RAW.
15917 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15918 (hash_tree): Likewise.
15919 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15921 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15924 2017-07-18 Tom de Vries <tom@codesourcery.com>
15926 PR middle-end/81464
15927 * omp-expand.c (expand_omp_for_static_chunk): Handle
15928 equal-argument loop exit phi.
15930 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15933 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15934 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15935 operand 2 predicate.
15936 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15937 operand 2 predicate.
15938 (ror,rol -> rorx splitters): Use const_int_operand as
15939 operand 2 predicate.
15941 2017-06-18 Richard Biener <rguenther@suse.de>
15943 PR tree-optimization/81410
15944 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15945 the gap in the ! slp_perm SLP case after each group.
15947 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15949 PR middle-end/81463
15950 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15953 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15955 PR middle-end/81462
15956 * predict.c (set_even_probabilities): Cleanup; do not affect
15957 probabilities that are already known.
15958 (combine_predictions_for_bb): Call even when count is set.
15960 2017-07-18 Nathan Sidwell <nathan@acm.org>
15962 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15965 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15968 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15969 optimization for loop niter analysis.
15971 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15974 * config/avr/avr.c (avr_optimize_casesi): Don't use
15975 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15977 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15979 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15980 body_cost_vec from _vect_peel_extended_info.
15981 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15982 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15985 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15987 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15989 2017-07-18 Richard Biener <rguenther@suse.de>
15991 PR tree-optimization/80620
15992 PR tree-optimization/81403
15993 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15994 info when re-using a VN table entry.
15996 2017-07-18 Richard Biener <rguenther@suse.de>
15998 PR tree-optimization/81418
15999 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16000 vectype_in. Verify that with lane-reducing reduction operations
16001 we have a single def-use cycle.
16003 2017-07-17 Carl Love <cel@us.ibm.com>
16005 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16007 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16008 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16009 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16010 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16011 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16012 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16013 VMULOSW): New enum "unspec" values.
16014 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16015 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16016 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16017 altivec_vmulosw): New patterns.
16018 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16019 VMULOSW): Add definitions.
16020 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16022 * config/alpha/alpha.c: Include predict.h.
16024 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16026 * tree-vrp.c (compare_assert_loc): Fix comparison function
16027 to return predictable results.
16029 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16031 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16033 (subdi3): Likewise.
16034 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16035 * doc/invoke.texi (mexpand-adddi): Update text.
16037 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16039 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16040 that also clobbers the CC register. The old expand code is moved
16042 (*arc_clzsi2): ... here.
16043 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16044 the CC register. The old expand code is moved to ...
16045 (arc_ctzsi2): ... here.
16047 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16049 * config/arc/arc.opt (mindexed-loads): Use initial value
16050 TARGET_INDEXED_LOADS_DEFAULT.
16051 (mauto-modify-reg): Use initial value
16052 TARGET_AUTO_MODIFY_REG_DEFAULT.
16053 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16054 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16055 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16056 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16058 2017-07-17 Martin Liska <mliska@suse.cz>
16061 * opts.c (finish_options): Do not allow -fgnu-tm
16062 w/ -fsanitize={kernel-,}address. Say sorry.
16064 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16067 * tree-loop-distribution.c (classify_partition): Only assert on
16068 numer of iterations.
16069 (merge_dep_scc_partitions): Delete prameter. Update function call.
16070 (distribute_loop): Remove code handling loop with unknown niters.
16071 (pass_loop_distribution::execute): Skip loop with unknown niters.
16073 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16076 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16077 function sort_partitions_by_post_order.
16079 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16081 PR tree-optimization/81374
16082 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16083 the max index of basic blocks, rather than number of basic blocks.
16085 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16087 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16089 (arc_legitimate_pic_operand_p): Likewise.
16090 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16092 (arc_needs_pcl_p): Likewise.
16093 (arc_legitimate_pc_offset_p): Likewise.
16094 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16095 function is also used in constrains.md.
16096 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16097 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16098 PLUS. Only return true/false in known cases, otherwise assert.
16099 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16100 is already called in arc_legitimate_constant_p.
16101 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16103 (LEGITIMATE_PIC_OPERAND_P): Use
16104 arc_raw_symbolic_reference_mentioned_p function.
16105 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16110 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16111 Andrew Burgess <andrew.burgess@embecosm.com>
16113 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16114 (arc_return_address_register): New function.
16115 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16116 (arc_handle_fndecl_attribute): Add naked attribute.
16117 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16118 (TARGET_WARN_FUNC_RETURN): Likewise.
16119 (arc_allocate_stack_slots_for_args): New function.
16120 (arc_warn_func_return): Likewise.
16121 (machine_function): Change type fn_type.
16122 (arc_compute_function_type): Consider new naked function type,
16123 change function return type.
16124 (arc_must_save_register): Adapt to handle new
16125 arc_compute_function_type's return type.
16126 (arc_expand_prologue): Likewise.
16127 (arc_expand_epilogue): Likewise.
16128 (arc_return_address_regs): Delete.
16129 (arc_return_address_register): New function.
16130 (arc_epilogue_uses): Use above function.
16131 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16132 (arc_function_type): Change encoding, add naked type.
16133 (ARC_INTERRUPT_P): Change to handle the new encoding.
16134 (ARC_FAST_INTERRUPT_P): Likewise.
16135 (ARC_NORMAL_P): Define.
16136 (ARC_NAKED_P): Likewise.
16137 (arc_compute_function_type): Delete prototype.
16138 * config/arc/arc.md (in_ret_delay_slot): Use
16139 arc_return_address_register function.
16140 (simple_return): Likewise.
16141 (p_return_i): Likewise.
16143 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16145 PR tree-optimization/81428
16146 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16147 can't be built for those types.
16149 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16151 Remove stuff dead since r239246.
16153 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16154 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16155 (avr_inform_devices): Remove dead stuff.
16157 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16159 * config/arm/arm_neon.h: Fix softp typo.
16161 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16163 PR tree-optimization/81365
16164 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16165 aggregate moves onto bb predecessor edges, make sure there are no
16166 loads that could alias the lhs in between the start of bb and the
16169 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16172 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16173 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16174 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16176 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16178 PR tree-optimization/81396
16179 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16180 (init_symbolic_number): Initialize it to 1.
16181 (perform_symbolic_merge): Add n_ops from both operands into the new
16183 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16184 without base_addr as useless if they need more than one operation.
16185 (bswap_replace): Handle !bswap case for NULL base_addr.
16187 2017-07-17 Tom de Vries <tom@codesourcery.com>
16190 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16193 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16195 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16196 conditional builtin define __FIX_LEON3FT_B2BST.
16198 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16200 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16201 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16204 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16206 PR rtl-optimization/81424
16207 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16208 to remove potential trapping from operands if -fnon-call-exceptions.
16210 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16212 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16213 profile_proability for scalling.
16214 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16216 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16218 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16220 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16222 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16223 fixpoint arithmetics.
16225 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16227 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16228 fixpoint arithmetics.
16230 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16232 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16233 fixpoint arithmetics.
16235 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16237 * profile-count.h (profile_probability::from_reg_br_prob_note,
16238 profile_probability::to_reg_br_prob_note): New functions.
16239 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16240 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16241 * predict.c (probability_reliable_p): Update.
16242 (edge_probability_reliable_p): Update.
16243 (br_prob_note_reliable_p): Update.
16244 (invert_br_probabilities): Update.
16245 (add_reg_br_prob_note): New function.
16246 (combine_predictions_for_insn): Update.
16247 * asan.c (asan_clear_shadow): Update.
16248 * cfgbuild.c (compute_outgoing_frequencies): Update.
16249 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16250 (update_br_prob_note): Update.
16251 (rtl_verify_edges): Update.
16252 (purge_dead_edges): Update.
16253 (fixup_reorder_chain): Update.
16254 * emit-rtl.c (try_split): Update.
16255 * ifcvt.c (cond_exec_process_insns): Update.
16256 (cond_exec_process_if_block): Update.
16257 (dead_or_predicable): Update.
16258 * internal-fn.c (expand_addsub_overflow): Update.
16259 (expand_neg_overflow): Update.
16260 (expand_mul_overflow): Update.
16261 * loop-doloop.c (doloop_modify): Update.
16262 * loop-unroll.c (compare_and_jump_seq): Update.
16263 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16264 * predict.h: Update.
16265 * reorg.c (mostly_true_jump): Update.
16267 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16268 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16269 * config/arc/arc.c: (emit_unlikely_jump): Update.
16270 * config/arm/arm.c: (emit_unlikely_jump): Update.
16271 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16272 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16273 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16274 (ix86_print_operand): Update.
16275 (ix86_split_fp_branch): Update.
16276 (predict_jump): Update.
16277 * config/ia64/ia64.c (ia64_print_operand): Update.
16278 * config/mmix/mmix.c (mmix_print_operand): Update.
16279 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16280 (rs6000_expand_split_stack_prologue): Update.
16281 * config/rs6000/rs6000.c: Update.
16282 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16283 (s390_expand_vec_movstr): Update.
16284 (s390_expand_cs_tdsi): Update.
16285 (s390_expand_split_stack_prologue): Update.
16286 * config/sh/sh.c (sh_print_operand): Update.
16287 (expand_cbranchsi4): Update.
16288 (expand_cbranchdi4): Update.
16289 * config/sparc/sparc.c (output_v9branch): Update.
16290 * config/spu/spu.c (get_branch_target): Update.
16291 (ea_load_store_inline): Update.
16292 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16293 * config/tilepro/tilepro.c: Update.
16295 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16297 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16298 (gimplify_save_expr): Likewise.
16300 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16302 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16304 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16306 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16308 * timevar.def (TV_IPA_FNSUMMARY): Define.
16310 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16312 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16313 to back store errata sensitive sequence from being generated.
16314 (sqrtdf2_fix): Likewise.
16316 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16318 * tree-ssa-threadupdate.c (compute_path_counts,
16319 update_joiner_offpath_counts): Use profile_probability.
16321 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16324 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16326 * config/arm/arm-c.c (arm_cpu_builtins): Define
16327 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16329 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16331 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16332 array entries to represent __ieee128 versions of the
16333 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16334 scalar_extract_sig, and scalar_insert_exp built-in functions.
16335 (altivec_resolve_overloaded_builtin): Add special case handling
16336 for the __builtin_scalar_insert_exp function, as represented by
16337 the P9V_BUILTIN_VEC_VSIEDP constant.
16338 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16339 exponent support for __ieee128 argument.
16340 (VSESQP): Add scalar extract signature support for __ieee128
16342 (VSTDCNQP): Add scalar test negative support for __ieee128
16344 (VSIEQP): Add scalar insert exponent support for __int128 argument
16345 with __ieee128 result.
16346 (VSIEQPF): Add scalar insert exponent support for __ieee128
16347 argument with __ieee128 result.
16348 (VSTDCQP): Add scalar test data class support for __ieee128
16350 (VSTDCNQP): Add overload support for scalar test negative with
16351 __ieee128 argument.
16352 (VSTDCQP): Add overload support for scalar test data class
16353 __ieee128 argument.
16354 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16355 UNSPEC_VSX_SXSIGDP.
16356 (UNSPEC_VSX_SIEXPQP): New constant.
16357 (xsxexpqp): New insn for VSX scalar extract exponent quad
16359 (xsxsigqp): New insn for VSX scalar extract significand quad
16361 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16362 precision with floating point argument.
16363 (xststdcqp): New expand for VSX scalar test data class quad
16365 (xststdcnegqp): New expand for VSX scalar test negative quad
16367 (xststdcqp): New insn to match expansions for VSX scalar test data
16368 class quad precision and VSX scalar test negative quad precision.
16369 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16370 special case operand checking to enforce that second operand of
16371 VSX scalar test data class with quad precision argument is a 7-bit
16373 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16374 prototypes and descriptions of __ieee128 versions of
16375 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16376 scalar_test_data_class, and scalar_test_neg built-in functions.
16378 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16380 PR tree-optimization/81162
16381 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16382 replace a negate with an add.
16384 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16386 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16388 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16390 * config/arm/arm-c.c (arm_cpu_builtins): Define
16391 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16393 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16395 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16396 (armv8-r): Set ARM Cortex-R52 as default CPU.
16397 * config/arm/arm-tables.opt: Regenerate.
16398 * config/arm/arm-tune.md: Regenerate.
16399 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16401 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16402 extension for -mcpu=cortex-r52.
16404 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16406 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16407 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16408 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16409 (fp-armv8): Define it as FP_ARMv8 only.
16410 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16411 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16413 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16414 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16415 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16416 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16417 than TARGET_FPU_ARMV8.
16418 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16419 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16420 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16422 * config/arm/neon.md (neon_vrint): Likewise.
16423 (neon_vcvt): Likewise.
16424 (neon_<fmaxmin_op><mode>): Likewise.
16425 (<fmaxmin><mode>3): Likewise.
16426 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16427 * config/arm/predicates.md (arm_cond_move_operator): Check against
16428 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16430 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16432 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16433 to top of function.
16435 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16437 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16438 loop in comment with memset.
16440 2017-07-14 Martin Liska <mliska@suse.cz>
16442 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16443 * dwarf2out.c (is_java): Remove the function.
16444 (output_pubname): Remove usage of the function.
16445 (lower_bound_default): Remove usage of DW_LANG_Java.
16446 (gen_compile_unit_die): Likewise.
16447 * gcc.c: Remove compiler defaults for .java and .zip files.
16448 * gimple-expr.c (remove_suffix): Change as there's no longer
16449 extension than 4-letter one.
16450 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16451 (gimplify_save_expr): Likewise.
16452 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16453 as it's possible even for other languages than Java.
16454 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16455 * lto-opts.c (lto_write_options): Remove reference to Java.
16456 * opts.c (strip_off_ending): Update file extension handling.
16457 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16458 * tree-eh.c (lower_resx): Likewise.
16459 * tree.c (free_lang_data_in_type): Remove dead code.
16460 (find_decls_types_r): Likewise.
16461 (build_common_builtin_nodes): Remove Java from a comment.
16462 (verify_type): Remove dead code.
16463 * varasm.c (assemble_external): Remove Java from a comment.
16465 2017-07-14 Martin Liska <mliska@suse.cz>
16467 * opts.c (finish_options): Add quotes.
16468 (common_handle_option): Likewise.
16470 2017-07-14 Martin Liska <mliska@suse.cz>
16472 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16473 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16474 Remove N_SO_PASCAL.
16475 * dwarf2out.c (lower_bound_default): Do not handle
16477 (gen_compile_unit_die): Likewise.
16478 * gcc.c: Remove default extension binding for GNU Pascal.
16479 * stmt.c: Remove Pascal language from a comment.
16480 * xcoffout.c: Likewise.
16482 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16485 * diagnostic-show-locus.c (fixit_cmp): New function.
16486 (layout::layout): Sort m_fixit_hints.
16487 (column_range::column_range): Assert that the values are valid.
16488 (struct char_span): New struct.
16489 (correction::overwrite): New method.
16490 (struct source_line): New struct.
16491 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16492 calls in terms of classes source_line and char_span, and
16493 correction::overwrite.
16494 (selftest::test_overlapped_fixit_printing_2): New function.
16495 (selftest::diagnostic_show_locus_c_tests): Call it.
16497 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16499 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16500 early if there is no lhs.
16502 2017-07-13 Martin Liska <mliska@suse.cz>
16504 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16505 (gen_reference_type_die): Likewise.
16506 * stor-layout.c: Remove Pascal-related comment.
16508 2017-07-13 Martin Liska <mliska@suse.cz>
16510 * opts.c (finish_options): Add quotes to error messages.
16511 (parse_sanitizer_options): Likewise.
16513 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16515 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16517 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16519 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16521 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16523 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16525 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16527 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16530 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16531 provides the hardware capability bits, define the macro
16532 __BUILTIN_CPU_SUPPORTS__.
16533 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16534 if GLIBC does not provide the hardware capability bits. Add a
16535 gcc_unreachable call if the built-in cpu function is neither
16536 __builtin_cpu_is nor __builtin_cpu_supports.
16537 (rs6000_get_function_versions_dispatcher): Change the warning
16538 that an old GLIBC is used which does not export the capability
16539 bits to be an error.
16540 * doc/extend.texi (target_clones attribute): Document the
16541 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16542 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16543 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16544 the macros defined by GCC if the newer GLIBC is available.
16546 2017-07-12 Jeff Law <law@redhat.com>
16548 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16549 remaining includes slightly.
16550 * config/riscv/riscv-builtins.c: Include profile-count.h.
16552 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16555 * config/avr/avr.c (avr_set_current_function): In diagnostic
16556 messages: Quote keywords and (parts of) identifiers.
16557 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16560 2017-07-12 Carl Love <cel@us.ibm.com>
16562 * config/rs6000/rs6000-c.c: Add support for built-in functions
16563 vector bool char vec_revb (vector bool char);
16564 vector bool short vec_revb (vector short char);
16565 vector bool int vec_revb (vector bool int);
16566 vector bool long long vec_revb (vector bool long long);
16567 * doc/extend.texi: Update the built-in documentation file for the
16568 new built-in functions.
16570 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16572 * config/s390/s390.md: Remove movcc splitter.
16574 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16576 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16577 load/store on condition.
16579 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16582 * config/avr/avr.c (avr_encode_section_info)
16583 [progmem && !TREE_READONLY]: Error if progmem object needs
16586 2017-07-11 Michael Collison <michael.collison@arm.com>
16588 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16591 2017-07-11 Carl Love <cel@us.ibm.com>
16593 * config/rs6000/rs6000-c.c: Add support for builtins
16594 vector unsigned int vec_parity_lsbb (vector signed int);
16595 vector unsigned int vec_parity_lsbb (vector unsigned int);
16596 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16597 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16598 vector unsigned long long vec_parity_lsbb (vector signed long long);
16599 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16600 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16601 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16602 * doc/extend.texi: Update the built-in documentation file for the
16603 new built-in functions.
16605 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16607 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16608 (layout::m_primary_loc): New field.
16609 (layout::layout): Initialize new field. Move location filtering
16610 logic from here to...
16611 (layout::maybe_add_location_range): ...this new method. Add
16612 support for filtering to just the lines already specified by other
16614 (layout::will_show_line_p): New method.
16615 (gcc_rich_location::add_location_if_nearby): New method.
16616 (selftest::test_add_location_if_nearby): New test function.
16617 (selftest::diagnostic_show_locus_c_tests): Call it.
16618 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16621 2017-07-11 Tom de Vries <tom@codesourcery.com>
16623 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16624 (bb_first_real_insn): New function.
16625 (nvptx_single): Add extra initialization of broadcasted condition
16628 2017-07-11 Nathan Sidwell <nathan@acm.org>
16630 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16632 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16634 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16635 Binutils doc as TEXI will mess them up.
16636 * doc/invoke.texi (AVR Options): Same here.
16638 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16640 * config/sparc/sparc.opt (mfix-ut700): New option.
16641 (mfix-gr712rc): Likewise.
16642 (sparc_fix_b2bst): New variable.
16643 * doc/invoke.texi (SPARC options): Document them.
16644 (ARM options): Fix warnings.
16645 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16646 instructions to prevent sequences that can trigger the store-store
16647 errata for certain LEON3FT processors.
16648 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16649 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16650 * config/sparc/sparc.md (fix_b2bst): New attribute.
16651 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16653 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16656 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16658 (*rsqrtsf2_sse): Ditto.
16660 (div<mode>3): Macroize insn from divdf3 and divsf3
16661 using MODEF mode iterator.
16663 2017-07-10 Martin Sebor <msebor@redhat.com>
16665 PR tree-optimization/80397
16666 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16667 instead of testing for equality to INTEGER_TYPE.
16669 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16671 * config.gcc: Remove uclibc from arc target spec.
16673 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16675 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16677 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16680 * lto-wrapper.c (remove_option): New function.
16681 (merge_and_complain): Merge PIC/PIE options more realistically.
16683 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16685 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16689 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16690 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16691 * config.in: Regenerate.
16692 * configure: Regenerate.
16693 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16694 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16695 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16696 (TARGET_GASISR_PROLOGUES): ...target mask.
16697 * common/config/avr/avr-common.c
16698 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16699 Set -mgas-isr-prologues.
16700 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16701 INSERT_PASS_BEFORE for it.
16702 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16703 * config/avr/avr.c (avr_option_override)
16704 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16705 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16706 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16707 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16708 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16710 (make_avr_pass_pre_proep): New function.
16711 (emit_push_sfr) <treg>: Add argument to function and use it
16712 instead of TMP_REG.
16713 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16714 and set machine->gasisr.yes.
16715 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16716 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16717 __gcc_isr.n_pushed to .L__stack_usage.
16718 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16719 (avr_asm_final_postscan_insn): ...this new static function.
16720 * config/avr/avr.h (machine_function)
16721 <is_no_gccisr, use_L__stack_usage>: New fields.
16722 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16723 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16724 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16725 (gasisr, *gasisr): New expander and insn.
16726 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16727 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16728 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16730 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16732 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16735 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16737 Move jump-tables out of .text again.
16740 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16741 (ASM_OUTPUT_ADDR_VEC): New function.
16742 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16743 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16744 INSN_ADDRESSes as asm comment.
16745 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16746 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16747 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16748 * config/avr/avr.md (*tablejump): Adjust comment.
16749 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16750 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16752 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16753 (avr_output_addr_vec): New proto.
16754 (avr_log_t) <insn_addresses>: New field.
16756 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16759 * config/i386/i386.c (ix86_function_arg_advance): Set
16760 outgoing_args_on_stack to true if there are outgoing arguments
16762 (ix86_function_arg): Likewise.
16763 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16764 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16765 * config/i386/i386.h (machine_function): Add
16766 outgoing_args_on_stack.
16768 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16770 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16771 supporting pthreds.
16772 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16774 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16776 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16777 (REAL_H): Remove $(MACHMODE_H).
16778 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16780 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16781 $(MACHMODE_H) and double-int.h.
16782 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16784 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16787 2017-07-07 Andrew Pinski <apinski@cavium.com>
16789 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16790 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16792 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16794 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16795 Add warning if GCC was not configured to link against a GLIBC that
16796 exports the hardware capability bits.
16797 (make_resolver_func): Make resolver function private and not a
16798 COMDAT function. Create the name with clone_function_name instead
16799 of make_unique_name.
16802 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16803 correct operand in doing the split.
16805 2017-07-07 Carl Love <cel@us.ibm.com>
16807 * config/rs6000/rs6000-c: Add support for built-in function
16808 vector unsigned short vec_pack_to_short_fp32 (vector float,
16810 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16811 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16812 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16813 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16814 (convert_4f32_8i16): Add define_expand.
16815 * doc/extend.texi: Update the built-in documentation file for the
16816 new built-in function.
16818 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16820 * config/sparc/m8.md: New file.
16821 * config/sparc/sparc.md: Include m8.md.
16823 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16825 * config/sparc/sparc.opt: New option -mvis4b.
16826 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16827 (sparc_option_override): Handle VIS4B.
16828 (enum sparc_builtins): Define
16829 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16830 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16831 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16832 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16833 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16834 (check_constant_argument): New function.
16835 (sparc_vis_init_builtins): Define builtins
16836 __builtin_vis_dictunpack{8,16,32},
16837 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16838 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16839 __builtin_vis_fpcmpde{8,16,32}shl and
16840 __builtin_vis_fpcmpur{8,16,32}shl.
16841 (sparc_expand_builtin): Check that the constant operands to
16842 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16843 constant and in range.
16844 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16846 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16847 (SPARC_IMM5_P): Likewise.
16848 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16849 (enabled): Handle vis4b.
16850 (UNSPEC_DICTUNPACK): New unspec.
16851 (UNSPEC_FPCMPSHL): Likewise.
16852 (UNSPEC_FPUCMPSHL): Likewise.
16853 (UNSPEC_FPCMPDESHL): Likewise.
16854 (UNSPEC_FPCMPURSHL): Likewise.
16855 (cpu_feature): New CPU feature `vis4b'.
16856 (dictunpack{8,16,32}): New insns.
16857 (FPCSMODE): New mode iterator.
16858 (fpcscond): New code iterator.
16859 (fpcsucond): Likewise.
16860 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16861 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16862 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16863 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16864 * config/sparc/constraints.md: Define constraints `q' for unsigned
16865 2-bit integer constants and `t' for unsigned 5-bit integer
16867 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16869 (imm5_operand_dictunpack16): Likewise.
16870 (imm5_operand_dictunpack32): Likewise.
16871 (imm2_operand): Likewise.
16872 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16873 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16874 ditunpack* and fpcmp*shl builtins.
16876 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16878 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16879 * config.in: Add HAVE_AS_SPARC6 define.
16880 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16882 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16884 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16885 (CPP_CPU_SPEC): Handle m8.
16886 (ASM_CPU_SPEC): Likewise.
16887 * config/sparc/sparc-opts.h (enum processor_type): Add
16889 * config/sparc/sparc.c (m8_costs): New struct.
16890 (sparc_option_override): Handle TARGET_CPU_m8.
16891 (sparc32_initialize_trampoline): Likewise.
16892 (sparc64_initialize_trampoline): Likewise.
16893 (sparc_issue_rate): Likewise.
16894 (sparc_register_move_cost): Likewise.
16895 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16896 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16897 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16898 (CPP_CPU_SPEC): Handle M8.
16899 (ASM_CPU_SPEC): Likewise.
16900 (AS_M8_FLAG): Define.
16901 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16902 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16903 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16905 * configure: Regenerate.
16906 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16909 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16911 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16913 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16914 ("*movdi_insn_sp32"): Do not set v3pipe.
16915 ("*movsi_insn"): Likewise.
16916 ("*movdi_insn_sp64"): Likewise.
16917 ("*movsf_insn"): Likewise.
16918 ("*movdf_insn_sp32"): Likewise.
16919 ("*movdf_insn_sp64"): Likewise.
16920 ("*zero_extendsidi2_insn_sp64"): Likewise.
16921 ("*sign_extendsidi2_insn"): Likewise.
16922 ("*mov<VM32:mode>_insn"): Likewise.
16923 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16924 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16925 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16926 ("<vlop:code><VL:mode>3"): Likewise.
16927 ("*not_<vlop:code><VL:mode>3"): Likewise.
16928 ("*nand<VL:mode>_vis"): Likewise.
16929 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16930 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16931 ("one_cmpl<VL:mode>2"): Likewise.
16932 ("faligndata<VM64:mode>_vis"): Likewise.
16933 ("alignaddrsi_vis"): Likewise.
16934 ("alignaddrdi_vis"): Likweise.
16935 ("alignaddrlsi_vis"): Likewise.
16936 ("alignaddrldi_vis"): Likewise.
16937 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16938 ("bmaskdi_vis"): Likewise.
16939 ("bmasksi_vis"): Likewise.
16940 ("bshuffle<VM64:mode>_vis"): Likewise.
16941 ("cmask8<P:mode>_vis"): Likewise.
16942 ("cmask16<P:mode>_vis"): Likewise.
16943 ("cmask32<P:mode>_vis"): Likewise.
16944 ("pdistn<P:mode>_vis"): Likewise.
16945 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16947 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16949 * config/sparc/sparc.md ("subtype"): New insn attribute.
16950 ("*wrgsr_sp64"): Set insn subtype.
16951 ("*rdgsr_sp64"): Likewise.
16952 ("alignaddrsi_vis"): Likewise.
16953 ("alignaddrdi_vis"): Likewise.
16954 ("alignaddrlsi_vis"): Likewise.
16955 ("alignaddrldi_vis"): Likewise.
16956 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16957 ("fexpand_vis"): Likewise.
16958 ("fpmerge_vis"): Likewise.
16959 ("faligndata<VM64:mode>_vis"): Likewise.
16960 ("bshuffle<VM64:mode>_vis"): Likewise.
16961 ("cmask8<P:mode>_vis"): Likewise.
16962 ("cmask16<P:mode>_vis"): Likewise.
16963 ("cmask32<P:mode>_vis"): Likewise.
16964 ("fchksm16_vis"): Likewise.
16965 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16966 ("fmean16_vis"): Likewise.
16967 ("fp<plusminus_insn>64_vis"): Likewise.
16968 ("<plusminus_insn>v8qi3"): Likewise.
16969 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16970 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16971 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16972 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16973 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16974 ("*movqi_insn"): Likewise.
16975 ("*movhi_insn"): Likewise.
16976 ("*movsi_insn"): Likewise.
16977 ("movsi_pic_gotdata_op"): Likewise.
16978 ("*movdi_insn_sp32"): Likewise.
16979 ("*movdi_insn_sp64"): Likewise.
16980 ("movdi_pic_gotdata_op"): Likewise.
16981 ("*movsf_insn"): Likewise.
16982 ("*movdf_insn_sp32"): Likewise.
16983 ("*movdf_insn_sp64"): Likewise.
16984 ("*zero_extendhisi2_insn"): Likewise.
16985 ("*zero_extendqihi2_insn"): Likewise.
16986 ("*zero_extendqisi2_insn"): Likewise.
16987 ("*zero_extendqidi2_insn"): Likewise.
16988 ("*zero_extendhidi2_insn"): Likewise.
16989 ("*zero_extendsidi2_insn_sp64"): Likewise.
16990 ("ldfsr"): Likewise.
16991 ("prefetch_64"): Likewise.
16992 ("prefetch_32"): Likewise.
16993 ("tie_ld32"): Likewise.
16994 ("tie_ld64"): Likewise.
16995 ("*tldo_ldub_sp32"): Likewise.
16996 ("*tldo_ldub1_sp32"): Likewise.
16997 ("*tldo_ldub2_sp32"): Likewise.
16998 ("*tldo_ldub_sp64"): Likewise.
16999 ("*tldo_ldub1_sp64"): Likewise.
17000 ("*tldo_ldub2_sp64"): Likewise.
17001 ("*tldo_ldub3_sp64"): Likewise.
17002 ("*tldo_lduh_sp32"): Likewise.
17003 ("*tldo_lduh1_sp32"): Likewise.
17004 ("*tldo_lduh_sp64"): Likewise.
17005 ("*tldo_lduh1_sp64"): Likewise.
17006 ("*tldo_lduh2_sp64"): Likewise.
17007 ("*tldo_lduw_sp32"): Likewise.
17008 ("*tldo_lduw_sp64"): Likewise.
17009 ("*tldo_lduw1_sp64"): Likewise.
17010 ("*tldo_ldx_sp64"): Likewise.
17011 ("*mov<VM32:mode>_insn"): Likewise.
17012 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17013 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17015 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17017 * config/sparc/sparc.md ("type"): New insn type viscmp.
17018 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17020 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17021 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17022 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17023 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17025 ("n7_vis_logical_11cycle"): Likewise.
17026 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17027 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17028 * config/sparc/niagara.md ("niag_vis"): Likewise.
17029 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17030 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17032 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17034 * config/sparc/sparc.md: New instruction type `bmask'.
17035 (bmaskdi_vis): Use the `bmask' type.
17036 (bmasksi_vis): Likewise.
17037 * config/sparc/ultra3.md (us3_array): Likewise.
17038 * config/sparc/niagara7.md (n7_array): Likewise.
17039 * config/sparc/niagara4.md (n4_array): Likewise.
17040 * config/sparc/niagara2.md (niag2_vis): Likewise.
17041 (niag3_vis): Likewise.
17042 * config/sparc/niagara.md (niag_vis): Likewise.
17044 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17046 * ipa-comdats.c: Remove optimize check from gate.
17047 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17048 for functions not optimized.
17049 (ipa_fn_summary_read): Skip optimize check.
17050 (ipa_fn_summary_write): Likewise.
17051 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17053 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17055 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17056 (check_callers): Give up on caller which is not optimized.
17057 (inline_small_functions): Likewise.
17058 (ipa_inline): Do not give up when not optimizing.
17059 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17060 away unoptimizes cdtors.
17061 (whole_program_function_and_variable_visibility): Do
17062 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17063 * ipa.c (process_references): Do not check optimize.
17064 (symbol_table::remove_unreachable_nodes): Update optimize check.
17065 (set_writeonly_bit): Update optimize check.
17066 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17067 (pass_ipa_single_use::gate): Remove.
17069 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17071 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17072 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17073 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17074 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17075 permute_load, permute_store, adjust_extract, adjust_splat,
17076 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17077 replace_swap_with_copy, dump_swap_insn_table,
17078 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17079 recombine_lvx_pattern, recombine_stvx_pattern,
17080 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17081 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17082 to file rs6000-p8swap.c.
17083 * config/rs6000/rs6000-p8swap.c: New file.
17084 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17085 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17086 and rs6000*-*-* targets.
17088 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17090 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17092 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17094 * lto-wrapper.c (merge_and_complain): Do not merge
17095 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17096 fsigned_zeros, ftrapping_math, fwrapv.
17097 (append_compiler_options): Do not track these options.
17098 (append_linker_options): Likewie
17100 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17102 * cgraphunit.c (cgraph_node::finalize_function): When
17103 !flag_toplevel_reorde set no_reorder flag.
17104 (varpool_node::finalize_decl): Likewise.
17105 (symbol_table::compile): Drop no toplevel reorder path.
17107 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17109 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17110 edges; zero probability is not better than uninitialized.
17112 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17114 * asan.h (asan_sanitize_allocas_p): Declare.
17115 * asan.c (asan_sanitize_allocas_p): New function.
17116 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17117 (handle_builtin_alloca): Likewise.
17118 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17119 if !asan_sanitize_allocas_p.
17120 * params.def (asan-instrument-allocas): Add new option.
17121 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17122 * opts.c (common_handle_option): Disable allocas sanitization for
17125 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17127 * asan.c: Include gimple-fold.h.
17128 (get_last_alloca_addr): New function.
17129 (handle_builtin_stackrestore): Likewise.
17130 (handle_builtin_alloca): Likewise.
17131 (asan_emit_allocas_unpoison): Likewise.
17132 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17133 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17134 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17135 (instrument_builtin_call): Pass gimple iterator to
17136 get_mem_refs_of_builtin_call.
17137 (last_alloca_addr): New global.
17138 * asan.h (asan_emit_allocas_unpoison): Declare.
17139 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17140 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17141 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17142 if function calls alloca.
17143 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17144 * gimple-fold.h (replace_call_with_value): Declare.
17145 * internal-fn.c: Include asan.h.
17146 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17147 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17149 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17151 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17152 (C_SELFTEST_FLAGS): New.
17153 (CPP_SELFTEST_FLAGS): New.
17154 (SELFTEST_DEPS): New, from deps of s-selftest.
17155 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17156 (CPP_SELFTEST_DEPS): New.
17157 (selftest): Add dependency on s-selftest-c++.
17158 (s-selftest): Rename to...
17159 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17160 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17161 than SELFTEST_FLAGS.
17162 (selftest-gdb): Rename to...
17163 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17165 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17166 (selftest-valgrind): Rename to...
17167 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17169 (selftest-valgrind): Reintroduce as an alias for
17170 selftest-c-valgrind.
17171 (s-selftest-c++): New.
17172 (selftest-c++-gdb): New.
17173 (selftest-c++-valgrind): New.
17175 2017-07-06 Olivier Hainque <hainque@adacore.com>
17177 * gcc.c (process_command): When deciding if undefined variables
17178 should be ignored when processing specs, accept "gcc -v" as well.
17180 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17182 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17183 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17185 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17187 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17188 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17189 * config/arm/arm-tables.opt: Regenerate.
17190 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17192 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17194 2017-07-06 Carl Love <cel@us.ibm.com>
17196 * ChangeLog: Clean up from mid air collision
17198 2017-07-06 Carl Love <cel@us.ibm.com>
17200 * config/rs6000/rs6000-c.c: Add support for built-in functions
17201 vector signed int vec_subc (vector signed int, vector signed int);
17202 vector signed __int128 vec_subc (vector signed __int128,
17203 vector signed __int128);
17204 vector unsigned __int128 vec_subc (vector unsigned __int128,
17205 vector unsigned __int128);
17206 vector signed int vec_sube (vector signed int, vector signed int,
17207 vector signed int);
17208 vector unsigned int vec_sube (vector unsigned int,
17209 vector unsigned int,
17210 vector unsigned int);
17211 vector signed __int128 vec_sube (vector signed __int128,
17212 vector signed __int128,
17213 vector signed__int128);
17214 vector unsigned __int128 vec_sube (vector unsigned __int128,
17215 vector unsigned __int128,
17216 vector unsigned __int128);
17217 vector signed int vec_subec (vector signed int, vector signed int,
17218 vector signed int);
17219 vector unsigned int vec_subec (vector unsigned int,
17220 vector unsigned int,
17221 vector unsigned int);
17222 vector signed __int128 vec_subec (vector signed __int128,
17223 vector signed __int128,
17224 vector signed__int128);
17225 vector unsigned __int128 vec_subec (vector unsigned __int128,
17226 vector unsigned __int128,
17227 vector unsigned __int128);
17228 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17229 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17230 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17231 BU_ALTIVEC_OVERLOAD_X definitions.
17232 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17233 * doc/extend.texi: Update the built-in documentation file for the new
17234 built-in functions.
17236 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17239 * diagnostic-show-locus.c (layout::layout): Use start and finish
17240 spelling location for the start and finish of each range.
17241 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17242 Add unused aspect param.
17243 * input.c (expand_location_1): Add "aspect" param, and use it
17244 to access the correct part of the location.
17245 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17247 (expand_location_to_spelling_point): Likewise.
17248 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17249 param, and pass it to expand_location_1.
17251 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17253 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17254 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17255 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17256 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17257 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17258 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17259 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17260 _mm_maskz_getmant_ss): New intrinsics.
17261 (__builtin_ia32_getexpss128_mask): Changed to ...
17262 __builtin_ia32_getexpss128_round ... this.
17263 (__builtin_ia32_getexpsd128_mask): Changed to ...
17264 __builtin_ia32_getexpsd128_round ... this.
17265 * config/i386/i386-builtin-types.def
17266 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17267 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17268 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17269 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17270 __builtin_ia32_getmantss_mask_round): New builtins.
17271 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17272 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17273 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17274 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17275 * config/i386/sse.md
17276 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17277 avx512f_sgetexp<mode><mask_scalar_name>
17278 <round_saeonly_scalar_name> ... this.
17279 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17280 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17281 vgetexp<ssescalarmodesuffix>
17282 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17283 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17284 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17285 avx512f_vgetmant<mode><mask_scalar_name>
17286 <round_saeonly_scalar_name> ... this.
17287 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17288 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17289 vgetmant<ssescalarmodesuffix>
17290 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17291 %0<mask_scalar_operand4>, %1, %2
17292 <round_saeonly_scalar_mask_op4>, %3} ... this.
17293 * config/i386/subst.md (mask_scalar_operand4,
17294 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17295 round_saeonly_scalar_nimm_predicate): New subst attributes.
17297 2017-07-06 Julia Koval <julia.koval@intel.com>
17299 * config/i386/i386.c (ix86_erase_embedded_rounding):
17300 Remove code for old rounding pattern.
17302 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17304 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17306 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17308 * doc/sourcebuild.texi (Test Directives, Variants of
17309 dg-require-support): Add documentation for dg-require-stack-check.
17311 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17313 * config/i386/subst.md (mask_scalar, round_scalar,
17314 round_saeonly_scalar): New meta-templates.
17315 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17316 round_scalar_mask_operand3, round_scalar_mask_op3,
17317 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17318 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17319 round_saeonly_scalar_constraint,
17320 round_saeonly_scalar_prefix): New subst attribute.
17321 * config/i386/sse.md
17322 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17323 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17324 <round_scalar_name> ... this.
17325 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17326 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17327 <round_scalar_name> ... this.
17328 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17329 <sse>_vm<code><mode>3<mask_scalar_name>
17330 <round_saeonly_scalar_name> ... this.
17331 (v<plusminus_mnemonic><ssescalarmodesuffix>
17332 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17333 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17334 v<plusminus_mnemonic><ssescalarmodesuffix>
17335 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17336 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17337 (v<multdiv_mnemonic><ssescalarmodesuffix>
17338 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17339 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17340 v<multdiv_mnemonic><ssescalarmodesuffix>
17341 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17342 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17343 (v<maxmin_float><ssescalarmodesuffix>
17344 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17345 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17346 v<maxmin_float><ssescalarmodesuffix>
17347 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17348 %0<mask_scalar_operand3>, %1, %<iptr>2
17349 <round_saeonly_scalar_mask_op3>} ... this.
17351 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17353 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17354 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17356 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17357 Alan Hayward <alan.hayward@arm.com>
17358 David Sherwood <david.sherwood@arm.com>
17360 * combine.c (simplify_if_then_else): Remove "enum" before
17362 * compare-elim.c (can_eliminate_compare): Likewise.
17363 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17365 (aarch64_lookup_simd_builtin_type): Likewise.
17366 (aarch64_simd_builtin_type): Likewise.
17367 (aarch64_init_simd_builtin_types): Likewise.
17368 (aarch64_simd_expand_args): Likewise.
17369 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17371 (aarch64_reverse_mask): Likewise.
17372 (aarch64_simd_emit_reg_reg_move): Likewise.
17373 (aarch64_gen_adjusted_ldpstp): Likewise.
17374 (aarch64_ccmp_mode_to_code): Likewise.
17375 (aarch64_operands_ok_for_ldpstp): Likewise.
17376 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17377 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17379 (aarch64_min_divisions_for_recip_mul): Likewise.
17380 (aarch64_reassociation_width): Likewise.
17381 (aarch64_get_condition_code_1): Likewise.
17382 (aarch64_simd_emit_reg_reg_move): Likewise.
17383 (aarch64_simd_attr_length_rglist): Likewise.
17384 (aarch64_reverse_mask): Likewise.
17385 (aarch64_operands_ok_for_ldpstp): Likewise.
17386 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17387 (aarch64_gen_adjusted_ldpstp): Likewise.
17388 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17390 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17391 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17392 (arm_lookup_simd_builtin_type): Likewise.
17393 (arm_simd_builtin_type): Likewise.
17394 (arm_init_simd_builtin_types): Likewise.
17395 (arm_expand_builtin_args): Likewise.
17396 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17397 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17398 (ft32_setup_incoming_varargs): Likewise.
17399 (ft32_function_arg): Likewise.
17400 (ft32_function_arg_advance): Likewise.
17401 (ft32_pass_by_reference): Likewise.
17402 (ft32_arg_partial_bytes): Likewise.
17403 (ft32_valid_pointer_mode): Likewise.
17404 (ft32_addr_space_pointer_mode): Likewise.
17405 (ft32_addr_space_legitimate_address_p): Likewise.
17406 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17408 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17409 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17410 (iamcu_alignment): Likewise.
17411 (canonicalize_vector_int_perm): Likewise.
17412 (ix86_noce_conversion_profitable_p): Likewise.
17413 (ix86_mpx_bound_mode): Likewise.
17414 (ix86_operands_ok_for_move_multiple): Likewise.
17415 * config/microblaze/microblaze-protos.h
17416 (microblaze_expand_conditional_branch_reg): Likewise.
17417 * config/microblaze/microblaze.c
17418 (microblaze_expand_conditional_branch_reg): Likewise.
17419 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17421 (rs6000_reassociation_width): Likewise.
17422 (rs6000_invalid_binary_op): Likewise.
17423 (fusion_p9_p): Likewise.
17424 (emit_fusion_p9_load): Likewise.
17425 (emit_fusion_p9_store): Likewise.
17426 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17428 (riscv_hard_regno_mode_ok_p): Likewise.
17429 (riscv_address_insns): Likewise.
17430 (riscv_split_symbol): Likewise.
17431 (riscv_legitimize_move): Likewise.
17432 (riscv_function_value): Likewise.
17433 (riscv_hard_regno_nregs): Likewise.
17434 (riscv_expand_builtin): Likewise.
17435 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17436 (riscv_build_integer): Likewise.
17437 (riscv_split_integer): Likewise.
17438 (riscv_legitimate_constant_p): Likewise.
17439 (riscv_cannot_force_const_mem): Likewise.
17440 (riscv_regno_mode_ok_for_base_p): Likewise.
17441 (riscv_valid_base_register_p): Likewise.
17442 (riscv_valid_offset_p): Likewise.
17443 (riscv_valid_lo_sum_p): Likewise.
17444 (riscv_classify_address): Likewise.
17445 (riscv_legitimate_address_p): Likewise.
17446 (riscv_address_insns): Likewise.
17447 (riscv_load_store_insns): Likewise.
17448 (riscv_force_binary): Likewise.
17449 (riscv_split_symbol): Likewise.
17450 (riscv_force_address): Likewise.
17451 (riscv_legitimize_address): Likewise.
17452 (riscv_move_integer): Likewise.
17453 (riscv_legitimize_const_move): Likewise.
17454 (riscv_legitimize_move): Likewise.
17455 (riscv_address_cost): Likewise.
17456 (riscv_subword): Likewise.
17457 (riscv_output_move): Likewise.
17458 (riscv_canonicalize_int_order_test): Likewise.
17459 (riscv_emit_int_order_test): Likewise.
17460 (riscv_function_arg_boundary): Likewise.
17461 (riscv_pass_mode_in_fpr_p): Likewise.
17462 (riscv_pass_fpr_single): Likewise.
17463 (riscv_pass_fpr_pair): Likewise.
17464 (riscv_get_arg_info): Likewise.
17465 (riscv_function_arg): Likewise.
17466 (riscv_function_arg_advance): Likewise.
17467 (riscv_arg_partial_bytes): Likewise.
17468 (riscv_function_value): Likewise.
17469 (riscv_pass_by_reference): Likewise.
17470 (riscv_setup_incoming_varargs): Likewise.
17471 (riscv_print_operand): Likewise.
17472 (riscv_elf_select_rtx_section): Likewise.
17473 (riscv_save_restore_reg): Likewise.
17474 (riscv_for_each_saved_reg): Likewise.
17475 (riscv_register_move_cost): Likewise.
17476 (riscv_hard_regno_mode_ok_p): Likewise.
17477 (riscv_hard_regno_nregs): Likewise.
17478 (riscv_class_max_nregs): Likewise.
17479 (riscv_memory_move_cost): Likewise.
17480 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17481 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17482 (rl78_addr_space_address_mode): Likewise.
17483 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17485 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17486 (rs6000_reassociation_width): Likewise.
17487 (rs6000_invalid_binary_op): Likewise.
17488 (fusion_p9_p): Likewise.
17489 (emit_fusion_p9_load): Likewise.
17490 (emit_fusion_p9_store): Likewise.
17491 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17492 (ok_for_simple_move_operands): Likewise.
17493 (ok_for_simple_move_strict_operands): Likewise.
17494 (ok_for_simple_arith_logic_operands): Likewise.
17495 (visium_legitimize_reload_address): Likewise.
17496 (visium_select_cc_mode): Likewise.
17497 (output_cbranch): Likewise.
17498 (visium_split_double_move): Likewise.
17499 (visium_expand_copysign): Likewise.
17500 (visium_expand_int_cstore): Likewise.
17501 (visium_expand_fp_cstore): Likewise.
17502 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17503 (visium_function_arg): Likewise.
17504 (visium_function_arg_advance): Likewise.
17505 (visium_libcall_value): Likewise.
17506 (visium_setup_incoming_varargs): Likewise.
17507 (visium_legitimate_constant_p): Likewise.
17508 (visium_legitimate_address_p): Likewise.
17509 (visium_legitimize_address): Likewise.
17510 (visium_secondary_reload): Likewise.
17511 (visium_register_move_cost): Likewise.
17512 (visium_memory_move_cost): Likewise.
17513 (prepare_move_operands): Likewise.
17514 (ok_for_simple_move_operands): Likewise.
17515 (ok_for_simple_move_strict_operands): Likewise.
17516 (ok_for_simple_arith_logic_operands): Likewise.
17517 (visium_function_value_1): Likewise.
17518 (rtx_ok_for_offset_p): Likewise.
17519 (visium_legitimize_reload_address): Likewise.
17520 (visium_split_double_move): Likewise.
17521 (visium_expand_copysign): Likewise.
17522 (visium_expand_int_cstore): Likewise.
17523 (visium_expand_fp_cstore): Likewise.
17524 (visium_split_cstore): Likewise.
17525 (visium_select_cc_mode): Likewise.
17526 (visium_split_cbranch): Likewise.
17527 (output_cbranch): Likewise.
17528 (visium_print_operand_address): Likewise.
17529 * expmed.c (flip_storage_order): Likewise.
17530 * expmed.h (emit_cstore): Likewise.
17531 (flip_storage_order): Likewise.
17532 * genrecog.c (validate_pattern): Likewise.
17533 * hsa-gen.c (gen_hsa_addr): Likewise.
17534 * internal-fn.c (expand_arith_overflow): Likewise.
17535 * ira-color.c (allocno_copy_cost_saving): Likewise.
17536 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17537 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17538 (process_invariant_for_inheritance): Likewise.
17539 * lra-eliminations.c (move_plus_up): Likewise.
17540 * omp-low.c (lower_oacc_reductions): Likewise.
17541 * simplify-rtx.c (simplify_subreg): Likewise.
17542 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17543 (TARGET_CHKP_BOUND_MODE): Likewise..
17544 * targhooks.c (default_chkp_bound_mode): Likewise.
17545 (default_setup_incoming_vararg_bounds): Likewise.
17546 * targhooks.h (default_chkp_bound_mode): Likewise.
17547 (default_setup_incoming_vararg_bounds): Likewise.
17548 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17549 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17550 (have_whole_vector_shift): Likewise.
17551 * tree-vect-stmts.c (vectorizable_load): Likewise.
17552 * doc/tm.texi: Regenerate.
17554 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17556 Graceful degrade if Binutils PR21472 is not available.
17559 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17560 .rodata in flash test fails.
17561 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17562 * confgure: Regenerate.
17563 * config.in: Regenerate.
17564 * config/avr/avr.c (avr_asm_named_section)
17565 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17566 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17567 (avr_asm_init_sections): Same.
17569 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17571 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17572 (fma<VH:mode>4_intrinsic): Likewise.
17573 (*fmsub<VCVTF:mode>4): Likewise.
17574 (*fmsub<VH:mode>4_intrinsic): Likewise.
17576 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17579 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17580 Don't depend on "optimize > 0".
17581 (out_movhi_r_mr, out_movqi_mr_r): Same.
17582 (out_movhi_mr_r, out_movqi_r_mr): Same.
17583 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17584 io_address_operand on "optimize > 0".
17586 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17588 * tree-loop-distribution.c: Add general explanantion on the pass.
17589 (generate_loops_for_partition): Mark distributed loop.
17590 (pg_add_dependence_edges): New parameter. Handle alias data
17591 dependence specially and record it in the parameter if asked.
17592 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17593 (init_partition_graph_vertices, add_partition_graph_edge): New.
17594 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17595 (free_partition_graph_vdata, build_partition_graph): New.
17596 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17597 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17598 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17599 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17600 (version_for_distribution_p, finalize_partitions): New.
17601 (distribute_loop): Handle alias data dependence specially. Factor
17602 out loop fusion code as functions and call these functions.
17604 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17606 * tree-loop-distribution.c (classify_partition): New parameter and
17607 better handle reduction statement.
17608 (rdg_build_partitions): Revise comment.
17609 (distribute_loop): Compute statements in all partitions and pass it
17610 to classify_partition.
17612 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17614 * tree-loop-distribution.c (enum partition_type): New.
17615 (struct partition): New field type.
17616 (partition_merge_into): Add parameter. Update partition type.
17617 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17618 (build_rdg_partition_for_vertex): Compute partition type.
17619 (rdg_build_partitions): Dump partition type.
17620 (distribute_loop): Update calls to partition_merge_into.
17622 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17624 * tree-loop-distribution.c (struct ddr_hasher): New.
17625 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17627 (classify_partition): Call get_data_dependence.
17628 (pg_add_dependence_edges): Ditto.
17629 (distribute_loop): Release data dependence hash table.
17631 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17633 * tree-loop-distribution.c (ref_base_address): Delete.
17634 (similar_memory_accesses): Rename ...
17635 (share_memory_accesses): ... to this. Check if partitions access
17636 the same memory reference.
17637 (distribute_loop): Call share_memory_accesses.
17639 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17641 * tree-loop-distribution.c (struct partition): New field recording
17642 its data reference.
17643 (partition_alloc, partition_free): Init and release data refs.
17644 (partition_merge_into): Merge data refs.
17645 (build_rdg_partition_for_vertex): Collect data refs for partition.
17646 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17648 (distribute_loop): Remve data refs from vertice data of partition
17651 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17653 * tree-loop-distribution.c (params.h): Include header file.
17654 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17655 (datarefs_vec): New global var.
17656 (create_rdg_vertices): Use datarefs_vec directly.
17657 (free_rdg): Don't free data references.
17658 (build_rdg): Update use. Don't free data references.
17659 (distribute_loop): Compute global variable for data references.
17660 Bail out if there are too many data references.
17662 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17664 * tree-loop-distribution.c (loop_nest): New global var.
17665 (build_rdg): Use loop directly, rather than loop nest.
17666 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17668 (distribute_loop): Compute global variable loop nest. Update use.
17670 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17672 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17673 (partition_merge_into): New parameter. Dump reason for fusion.
17674 (distribute_loop): Update use of partition_merge_into.
17676 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17678 * tree-loop-distribution.c (bb_top_order_index): New.
17679 (bb_top_order_index_size, bb_top_order_cmp): New.
17680 (stmts_from_loop): Use topological order.
17681 (pass_loop_distribution::execute): Compute and release topological
17682 order for basic blocks.
17684 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17686 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17689 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17691 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17692 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17693 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17694 * internal-fn.def (LOOP_DIST_ALIAS): New.
17695 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17696 (fold_loop_internal_call): ... this.
17697 (vect_loop_dist_alias_call): New function.
17698 (set_uid_loop_bbs): Call fold_loop_internal_call.
17699 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17702 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17705 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17706 Require dead FLAGS_REG at the beginning of a peephole.
17708 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17711 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17712 arguments in the call to __builtin_ia32_sbb_u32.
17713 (_subborrow_u64): Swap _X and _Y arguments in the call to
17714 __builtin_ia32_sbb_u64.
17716 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17719 * tree-vrp.c (compare_assert_loc): Turn into a function template
17720 with stable template parameter. Only test if a->e is NULL,
17721 !a->e == !b->e has been verified already. Use e == NULL or
17722 e != NULL instead of e or ! e tests. If stable is true, don't use
17723 iterative_hash_expr, on the other side allow a or b or both NULL
17724 and sort the NULLs last.
17725 (process_assert_insertions): Sort using compare_assert_loc<false>
17726 instead of compare_assert_loc, later sort using
17727 compare_assert_loc<true> before calling process_assert_insertions_for
17728 in a loop. Use break instead of continue once seen NULL pointer.
17730 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17732 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17733 Cortex-R7 and Cortex-R8 processors.
17735 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17737 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17738 uninitialized while src is not.
17740 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17742 * common/config/arm/arm-common.c: Adjust include path for
17744 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17745 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17746 (arm-cpu-data.h): Likewise.
17747 (arm-cpu-cdata.h): Likewise.
17748 * config/arm/arm-cpu.h: Delete.
17749 * config/arm/arm-cpu-cdata.h: Delete.
17750 * config/arm/arm-cpu-data.h: Delete.
17752 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17754 * config/arm/arm-cpus.in (cortex-a55): New.
17755 (cortex-a75): Likewise.
17756 (cortex-a75.cortex-a55): Likewise.
17757 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17759 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17760 * config/arm/arm-cpu-cdata.h: Regenerate.
17761 * config/arm/arm-cpu-data.h: Regenerate.
17762 * config/arm/arm-cpu.h: Regenerate.
17763 * config/arm/arm-tables.opt: Regenerate.
17764 * config/arm/arm-tune.md: Regenerate.
17766 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17768 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17770 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17772 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17775 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17777 PR tree-optimization/81292
17778 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17779 full_string_p, also call adjust_related_strinfos if the adjustment
17780 is simple, otherwise invalidate related strinfos.
17782 2017-07-04 Martin Liska <mliska@suse.cz>
17785 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17786 newly created variable as DECL_IGNORED_P.
17788 2017-07-04 Martin Liska <mliska@suse.cz>
17791 * ipa-inline.c (inline_small_functions):
17792 Use xstrdup_for_dump.
17794 2017-07-04 Tom de Vries <tom@codesourcery.com>
17796 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17798 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17801 * config/darwin.c (darwin_function_switched_text_sections):
17804 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17806 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17808 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17810 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17812 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17814 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17815 min_profitable_iters, and th as inclusive lower bounds.
17816 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17817 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17818 for min_profitable_iters and min_profitable_estimate.
17819 (vect_transform_loop): Treat th as an inclusive lower bound.
17820 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17822 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17825 * config/darwin.c (darwin_function_switched_text_sections):
17826 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17827 in two pieces, and suppress the use of buf.
17829 2017-07-03 Nathan Sidwell <nathan@acm.org>
17831 * hash-table.h (hash_table_mod1): Fix indentation.
17833 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17835 PR middle-end/81290
17836 * predict.c (force_edge_cold): Be more careful about propagation
17838 * profile-count.h (profile_probability::guessed,
17839 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17841 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17843 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17845 * doc/invoke.texi (rcpc architecture extension): Document it.
17847 2017-07-03 Richard Biener <rguenther@suse.de>
17849 PR tree-optimization/60510
17850 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17851 the scalar reduction PHI and use it.
17852 (vectorizable_reduction): Properly guard the single_defuse_cycle
17853 path for non-SLP reduction chains where we cannot use it.
17854 Rework reduc_def/index and vector type deduction. Rework
17855 vector operand gathering during reduction op code-gen.
17856 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17857 chains dissolve the chain and leave it to non-SLP reduction
17860 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17862 * tree-data-ref.h (dr_alignment): Declare.
17863 * tree-data-ref.c (dr_alignment): New function.
17864 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17865 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17867 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17869 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17871 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17872 and base_misalignment fields.
17873 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17874 * tree-data-ref.c: Include builtins.h.
17875 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17876 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17877 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17878 * tree-vect-data-refs.c: Include tree-cfg.h.
17879 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17880 fields instead of calculating an alignment here.
17881 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17882 innermost_loop_behavior fields.
17884 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17886 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17888 (DR_STEP_ALIGNMENT): New macro.
17889 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17890 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17891 (create_data_ref): Print it.
17892 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17893 to tell whether the step preserves vector (mis)alignment.
17894 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17895 Move the check for an integer step and generalise to all INTEGER_CST.
17896 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17897 Print the outer step alignment.
17899 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17901 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17902 with offset_alignment.
17903 (DR_ALIGNED_TO): Delete.
17904 (DR_OFFSET_ALIGNMENT): New macro.
17905 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17906 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17907 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17908 (create_data_ref): Likewise.
17909 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17910 (vect_analyze_data_refs): Likewise.
17911 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17912 creating dummy innermost behavior.
17914 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17916 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17917 with a "innermost_loop_behavior *" and refeence tree.
17918 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17919 (create_data_ref): Update call accordingly.
17920 * tree-predcom.c (find_looparound_phi): Likewise.
17922 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17924 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17925 fields with dr_wrt_vec_loop.
17926 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17927 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17928 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17929 (vect_dr_behavior): New function.
17930 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17931 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17932 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17933 track whether the step preserves the misalignment.
17934 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17935 Use vect_dr_behavior.
17936 (vect_setup_realignment): Update call accordingly.
17937 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17938 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17939 call to vect_create_addr_base_for_vector_ref.
17940 (vect_create_cond_for_align_checks): Likewise.
17941 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17942 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17943 (vect_recog_mask_conversion_pattern): Likewise.
17944 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17945 (new_stmt_vec_info): Remove redundant zeroing.
17947 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17949 * common/config/arm/arm-common.c (arm_be8_option): New function.
17950 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17951 (ISA_ARMv6): Add isa_bit_be8.
17952 * config/arm/arm.h (arm_be8_option): Add prototype.
17953 (BE8_SPEC_FUNCTION): New define.
17954 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17955 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17956 (mlittle-endian): Similarly.
17957 (mbe8, mbe32): New options.
17958 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17959 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17961 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17963 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17965 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17967 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17968 (cleanup_tree_cfg_bb): Use it.
17969 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17971 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17973 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17976 * loop-doloop.c (add_test): Update profile.
17978 2017-07-03 Martin Liska <mliska@suse.cz>
17981 * sanopt.c (rewrite_usage_of_param): New function.
17982 (sanitize_rewrite_addressable_params): Likewise.
17983 (pass_sanopt::execute): Call rewrite_usage_of_param.
17985 2017-07-03 Richard Biener <rguenther@suse.de>
17987 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17988 back to using VIEW_CONVERT_EXPR.
17990 2017-07-03 Martin Liska <mliska@suse.cz>
17993 * doc/extend.texi: Document when a resolver function is
17994 generated for target_clones.
17996 2017-07-03 Martin Liska <mliska@suse.cz>
17998 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17999 and shadow memory of auto variables which are subject of
18000 use-after-scope sanitization.
18001 (asan_expand_mark_ifn): Add do set only when is_poison.
18003 2016-07-03 Richard Biener <rguenther@suse.de>
18005 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18007 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18008 (vectorizable_reduction): Perform reduction PHI creation when
18009 visiting a reduction PHI and adjust and simplify code generation
18010 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18011 (vect_transform_loop): Visit reduction PHIs.
18012 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18013 defs into the SLP tree.
18014 (vect_build_slp_tree): Reduction defs terminate the recursion.
18015 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18017 (vect_get_vec_defs_for_stmt_copy): Export.
18018 (vect_get_vec_defs): Likewise.
18019 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18021 (vect_get_vec_defs_for_stmt_copy): Declare.
18022 (vect_get_vec_defs): Likewise.
18024 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18026 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18027 parameter with a "loop" parameter and use it instead of the
18028 loop containing DR_STMT. Don't check simple_iv when doing
18029 BB analysis. Describe the two analysis modes in the comment.
18031 2017-07-03 Tom de Vries <tom@codesourcery.com>
18033 PR tree-optimization/69468
18034 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18035 (find_same_succ_bb): Handle ignore_edge_flags.
18037 2017-07-03 Tom de Vries <tom@codesourcery.com>
18039 PR tree-optimization/81192
18040 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18042 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18044 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18046 2017-07-03 Tom de Vries <tom@codesourcery.com>
18048 PR tree-optimization/81192
18049 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18050 BB_SAME_SUCC (bb) == NULL.
18052 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18054 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18057 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18059 * dumpfile.c: Include profile-count.h
18060 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18062 (insert_cond_bb): Update profile.
18063 * tree-cfg.h (insert_cond_bb): Update prototype.
18064 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18065 * tree-dump.c: Do not include tree-cfg.
18067 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18069 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18071 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18073 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18076 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18078 * tree-complex.c (expand_complex_div_wide): update profile.
18080 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18081 Alan Hayward <alan.hayward@arm.com>
18082 David Sherwood <david.sherwood@arm.com>
18084 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18085 (CORETYPES_H): New define.
18086 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18087 (insn-modes-inline.h, s-modes-inline-h): New rules.
18088 (generated_files): Add insn-modes-inline.h.
18089 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18090 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18091 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18092 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18093 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18094 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18095 (build/gencodes.o, build/genconditions.o): Likewise.
18096 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18097 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18098 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18099 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18100 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18101 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18102 * coretypes.h: Include everything up to real.h for generators.
18103 Include insn-modes.h first. Include wide-int-print.h after
18104 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18105 * machmode.h: Don't include insn-modes.h here.
18106 * function-tests.c: Remove includes of signop.h, machmode.h,
18107 double-int.h and wide-int.h.
18109 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18111 * optc-save-gen.awk: Likewise.
18112 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18113 * godump.c: Remove include of wide-int-print.h.
18114 * pretty-print.h: Likewise.
18115 * wide-int-print.cc: Likewise.
18116 * wide-int.cc: Likewise.
18117 * hash-map-tests.c: Remove include of signop.h.
18118 * hash-set-tests.c: Likewise.
18119 * rtl-tests.c: Likewise.
18120 * mkconfig.sh: Remove include of machmode.h.
18121 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18123 (emit_insn_modes_inline_h): ...this new function. Emit the code
18124 into an insn-modes-inline.h header file, adding appropriate
18125 include guards and end comments.
18126 (emit_insn_modes_c_header): Remove include of machmode.h.
18127 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18129 (main): Handle -i flag and call emit_insn_modes_inline_h when
18132 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18134 * tree-ssa-strlen.c (strinfo): Rename the length field to
18135 nonzero_chars. Add a full_string_p field.
18136 (compare_nonzero_chars, zero_length_string_p): New functions.
18137 (get_addr_stridx): Add an offset_out parameter.
18138 Use compare_nonzero_chars.
18139 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18140 (new_strinfo): Update after above changes to strinfo.
18141 (set_endptr_and_length): Set full_string_p.
18142 (get_string_length): Update after above changes to strinfo.
18143 (unshare_strinfo): Update call to new_strinfo.
18144 (maybe_invalidate): Likewise.
18145 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18146 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18147 as a uhwi instead of an shwi. Update after above changes to
18148 strinfo and new_strinfo.
18149 (zero_length_string): Assert that chainsi contains full strings.
18150 Use zero_length_string_p. Update call to new_strinfo.
18151 (adjust_related_strinfos): Update after above changes to strinfo.
18152 Copy full_string_p from origsi.
18153 (adjust_last_stmt): Use zero_length_string_p.
18154 (handle_builtin_strlen): Update after above changes to strinfo and
18155 new_strinfo. Install the lhs as the string length if the previous
18156 entry didn't describe a full string.
18157 (handle_builtin_strchr): Update after above changes to strinfo
18159 (handle_builtin_strcpy): Likewise.
18160 (handle_builtin_strcat): Likewise.
18161 (handle_builtin_malloc): Likewise.
18162 (handle_pointer_plus): Likewise.
18163 (handle_builtin_memcpy): Likewise. Track nonzero characters
18164 that aren't necessarily followed by a nul terminator.
18165 (handle_char_store): Likewise.
18167 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18169 PR tree-optimization/80769
18170 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18171 for malloc and calloc. Document the new invariant that all related
18172 strinfos have delayed lengths or none do.
18173 (verify_related_strinfos): Move earlier in file.
18174 (set_endptr_and_length): New function, split out from...
18175 (get_string_length): ...here. Also set the lengths of related
18177 (zero_length_string): Assert that chainsi has known (rather than
18179 (adjust_related_strinfos): Likewise.
18181 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18183 PR tree-optimization/81136
18184 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18185 assert that two references with the same misalignment have the same
18186 compile-time misalignment if those compile-time misalignments
18189 2017-07-01 Andi Kleen <ak@linux.intel.com>
18191 * print-tree.c (print_node): Print all attributes.
18193 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18195 * cfg.c (scale_bbs_frequencies): New function.
18196 * cfg.h (scale_bbs_frequencies): Declare it.
18197 * cfgloopanal.c (single_likely_exit): Cleanup.
18198 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18200 (scale_loop_profile): Likewise.
18201 (loop_version): Likewise.
18202 (create_empty_loop_on_edge): Update.
18203 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18204 scale_loop_frequencies, scale_loop_profile, loopify,
18205 loop_version): Update prototypes.
18206 * modulo-sched.c (sms_schedule): Update.
18207 * predict.c (unlikely_executed_edge_p): Also check probability.
18208 (probably_never_executed_edge_p): Fix typo.
18209 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18210 * tree-parloops.c (gen_parallel_loop): Update.
18211 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18212 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18213 * tree-ssa-loop-split.c (split_loop): Update.
18214 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18215 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18216 (vect_loop_versioning): Update.
18217 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18219 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18221 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18223 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18225 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18226 to keep profile consistent.
18228 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18230 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18231 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18232 * profile-count.h (max_safe_multiplier): Make unsigned.
18233 (profile_count::guessed_zero): New.
18235 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18237 * bb-reorder.c (fix_up_crossing_landing_pad,
18238 fix_crossing_conditional_branches): Use make_single_succ_edge
18239 to keep profile consistent.
18241 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18243 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18246 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18249 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18250 the right scopes, make sure cond_jump isn't preserved between multiple
18251 iterations. Search for fallthru edge whenever there are 3+ edges and
18252 use find_fallthru_edge for it.
18254 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18256 Patch by Alexander Monakov <amonakov@ispras.ru>
18257 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18258 probabilities consistently.
18260 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18262 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18263 emit_cmp_and_jump_insns.
18265 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18268 * tree-inline.c (expand_call_inline): Combine profile statuses.
18270 2017-06-30 Andrew Pinski <apinski@cavium.com>
18272 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18273 fold_stmt returned true.
18275 2017-06-30 Nathan Sidwell <nathan@acm.org>
18277 * ggc.h (empty_string): Delete.
18278 * cfgexpand.c (expand_asm_stmt): Use plain "".
18279 * optabs.c (expand_asm_memory_barrier): Likewise.
18280 * stringpool.c (empty_string): Delete.
18281 (digit_vector, digit_string): Delete.
18282 (ggc_alloc_string): Use plain "", don't optimize single digit
18283 strings. Use ggc_alloc_atomic.
18285 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18287 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18288 comparison set and one other set, use the cost of the non-comparison
18291 2017-06-30 Nathan Sidwell <nathan@acm.org>
18293 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18296 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18298 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18299 loops. Remove now unneeded calls to gimple_switch_set_label() that
18300 just set removed labels to NULL_TREE.
18302 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18304 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18305 (set_range_info): ...here. Only call set_range_info_raw if domain
18307 (set_nonzero_bits): Call set_range_info_raw.
18308 * tree-ssanames.h (set_range_info_raw): New.
18310 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18313 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18314 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18315 of nonimmediate_operand and <store_mask_constraint> instead of m
18316 for the input operand. For V8FI iterator, always split if input
18317 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18318 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18319 <store_mask_predicate> instead of register_operand and
18320 <store_mask_constraint> instead of v for the input operand. Make
18321 sure both operands aren't MEMs for if not <mask_applied>.
18323 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18325 * lto-wrapper.c (copy_file) Close both file descriptors before
18328 2017-06-30 Martin Liska <mliska@suse.cz>
18331 * multiple_target.c (create_dispatcher_calls): Make ifunc
18332 also for function that don't have calls or are not referenced.
18334 2017-06-30 Richard Biener <rguenther@suse.de>
18336 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18337 analyze the first scalar stmt. Move vector type computation
18338 for the BB case here from ...
18339 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18340 live operation processing in the SLP case properly.
18342 2017-06-30 Richard Biener <rguenther@suse.de>
18344 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18346 2017-06-30 Martin Liska <mliska@suse.cz>
18349 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18350 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18352 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18354 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18357 2017-06-30 Richard Biener <rguenther@suse.de>
18359 PR tree-optimization/81249
18360 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18361 condition reduction result to original scalar type.
18363 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18365 * profile-count.h (enum profile_quality): Fix typos and whitespace
18368 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18370 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18371 type for branch probabilities.
18373 2017-06-29 Julian Brown <julian@codesourcery.com>
18374 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18376 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18377 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18378 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18379 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18381 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18383 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18384 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18385 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18386 CC usage from generic code to here.
18387 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18388 CC usage into the target macros.
18390 2017-06-29 Maya Rashish <coypu@sdf.org>
18392 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18395 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18397 * arm/arm-builtins.c: Include profile-count.h
18398 * except.c (sjlj_emit_function_enter): Use
18399 profile_probability::unlikely.
18401 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18403 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18404 and tocrel_offset be pointer args rather than implicitly using
18406 (legitimate_constant_pool_address_p, rs6000_emit_move,
18407 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18408 tocrel_offset and use in toc_relative_expr_p call.
18409 (print_operand, print_operand_address): Use static tocrel_base_oac
18410 and tocrel_offset_oac.
18411 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18414 2017-06-29 Maya Rashish <coypu@sdf.org>
18416 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18418 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18420 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18421 objects, take into account only the alignment of 'op0' and 'mode1' if
18424 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18426 * ccmp.c (ccmp_tree_comparison_p): New function.
18427 (ccmp_candidate_p): Update to use above function.
18428 (get_compare_parts): New function.
18429 (expand_ccmp_next): Update to use new functions.
18430 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18432 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18433 take mode as argument.
18434 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18435 * expr.c (expand_expr_real_1): Pass mode as argument.
18437 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18439 * combine.c (combine_instructions): Print insns to dump_file, together
18442 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18444 * asan.c (asan_emit_stack_protection): Update.
18445 (create_cond_insert_point): Update.
18446 * auto-profile.c (afdo_propagate_circuit): Update.
18447 * basic-block.h (struct edge_def): Turn probability to
18448 profile_probability.
18449 (EDGE_FREQUENCY): Update.
18450 * bb-reorder.c (find_traces_1_round): Update.
18451 (better_edge_p): Update.
18452 (sanitize_hot_paths): Update.
18453 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18454 (make_single_succ_edge): Update.
18455 (check_bb_profile): Update.
18456 (dump_edge_info): Update.
18457 (update_bb_profile_for_threading): Update.
18458 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18459 probabilitycount to 0.
18460 * cfgbuild.c (compute_outgoing_frequencies): Update.
18461 * cfgcleanup.c (try_forward_edges): Update.
18462 (outgoing_edges_match): Update.
18463 (try_crossjump_to_edge): Update.
18464 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18465 (expand_gimple_tailcall): Update.
18466 (construct_init_block): Use make_single_succ_edge.
18467 (construct_exit_block): Use make_single_succ_edge.
18468 * cfghooks.c (verify_flow_info): Update.
18469 (redirect_edge_succ_nodup): Update.
18470 (split_edge): Update.
18471 (account_profile_record): Update.
18472 * cfgloopanal.c (single_likely_exit): Update.
18473 * cfgloopmanip.c (scale_loop_profile): Update.
18474 (set_zero_probability): Remove.
18475 (duplicate_loop_to_header_edge): Update.
18476 * cfgloopmanip.h (loop_version): Update prototype.
18477 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18478 (force_nonfallthru_and_redirect): Update.
18479 (update_br_prob_note): Update.
18480 (rtl_verify_edges): Update.
18481 (purge_dead_edges): Update.
18482 (rtl_lv_add_condition_to_bb): Update.
18483 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18484 * cgraphunit.c (init_lowered_empty_function): Update.
18485 (cgraph_node::expand_thunk): Update.
18486 * cilk-common.c: Include profile-count.h
18487 * dojump.c (inv): Remove.
18488 (jumpifnot): Update.
18489 (jumpifnot_1): Update.
18490 (do_jump_1): Update.
18492 (do_jump_by_parts_greater_rtx): Update.
18493 (do_compare_rtx_and_jump): Update.
18494 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18495 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18496 * dwarf2cfi.c: Include profile-count.h
18497 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18498 (sjlj_emit_dispatch_table): Likewise.
18499 * explow.c: Include profile-count.h
18500 * expmed.c (emit_store_flag_force): Update.
18501 (do_cmp_and_jump): Update.
18502 * expr.c (compare_by_pieces_d::generate): Update.
18503 (compare_by_pieces_d::finish_mode): Update.
18504 (emit_block_move_via_loop): Update.
18505 (store_expr_with_bounds): Update.
18506 (store_constructor): Update.
18507 (expand_expr_real_2): Update.
18508 (expand_expr_real_1): Update.
18509 * expr.h (try_casesi, try_tablejump): Update prototypes.
18510 * gimple-pretty-print.c (dump_probability): Update.
18511 (dump_profile): New.
18512 (dump_gimple_label): Update.
18513 (dump_gimple_bb_header): Update.
18514 * graph.c (draw_cfg_node_succ_edges): Update.
18515 * hsa-gen.c (convert_switch_statements): Update.
18516 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18517 (find_if_case_1): Update.
18518 (find_if_case_2): Update.
18519 * internal-fn.c (expand_arith_overflow_result_store): Update.
18520 (expand_addsub_overflow): Update.
18521 (expand_neg_overflow): Update.
18522 (expand_mul_overflow): Update.
18523 (expand_vector_ubsan_overflow): Update.
18524 * ipa-cp.c (good_cloning_opportunity_p): Update.
18525 * ipa-split.c (split_function): Use make_single_succ_edge.
18526 * ipa-utils.c (ipa_merge_profiles): Update.
18527 * loop-doloop.c (add_test): Update.
18528 (doloop_modify): Update.
18529 * loop-unroll.c (compare_and_jump_seq): Update.
18530 (unroll_loop_runtime_iterations): Update.
18531 * lra-constraints.c (lra_inheritance): Update.
18532 * lto-streamer-in.c (input_cfg): Update.
18533 * lto-streamer-out.c (output_cfg): Update.
18534 * mcf.c (adjust_cfg_counts): Update.
18535 * modulo-sched.c (sms_schedule): Update.
18536 * omp-expand.c (expand_omp_for_init_counts): Update.
18537 (extract_omp_for_update_vars): Update.
18538 (expand_omp_ordered_sink): Update.
18539 (expand_omp_for_ordered_loops): Update.
18540 (expand_omp_for_generic): Update.
18541 (expand_omp_for_static_nochunk): Update.
18542 (expand_omp_for_static_chunk): Update.
18543 (expand_cilk_for): Update.
18544 (expand_omp_simd): Update.
18545 (expand_omp_taskloop_for_outer): Update.
18546 (expand_omp_taskloop_for_inner): Update.
18547 * omp-simd-clone.c (simd_clone_adjust): Update.
18548 * optabs.c (expand_doubleword_shift): Update.
18549 (expand_abs): Update.
18550 (emit_cmp_and_jump_insn_1): Update.
18551 (expand_compare_and_swap_loop): Update.
18552 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18553 * predict.c (predictable_edge_p): Update.
18554 (edge_probability_reliable_p): Update.
18555 (set_even_probabilities): Update.
18556 (combine_predictions_for_insn): Update.
18557 (combine_predictions_for_bb): Update.
18558 (propagate_freq): Update.
18559 (estimate_bb_frequencies): Update.
18560 (force_edge_cold): Update.
18561 * profile-count.c (profile_count::dump): Add missing space into dump.
18562 (profile_count::debug): Add newline.
18563 (profile_count::differs_from_p): Explicitly convert to unsigned.
18564 (profile_count::stream_in): Update.
18565 (profile_probability::dump): New member function.
18566 (profile_probability::debug): New member function.
18567 (profile_probability::differs_from_p): New member function.
18568 (profile_probability::differs_lot_from_p): New member function.
18569 (profile_probability::stream_in): New member function.
18570 (profile_probability::stream_out): New member function.
18571 * profile-count.h (profile_count_quality): Rename to ...
18572 (profile_quality): ... this one.
18573 (profile_probability): New.
18574 (profile_count): Update.
18575 * profile.c (compute_branch_probabilities): Update.
18576 * recog.c (peep2_attempt): Update.
18577 * sched-ebb.c (schedule_ebbs): Update.
18578 * sched-rgn.c (find_single_block_region): Update.
18579 (compute_dom_prob_ps): Update.
18580 (schedule_region): Update.
18581 * sel-sched-ir.c (compute_succs_info): Update.
18582 * stmt.c (struct case_node): Update.
18583 (do_jump_if_equal): Update.
18584 (get_outgoing_edge_probs): Update.
18585 (conditional_probability): Update.
18586 (emit_case_dispatch_table): Update.
18587 (expand_case): Update.
18588 (expand_sjlj_dispatch_table): Update.
18589 (emit_case_nodes): Update.
18590 * targhooks.c: Update.
18591 * tracer.c (better_p): Update.
18592 (find_best_successor): Update.
18593 * trans-mem.c (expand_transaction): Update.
18594 * tree-call-cdce.c: Update.
18595 * tree-cfg.c (gimple_split_edge): Upate.
18596 (move_sese_region_to_fn): Upate.
18597 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18598 * tree-eh.c (lower_resx): Upate.
18599 (cleanup_empty_eh_move_lp): Upate.
18600 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18601 * tree-inline.c (copy_edges_for_bb): Update.
18602 (copy_cfg_body): Update.
18603 * tree-parloops.c (gen_parallel_loop): Update.
18604 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18605 (gimple_gen_time_profiler): Update.
18606 * tree-ssa-dce.c (remove_dead_stmt): Update.
18607 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18608 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18609 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18610 (unloop_loops): Update.
18611 (try_peel_loop): Update.
18612 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18613 * tree-ssa-loop-split.c (connect_loops): Update.
18614 (split_loop): Update.
18615 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18616 (hoist_guard): Update.
18617 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18618 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18619 (value_replacement): Update.
18620 * tree-ssa-reassoc.c (branch_fixup): Update.
18621 * tree-ssa-tail-merge.c (replace_block_by): Update.
18622 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18623 (create_edge_and_update_destination_phis): Update.
18624 (compute_path_counts): Update.
18625 (recompute_probabilities): Update.
18626 (update_joiner_offpath_counts): Update.
18627 (freqs_to_counts_path): Update.
18628 (duplicate_thread_path): Update.
18629 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18630 (struct switch_conv_info): Update.
18631 (gen_inbound_check): Update.
18632 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18633 (vect_do_peeling): Update.
18634 (vect_loop_versioning): Update.
18635 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18636 (optimize_mask_stores): Update.
18637 * ubsan.c (ubsan_expand_null_ifn): Update.
18638 * value-prof.c (gimple_divmod_fixed_value): Update.
18639 (gimple_divmod_fixed_value_transform): Update.
18640 (gimple_mod_pow2): Update.
18641 (gimple_mod_pow2_value_transform): Update.
18642 (gimple_mod_subtract): Update.
18643 (gimple_mod_subtract_transform): Update.
18644 (gimple_ic): Update.
18645 (gimple_stringop_fixed_value): Update.
18646 (gimple_stringops_transform): Update.
18647 * value-prof.h: Update.
18649 2017-06-29 Carl Love <cel@us.ibm.com>
18651 * config/rs6000/rs6000-c.c: Add support for built-in functions
18652 vector signed int vec_signed (vector float);
18653 vector signed long long vec_signed (vector double);
18654 vector signed int vec_signed2 (vector double, vector double);
18655 vector signed int vec_signede (vector double);
18656 vector signed int vec_signedo (vector double);
18657 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18658 instruction generator.
18659 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18660 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18661 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18663 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18664 vunsignede_v2df): Add define_expands.
18665 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18666 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18667 VEC_UNSIGNEDO): Add definitions.
18668 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18669 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18670 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18671 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18672 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18673 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18674 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18675 vec_unsignede, vec_unsignedo): Add builtin defines.
18676 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18678 * doc/extend.texi: Update the built-in documentation file for the
18679 new built-in functions.
18681 2017-06-29 Richard Biener <rguenther@suse.de>
18683 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18684 reduction chains to LOOP_VINFO_REDUCTIONS.
18685 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18686 SLP reductions after processing reduction chains.
18688 2017-06-29 Nathan Sidwell <nathan@acm.org>
18690 * builtins.c (fold_builtin_FUNCTION): Use
18691 lang_hooks.decl_printable_name.
18693 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18695 PR middle-end/81194
18696 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18697 with only one label.
18698 * stmt.c (expand_case): Assert NCASES is greater than one.
18700 2017-06-29 Richard Biener <rguenther@suse.de>
18702 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18704 (group_case_labels): Likewise.
18705 (find_taken_edge): Push sanity checking on val to workers...
18706 (find_taken_edge_cond_expr): ... here
18707 (find_taken_edge_switch_expr): ... and here, handle cases
18708 with just a default label.
18709 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18710 (group_case_labels): Likewise.
18711 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18712 group_case_labels does anything cleanup the CFG again.
18714 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18716 PR tree-optimization/81196
18717 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18718 exit condition comparing two IVs.
18720 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18722 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18723 profile to the dummy entry at the end of the list of architectures.
18724 * config/arm/arm-cpu-cdata.h: Regenerated.
18726 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18727 Michael Collison <michael.collison@arm.com>
18730 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18732 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18733 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18734 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18735 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18736 of shift when the shift amount is masked with constant equal to
18737 the size of the mode.
18738 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18741 2017-06-29 Martin Liska <mliska@suse.cz>
18743 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18745 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18748 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18751 (edge_growth_cache_entry::edge_growth_cache_entry): New
18753 (reset_edge_growth_cache): Update to use constructor.
18755 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18757 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18758 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18759 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18761 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18763 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18764 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18766 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18768 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18769 (*-linux-uclibc*): Add t-uclibc tmake_file.
18770 * config/t-musl: New.
18771 * config/t-uclibc: New.
18773 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18775 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18777 (gen_comm_data): Emit architectural setting of arch_prof.
18778 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18780 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18781 (armv8-m.base, armv8-m.main): Likewise.
18782 * arm-protos.h (arm_build_target): Add profile field.
18783 (arch_option): Likewise.
18784 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18786 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18787 arm_active_target.profile.
18789 2017-06-28 Richard Biener <rguenther@suse.de>
18791 PR middle-end/81227
18792 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18793 TYPE_OVERFLOW_WRAPS.
18794 * match.pd (negate_expr_p): Likewise.
18795 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18796 fold_build2, not fold_binary.
18798 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18800 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18801 Convert memory address to Pmode.
18802 (aarch64_print_operand): Assert MEM operands are always Pmode.
18804 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18807 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18808 Remove redundant if.
18809 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18810 * config/arm/aarch-common-protos.h
18811 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18812 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18814 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18817 * multiple_target.c (create_dispatcher_calls): Set the default
18818 clone to be static, not public.
18820 2017-06-28 Richard Biener <rguenther@suse.de>
18822 * tree-vect-loop.c (vectorizable_reduction): Move special
18823 cond reduction IV var creation ...
18824 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18825 parameter. Use STMT_VINFO_VECTYPE.
18826 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18829 2017-06-28 Martin Liska <mliska@suse.cz>
18832 * ipa-visibility.c (non_local_p): Handle visibility.
18834 2017-06-28 Martin Liska <mliska@suse.cz>
18837 * common.opt: Add IntegerRange to various options.
18838 * opt-functions.awk (integer_range_info): New function.
18839 * optc-gen.awk: Add integer_range_info to cl_options struct.
18840 * opts-common.c (decode_cmdline_option): Handle
18841 CL_ERR_INT_RANGE_ARG.
18842 (cmdline_handle_error): Likewise.
18843 * opts.c (print_filtered_help): Show valid interval in
18844 when --help is provided.
18845 * opts.h (struct cl_option): Add range_min and range_max fields.
18846 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18848 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18850 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18851 (x * C EQ/NE y * C): New transformation.
18853 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18855 * genmultilib (combination_space): Accept '+' in option names.
18857 2017-06-28 Martin Liska <mliska@suse.cz>
18860 * asan.c (instrument_derefs): Bail out inner references
18861 that are hard register variables.
18863 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18866 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18867 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18869 2017-06-28 Richard Biener <rguenther@suse.de>
18871 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18872 (vect_get_slp_defs): Adjust.
18873 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18875 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18877 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18878 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18879 (vectorizable_reduction): Adjust.
18880 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18882 (vect_get_slp_defs): Likewise.
18883 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18884 (vectorizable_bswap): Adjust.
18885 (vectorizable_call): Likewise.
18886 (vectorizable_conversion): Likewise.
18887 (vectorizable_assignment): Likewise.
18888 (vectorizable_shift): Likewise.
18889 (vectorizable_operation): Likewise.
18890 (vectorizable_store): Likewise.
18891 (vectorizable_condition): Likewise.
18892 (vectorizable_comparison): Likewise.
18894 2017-06-28 Michael Collison <michael.collison@arm.com>
18897 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18899 (arm_gen_movmemqi): Removed unused variable 'i'.
18900 Convert 'for' loop into 'while' loop.
18901 (arm_expand_prologue): Remove last unnecessary set of insn.
18902 (thumb_pop): Remove unused variable 'pushed_words'.
18903 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18905 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18907 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18908 * config/s390/s390-protos.h: Add prototype of
18909 s390_rel_address_ok_p.
18910 * config/s390/s390.c (s390_got_symbol): New function.
18911 (s390_rel_address_ok_p): New function.
18912 (legitimize_pic_address): Use s390_rel_address_ok_p.
18913 (s390_load_got): Use s390_got_symbol.
18914 (s390_option_override): Issue error if
18915 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18916 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18918 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18920 2017-06-27 Andrew Pinski <apinski@cavium.com>
18922 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18923 (X * copysign (1.0, X)): New pattern.
18924 (X * copysign (1.0, -X)): New pattern.
18925 (copysign (-1.0, CST)): New pattern.
18927 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18929 * genmultilib (combination_space): Remove variable.
18930 Validate reuse rules against regular expression for any sequence
18931 of multilib options in any order.
18933 2017-06-27 Michael Collison <michael.collison@arm.com>
18935 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18936 call aarch64_split_simd_combine.
18937 * (aarch64_combine_internal<mode>): Delete pattern.
18938 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18939 Allow register and subreg operands.
18941 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18943 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18944 specific need, just fallback on defaults.
18945 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18947 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18948 Olivier Hainque <hainque@adacore.com>
18950 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18952 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18953 targets. Pick a default if no particular attempt applied.
18954 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18957 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18959 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18960 (x86_64-wrs-vxworks7): Likewise.
18962 2017-06-27 Marek Polacek <polacek@redhat.com>
18965 * ubsan.c (instrument_null): Check get_base_address's result for null.
18967 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18969 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18971 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18973 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18974 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18975 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18976 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18977 New function types.
18978 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18979 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18980 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18981 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18982 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18983 BUILT_IN_FEUPDATEENV): New builtins.
18984 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18985 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18986 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18987 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18989 (builtin_structptr_types): Adjust size.
18990 * tree.c (builtin_structptr_types): Add four entries.
18992 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18993 Olivier Hainque <hainque@adacore.com>
18995 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18996 (TLS_SYM): New local macro, forcing reference to __tls__ on
18997 link command lines for VxWorks 7 RTPs, triggering initialization
18999 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19000 OS features TLS support, true for RTPs on VxWorks 7.
19001 * config/vxworks.c (vxworks_override_options): Setup emutls
19004 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19006 * predict.c (test_prediction_value_range): Use -1U instead of -1
19007 to avoid narrowing conversion warning.
19008 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19009 to avoid narrowing conversion warning.
19010 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19012 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19014 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19016 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19017 64bit configurations.
19018 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19019 (SIZE_TYPE): Likewise.
19020 * config/vxworks.c (vxworks_emutls_var_fields): Use
19021 long_unsigned_type_node instead of unsigned_type_node as the offset
19022 field type, which is "pointer" mode in emutls.c.
19024 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19027 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19029 PR middle-end/81207
19030 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19031 gimple_vuse copying separately from gimple_vdef copying.
19033 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19035 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19038 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19039 Olivier Hainque <hainque@adacore.com>
19041 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19042 for all vxworks7 targets.
19043 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19044 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19045 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19046 variations for VX6/VX7 and 32/64bits later on in ...
19047 (VXWORKS_LIB_SPEC): Leverage new macros.
19048 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19049 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19051 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19053 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19054 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19056 2017-06-26 Carl Love <cel@us.ibm.com>
19058 * config/rs6000/rs6000-c.c: Add support for built-in functions
19059 vector bool char vec_reve (vector bool char);
19060 vector signed char vec_reve (vector signed char);
19061 vector unsigned char vec_reve (vector unsigned char);
19062 vector bool int vec_reve (vector bool int);
19063 vector signed int vec_reve (vector signed int);
19064 vector unsigned int vec_reve (vector unsigned int);
19065 vector bool long long vec_reve (vector bool long long);
19066 vector signed long long vec_reve (vector signed long long);
19067 vector unsigned long long vec_reve (vector unsigned long long);
19068 vector bool short vec_reve (vector bool short);
19069 vector signed short vec_reve (vector signed short);
19070 vector double vec_reve (vector double);
19071 vector float vec_reve (vector float);
19072 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19073 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19074 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19075 (altivec_vreve): New pattern.
19076 * config/rs6000/altivec.h (vec_reve): New define.
19077 * doc/extend.texi (vec_rev): Update the built-in documentation file
19078 for the new built-in functions.
19080 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19082 PR tree-optimization/71815
19083 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19085 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19087 (slsr_process_phi): Likewise.
19088 (replace_uncond_cands_and_profitable_phis): Don't replace a
19089 multiply candidate with a stride of 1 (copy or cast).
19090 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19092 (lowest_cost_path): Likewise.
19093 (total_savings): Likewise.
19095 2017-06-26 Richard Biener <rguenther@suse.de>
19098 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19099 Use def_builtin_pure for all gather builtins.
19101 2017-06-26 Richard Biener <rguenther@suse.de>
19103 PR tree-optimization/81203
19104 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19105 non-dominating BBs.
19107 2017-06-26 Marek Polacek <polacek@redhat.com>
19110 * doc/invoke.texi: Document -Wmultistatement-macros.
19112 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19114 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19115 arm_neon_ok_no_float_abi effective target.
19117 2017-06-26 Richard Biener <rguenther@suse.de>
19119 PR tree-optimization/80928
19120 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19121 (copy_bbs): Set BB_DUPLICATED flag early.
19122 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19124 (execute_on_shrinking_pred): Likewise.
19125 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19126 BB_DUPLICATED blocks.
19127 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19128 iterate over all PHIs considering removal of *gsi.
19130 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19132 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19135 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19137 * config/rs6000/rs6000-string.c: (expand_block_clear,
19138 do_load_for_compare, select_block_compare_mode,
19139 compute_current_alignment, expand_block_compare,
19140 expand_strncmp_align_check, expand_strn_compare,
19141 expand_block_move, rs6000_output_load_multiple)
19142 Move functions related to string/block move/compare
19143 to a separate file.
19144 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19145 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19146 for this function which is now used in two files.
19147 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19148 * config.gcc: Add rs6000-string.o to extra_objs for
19149 targets powerpc*-*-* and rs6000*-*-*.
19151 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19154 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19155 32-bit, since indexed is not valid for DImode.
19156 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19157 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19158 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19159 (define_peephole2 for Altivec d-form store): Likewise.
19162 * multiple_target.c (create_dispatcher_calls): Only create the
19163 dispatcher call if the function is the default clone of a
19164 versioned function.
19166 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19168 PR middle-end/80902
19169 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19170 a call, force the call to not be a tail call.
19172 2017-06-23 Jeff Law <law@redhat.com>
19174 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19177 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19179 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19180 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19181 handling for early expansion of vector shifts (sl,sr,sra,rl).
19182 (builtin_function_type): Add vector shift right instructions
19183 to the unsigned argument list.
19185 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19187 rtl-optimizatoin/79286
19188 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19189 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19190 trap. PIC register plus a const unspec without offset can never trap.
19192 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19194 * tree.h (builtin_structptr_type): New type.
19195 (builtin_structptr_types): Declare new array.
19196 * tree.c (builtin_structptr_types): New array.
19197 (free_lang_data, build_common_tree_nodes): Use it.
19199 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19202 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19205 2017-06-22 Matt Turner <mattst88@gmail.com>
19207 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19208 Lake models to skylake case. Assume skylake for unknown
19209 models with clflushopt.
19211 2017-06-22 Jeff Law <law@redhat.com>
19213 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19214 frame sizes that do not satisfy aarch64_uimm12_shift.
19216 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19218 * profile-count.h (apply_probability,
19219 apply_scale, probability_in): Fix checks for zero.
19221 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19223 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19224 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19226 2016-06-22 Richard Biener <rguenther@suse.de>
19228 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19229 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19230 REDUC_MAX_EXPR support.
19231 (vectorizable_reduction): Likewise.
19232 (vect_create_epilog_for_reduction): Likewise.
19234 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19236 * match.pd (A / (1 << B) -> A >> B): New.
19237 * generic-match-head.c: Include optabs-tree.h.
19238 * gimple-match-head.c: Likewise.
19239 * optabs-tree.h (target_supports_op_p): New.
19240 * optabs-tree.c (target_supports_op_p): New.
19242 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19244 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19245 $gcc_cv_ld --help output.
19246 (gcc_cv_ld_demangle): Likewise.
19247 (gcc_cv_ld_eh_frame_hdr): Likewise.
19248 (gcc_cv_ld_pie): Likewise.
19249 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19250 (gcc_cv_ld_buildid): Likewise.
19251 (gcc_cv_ld_sysroot): Likewise.
19252 (ld_bndplt_support): Likewise.
19253 (ld_pushpopstate_support): Likewise.
19254 * configure: Regenerate.
19255 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19257 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19260 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19261 operands indexes to avoid gap between operands and match_dups.
19263 2017-06-21 Andrew Pinski <apinski@cavium.com>
19265 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19266 Increment Arith_shift and Arith_shift_reg by 1.
19267 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19269 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19270 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19271 (aarch64_strip_extend): Add new argument and test for it.
19272 (aarch64_cheap_mult_shift_p): New function.
19273 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19274 add a cost if it is true.
19275 Update calls to aarch64_strip_extend.
19276 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19278 2017-06-21 Andrew Pinski <apinski@cavium.com>
19280 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19282 (thunderxt88): Likewise.
19283 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19284 (thunderx_prefetch_tune): New variable.
19285 (thunderx2t99_prefetch_tune): Update for the correct values.
19286 (thunderxt88_tunings): New variable.
19287 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19288 generic_prefetch_tune.
19289 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19291 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19293 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19294 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19295 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19296 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19297 (aarch64_atomic_cas<mode>, GPI): Likewise.
19299 2017-06-21 Martin Liska <mliska@suse.cz>
19301 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19302 statements on cold and hot labels.
19303 * predict.c (tree_estimate_probability_bb): Remove the
19304 prediction from this place.
19306 2017-06-21 Martin Liska <mliska@suse.cz>
19308 PR tree-optimization/79489
19309 * gimplify.c (maybe_add_early_return_predict_stmt): New
19311 (gimplify_return_expr): Call the function.
19312 * predict.c (tree_estimate_probability_bb): Remove handling
19314 * predict.def: Update comment about early return predictor.
19315 * gimple-predict.h (is_gimple_predict): New function.
19316 * predict.def: Change default value of early return to 66.
19317 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19319 * passes.def: Put pass_strip_predict_hints to the beginning of
19322 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19324 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19325 FUNCTION_DECL declarations.
19326 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19328 (dwaf2out_decl): Likewise.
19329 * godump.c (go_early_global_decl): Skip call to the real debug hook
19330 for FUNCTION_DECL declarations.
19331 * passes.c (rest_of_decl_compilation): Skip call to the
19332 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19333 -fdump-go-spec is passed.
19335 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19337 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19338 Reorder for compactness.
19339 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19340 (def_builtin_pure, def_builtin_pure2): New functions.
19341 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19343 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19345 * match.pd (nop_convert): New predicate.
19346 ((A +- CST1) +- CST2): Allow some NOP conversions.
19348 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19351 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19352 with ctors/dtors if GOVD_SHARED is set.
19354 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19356 * config/aarch64/aarch64.md (movti_aarch64):
19357 Emit mov rather than orr.
19358 (movtf_aarch64): Likewise.
19359 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19360 Emit mov rather than orr.
19362 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19364 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19365 Swap alternatives, make integer dup more expensive.
19367 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19369 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19370 Return true for non-tls symbols.
19372 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19374 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19375 (cortex-a75): Likewise.
19376 (cortex-a75.cortex-a55): Likewise.
19377 * config/aarch64/aarch64-tune.md: Regenerate.
19378 * doc/invoke.texi (-mtune): Document new values for -mtune.
19380 2017-06-21 Tom de Vries <tom@codesourcery.com>
19382 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19383 stack_size feature.
19384 (Effective-Target Keywords, Other attributes): Suggest using
19385 dg-add-options stack_size feature to get stack limit in stack_size
19386 effective target documentation.
19388 2017-06-21 Julian Brown <julian@codesourcery.com>
19389 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19391 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19392 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19393 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19395 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19396 attribute type list for neon_multiply.
19397 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19398 attribute type list for neon_multiply.
19399 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19400 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19401 attribute type list for neon_multiply.
19402 * config/arm/types.md (crypto_pmull): Add.
19403 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19404 attribute type list.
19406 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19408 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19411 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19413 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19414 to make sure not to dereference a NULL cost_classes_ptr pointer.
19416 2017-06-20 Carl Love <cel@us.ibm.com>
19418 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19419 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19420 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19421 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19422 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19423 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19424 VMULOSW): New enum "unspec" values.
19425 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19426 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19427 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19428 altivec_vmulosw): New patterns.
19429 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19430 VMULOSW): Add definitions.
19432 2017-06-20 Julia Koval <julia.koval@intel.com>
19434 * config/i386/i386.c: Fix rounding expand for new pattern.
19435 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19437 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19439 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19440 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19442 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19444 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19447 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19449 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19450 architecture, then by implementer ID.
19451 * config/aarch64/aarch64-tune.md: Regenerate.
19453 2017-06-20 Richard Biener <rguenther@suse.de>
19455 PR middle-end/81097
19456 * fold-const.c (split_tree): Fold to type before negating.
19458 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19460 * diagnostic-show-locus.c
19461 (selftest::test_fixit_deletion_affecting_newline): New function.
19462 (selftest::diagnostic_show_locus_c_tests): Call it.
19464 2017-06-20 Andreas Schwab <schwab@suse.de>
19467 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19470 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19472 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19473 __ARM_FEATURE_COPROC according to support.
19475 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19477 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19478 Rewritten to avoid overflow for > 32-bit pointers.
19481 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19482 by removing enum keyword.
19483 (ubsan_type_descriptor): Likewise. Formatting fix.
19486 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19487 splitter): Require TARGET_SSE2 in the condition.
19489 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19492 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19493 for doing vector set of SFmode on ISA 3.0.
19494 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19495 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19497 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19498 SFmode value into a V4SF variable that was extracted from another
19499 V4SF variable without converting the element to double precision
19500 and back to single precision vector format.
19501 (vsx_insert_extract_v4sf_p9_2): Likewise.
19503 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19505 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19506 in UWHI to avoid undefined overflow.
19509 * ubsan.h (enum ubsan_encode_value_phase): New.
19510 (ubsan_encode_value): Change second argument to
19511 enum ubsan_encode_value_phase with default value of
19512 UBSAN_ENCODE_VALUE_GENERIC.
19513 * ubsan.c (ubsan_encode_value): Change second argument to
19514 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19515 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19516 create_tmp_var_raw instead of create_tmp_var and use a
19518 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19519 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19520 ubsan_encode_value callers.
19523 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19524 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19525 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19527 2017-06-19 Richard Biener <rguenther@suse.de>
19529 PR middle-end/81118
19530 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19531 estimates if we changed anything.
19533 2017-06-19 Richard Biener <rguenther@suse.de>
19535 PR tree-optimization/80887
19536 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19537 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19538 simplified lookups, then reset mprts_hook.
19539 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19541 (try_to_simplify): Likewise.
19543 2017-06-19 Martin Liska <mliska@suse.cz>
19546 * gimplify.c (gimplify_switch_expr):
19547 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19549 2017-06-19 Martin Liska <mliska@suse.cz>
19551 * doc/install.texi: Document that PGO runs in 4 stages.
19553 2017-06-19 Martin Liska <mliska@suse.cz>
19556 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19557 to dispatcher function name.
19558 * multiple_target.c (replace_function_decl): New function.
19559 (create_dispatcher_calls): Redirect both edges and references.
19561 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19563 * profile-count.c (profile_count::dump): Dump quality.
19564 (profile_count::differs_from_p): Update for unsigned val.
19565 * profile-count.h (profile_count_quality): New enum.
19566 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19568 2017-06-19 Richard Biener <rguenther@suse.de>
19570 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19571 struct function as arg.
19572 (estimate_numbers_of_iterations): Export overload with loop arg.
19573 (free_numbers_of_iterations_estimates_loop): Use an overload of
19574 free_numbers_of_iterations_estimates instead.
19575 * tree-cfg.c (remove_bb): Adjust.
19576 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19577 * tree-parloops.c (gen_parallel_loop): Likewise.
19578 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19580 (tree_unroll_loops_completely): Likewise.
19581 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19582 Use an overload instead and export.
19583 (estimated_loop_iterations): Adjust.
19584 (max_loop_iterations): Likewise.
19585 (likely_max_loop_iterations): Likewise.
19586 (estimate_numbers_of_iterations): Take struct function as arg
19588 (loop_exits_before_overflow): Adjust.
19589 (free_numbers_of_iterations_estimates_loop): Use an overload.
19590 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19591 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19593 2017-06-19 Richard Biener <rguenther@suse.de>
19596 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19597 RANGE_EXPR conservatively.
19599 2017-06-16 Carl Love <cel@us.ibm.com>
19601 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19602 definitions for vec_float, vec_float2, vec_floato,
19603 vec_floate built-ins.
19604 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19605 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19607 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19608 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19609 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19610 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19611 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19612 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19613 vec_floato): Add builtin defines.
19614 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19615 Update the built-in documentation file for the new built-in
19618 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19620 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19621 (mthumb): Mark as the negative of -marm.
19623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19625 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19627 (ARM Options, -mtune): Document that this accepts the same
19628 extension options as -mcpu.
19629 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19631 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19633 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19634 permitted extensions.
19636 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19638 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19639 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19640 (armv8-m.main): Add option +nodsp.
19641 * config/arm/arm-cpu-cdata.h: Regenerated.
19643 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19645 * config/arm/t-fuchsia: New file.
19646 * config.gcc (arm*-*-fuchsia*): Use it.
19648 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19650 * config/arm/t-symbian: Rewrite for new option infrastructure.
19652 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19654 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19655 (MULTILIB_REQUIRED): Likewise.
19657 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19659 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19660 (MULTILIB_RESUE): Likewise.
19661 (MULTILIB_MATCHES): Likewise.
19662 (MULTLIB_REQUIRED): Likewise.
19664 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19666 * config/arm/t-rtems: Rewrite for new option framework.
19668 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19670 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19671 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19672 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19673 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19674 * config/arm/t-multilib: ... here.
19675 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19676 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19677 armv7-a and armv8*-a when A-profile libraries have not been built.
19678 * config/arm/t-rmprofile: Rewrite.
19680 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19682 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19683 with a backslash. Remove the backslash after substituting unescaped
19685 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19689 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19690 the driver to pass through the default setting of -mfloat-abi.
19691 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19693 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19694 (all_feat_combs): New rule.
19695 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19697 * config/arm/t-aprofile: Rewrite.
19699 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19701 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19702 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19703 fpu is not specified by the user/command-line.
19704 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19705 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19706 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19707 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19708 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19709 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19713 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19714 * config/arm/t-arm-elf: Rewritten.
19716 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19718 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19719 have some floating-point instructions.
19720 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19721 (TARGET_MAYBE_HARD_FLOAT): New macro.
19722 * config/arm/arm-builtins.c (arm_init_builtins): Use
19723 TARGET_MAYBE_HARD_FLOAT.
19724 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19726 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19728 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19729 (configargs.h): Include it.
19730 (arm_print_hint_for_fpu_option): New function.
19731 (arm_parse_fpu_option): New function.
19732 (candidate_extension): New class.
19733 (arm_canon_for_multilib): New function.
19734 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19735 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19736 (ARCH_CANONICAL_SPECS): New macro.
19737 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19739 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19741 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19742 are set after handling multilib fragments. Set target_cpu_default2
19745 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19747 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19749 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19750 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19752 (check_arch): Likewise.
19753 * config/arm/arm.c (arm_configure_build_target): Handle
19754 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19755 options in the default.
19757 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19759 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19760 when an option is an alias of another.
19761 * config/arm/parsecpu.awk (optalias): New parser token.
19762 (gen_comm_data): Mark non-alias options as such. Emit entries
19763 for extension aliases.
19764 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19765 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19766 (armv6kz, armv6zk, armv6t2): Likewise.
19767 (armv7): Make vfpv3-d16 an alias.
19768 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19770 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19771 Sort in canonical order.
19772 (armv8-a): Sort in canonical order.
19773 (armv8.1-a, armv8.2-a): Likewise.
19774 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19776 (cortex-a9): Sort in canonical order.
19777 * config/arm/arm.c (selftests.h): Include it.
19778 (arm_test_cpu_arch_data): New function.
19779 (arm_run_self_tests): New function.
19780 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19781 (targetm): Move declaration to the end of the file.
19782 * arm-cpu-cdata.h: Regenerated.
19784 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19786 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19787 call to target_mode_check describing the type of option passed.
19788 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19789 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19790 arm_parse_cpu_option_name to match parameters against list of
19792 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19793 arm_arch_core_flags data structure.
19794 * config/arm/arm-cpu_cdata.h: Regenerated.
19796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19798 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19800 (arm_print_hint_for_cpu_option): Likewise.
19801 (arm_print_hint_for_arch_option): Likewise.
19802 (arm_parse_cpu_option_name): Likewise.
19803 (arm_parse_arch_option_name): Likewise.
19804 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19805 of entries in the all_fpus list.
19806 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19807 (arm_parse_cpu_option_name): Declare.
19808 (arm_parse_arch_option_name): Declare.
19809 (arm_parse_option_features): Declare.
19810 (arm_intialize_isa): Declare.
19811 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19813 (gen_comm_data): ... here. Make definitions non-static.
19814 * config/arm/arm-cpu-data.h: Regenerated.
19815 * config/arm/arm-cpu-cdata.h: Regenerated.
19817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19819 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19820 (cpu_arch_extension): New structure.
19821 (cpu_arch_option, arch_option, cpu_option): New structures.
19822 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19823 architecture types.
19824 (gen_data): Generate new format data tables.
19825 * config/arm/arm.c (cpu_tune): New structure.
19826 (cpu_option, processors): Delete.
19827 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19828 (arm_print_hint_for_cpu_option): ... this and ...
19829 (arm_print_hint_for_arch_option): ... this.
19830 (arm_parse_arch_cpu_name): Delete. Replace with ...
19831 (arm_parse_cpu_option_name): ... this and ...
19832 (arm_parse_arch_option_name): ... this.
19833 (arm_unrecognized_feature): Change type of target parameter to
19835 (arm_parse_arch_cpu_features): Delete. Replace with ...
19836 (arm_parse_option_features): ... this.
19837 (arm_configure_build_target): Rework to use new configuration data
19839 (arm_print_tune_info): Rework for new configuration data tables.
19840 * config/arm/arm-cpu-data.h: Regenerated.
19841 * config/arm/arm-cpu.h: Regenerated.
19843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19845 * Makefile.in (OBJS): Move sbitmap.o from here ...
19846 (OBJS-libcommon): ... to here.
19848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19850 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19851 (ISA_ALL_CRYPTO): New macro.
19852 (ISA_ALL_SIMD): New macro
19853 (ISA_ALL_FP): New macro.
19854 * config/arm/arm.c (fpu_bitlist): Update initializer.
19855 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19857 (arm9e): Add fpu. Add option for nofp
19858 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19859 (arm926ej-s, arm1026ej-s): Likewise.
19860 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19861 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19862 neon-fp16, neon-vfpv4, nofp and nosimd.
19863 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19864 (cortex-a8): Add fpu. Add option for nofp.
19865 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19866 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19867 (cortex-r4f): Add fpu.
19868 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19869 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19871 (cortex-r8): Likewise.
19872 (cortex-m4): Add fpu. Add option for nofp.
19873 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19874 (cortex-a17.cortex-a7): Likewise.
19875 (cortex-a32): Add fpu. Add options for crypto and nofp.
19876 (cortex-a35, cortex-a53): Likewise.
19877 (cortex-a57): Add fpu. Add option for crypto.
19878 (cortex-a72, cortex-a73): Likewise.
19879 (exynos-m1): Likewise.
19880 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19881 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19882 (cortex-m33): Add fpu. Add option for nofp.
19883 * config/arm/arm-cpu-cdata.h: Regenerated
19884 * config/arm/arm-cpu-data.h: Regenerated.
19886 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19888 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19889 (armv5te, armv5tej): Likewise.
19890 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19891 (armv7): Add options fp and vfpv3-d16.
19892 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19893 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19895 (armv7ve): Likewise.
19896 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19897 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19898 (armv8-a): Add nocrypto option.
19899 (armv8.1-a, armv8.2-a): Likewise.
19900 (armv8-m.main): add options fp, fp.dp and nofp.
19902 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19904 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19906 (armv8-a+crc): Delete.
19907 (armv8.1-a): Add options simd, crypto and nofp.
19908 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19909 (armv8.2-a+fp16): Delete.
19910 (armv8-m.main): Add option dsp.
19911 (armv8-m.main+dsp): Delete.
19912 (cortex-a8): Add fpu. Add nofp option.
19913 (cortex-a9): Add fpu. Add nofp and nosimd options.
19914 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19915 link to main cpu and architecture data structures.
19916 (gen_comm_data): Only put isa attributes from the main architecture
19918 (option): New statement for architecture and CPU entries.
19919 * arm.c (struct cpu_option): New structure.
19920 (struct processors): Add entry for options.
19921 (arm_unrecognized_feature): New function.
19922 (arm_parse_arch_cpu_name): Ignore any characters after the first
19924 (arm_parse_arch_cpu_feature): New function.
19925 (arm_configure_build_target): Separate out any CPU and architecture
19926 features and parse separately. Don't error out if -mfpu=auto is
19927 used with only an architecture string.
19928 (arm_print_asm_arch_directives): New function.
19929 (arm_file_start): Call it.
19930 * config/arm/arm-cpu-cdata.h: Regenerated.
19931 * config/arm/arm-cpu-data.h: Likewise.
19932 * config/arm/arm-tables.opt: Likewise.
19934 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19936 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19937 assembler when it is not -mfpu=auto.
19939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19941 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19942 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19943 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19944 (ASM_CPU_SPEC): Rewrite.
19945 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19946 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19947 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19948 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19949 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19950 copied string is NUL-terminated. Also strip any characters prefixed
19952 (arm_rewrite_selected_arch): New function.
19953 (arm_rewrite_march): New function.
19955 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19957 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19958 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19959 (march, mcpu, mtune): Convert to string-based options.
19960 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19961 (arm_parse_arch_cpu_name): New function.
19962 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19963 identify selected architecture or CPU.
19964 (arm_option_save): New function.
19965 (TARGET_OPTION_SAVE): Redefine.
19966 (arm_option_restore): Restore string options.
19967 (arm_option_print): Print string options.
19969 2017-06-16 Martin Sebor <msebor@redhat.com>
19971 PR tree-optimization/80933
19972 PR tree-optimization/80934
19973 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19974 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19975 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19976 (gimple_fold_builtin): Call them.
19978 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19980 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19981 as unlikely; update profile.
19983 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19985 * predict.c (force_edge_cold): Handle declaring edges impossible
19988 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19990 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19992 (try_unroll_loop_completely): Fix reporting.
19994 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19996 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19998 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20001 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20002 if given a non-constant argument for an intrinsic which requires a
20005 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20007 * profile.c (compare_freqs): New function.
20008 (branch_prob): Sort edge list.
20009 (find_spanning_tree): Assume that the list is priority sorted.
20011 2017-06-16 Richard Biener <rguenther@suse.de>
20013 PR tree-optimization/81090
20014 * passes.def (pass_record_bounds): Remove.
20015 * tree-pass.h (make_pass_record_bounds): Likewise.
20016 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20017 make_pass_record_bounds): Likewise.
20018 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20019 not free niter estimates at the beginning but at the end.
20020 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20022 2017-06-16 Richard Biener <rguenther@suse.de>
20024 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20025 initializer to workaround ICE in host GCC 4.8.
20027 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20029 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20031 (clone_inlined_nodes): Update.
20033 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20035 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20036 prefetch settings, and enable prefetching by default at -O3.
20038 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20040 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20041 Set flag_prefetch_loop_arrays according to tuning data.
20043 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20045 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20046 New tune structure.
20047 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20048 [Unrelated to main purpose of the patch] Place the pointer field last
20049 to enable type checking errors when tune structure are wrongly merged.
20050 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20051 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20052 (thunderx2t99_prefetch_tune): New tune constants.
20053 (tune_params *_tunings): Update all tunings (no functional change).
20054 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20055 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20056 from tunings structures.
20058 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20061 * ubsan.c (instrument_null): Add T argument, use it instead
20062 of computing it based on IS_LHS.
20063 (instrument_object_size): Likewise.
20064 (pass_ubsan::execute): Adjust instrument_null and
20065 instrument_object_size callers to pass gimple_get_lhs or
20066 gimple_assign_rhs1 result to it. Use instrument_null instead of
20067 calling get_base_address and instrument_mem_ref. Handle
20068 aggregate call arguments for object-size sanitization.
20070 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20072 PR tree-optimization/81089
20073 * tree-vrp.c (is_masked_range_test): Validate operands of
20076 2017-06-15 Martin Sebor <msebor@redhat.com>
20079 * dumpfile.c (dump_register): Avoid calling memset to initialize
20080 a class with a default ctor.
20081 * gcc.c (struct compiler): Remove const qualification.
20082 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20083 * hash-table.h: Ditto.
20084 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20086 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20087 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20089 * params.h (struct param_info): Make struct members non-const.
20090 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20091 with default initialization.
20092 * vec.h (vec_copy_construct, vec_default_construct): New helper
20094 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20095 with vec_copy_construct.
20096 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20097 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20098 * doc/invoke.texi (-Wclass-memaccess): Document.
20100 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20102 * emit-rtl.h (is_leaf): Update comment about local
20103 register allocator.
20105 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20108 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20109 for a variable to have a section before checking if the section has a
20111 Set section to.persistent if persistent attribute is set.
20112 Warn if .persistent attribute is used on an automatic variable.
20114 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20116 PR rtl-optimization/80474
20117 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20119 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20121 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20124 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20126 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20127 (casesi): Emit bounds checking as RTL.
20128 (casesi_internal_mips16_<mode>): Remove bounds checking.
20130 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20132 * config/xtensa/xtensa.c (xtensa_option_override): Append
20133 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20134 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20135 xtensa_doloop_hooks): Define unconditionally.
20136 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20138 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20139 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20140 for it in xtensa_option_override.
20141 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20142 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20144 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20146 * doc/cppopts.texi: Document '-' special value to -MF.
20148 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20150 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20151 (cortex_a53_fconst): Likewise.
20152 (cortex_a53_fpmul): Likewise.
20153 (cortex_a53_f_load_64): Likewise.
20154 (cortex_a53_f_load_many): Likewise.
20155 (cortex_a53_advsimd_alu): Likewise.
20156 (cortex_a53_advsimd_alu_q): Likewise.
20157 (cortex_a53_advsimd_mul): Likewise.
20158 (cortex_a53_advsimd_mul_q): Likewise.
20159 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20160 Add missing fmul, r2f_cvt and fconst cases.
20162 2017-06-14 Richard Biener <rguenther@suse.de>
20164 PR middle-end/81088
20165 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20167 (fold_binary_loc): When associating do not treat pre-existing
20168 TREE_OVERFLOW on literal constants as a reason to allow
20169 TREE_OVERFLOW on associated literal constants.
20171 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20173 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20174 (MASK_FEATURES): New macro.
20175 * config/sparc/sparc.c (sparc_option_override): Remove the special
20176 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20178 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20180 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20181 a division of 0 if non-call exceptions are enabled.
20183 2017-06-14 Andrew Pinski <apinski@cavium.com>
20184 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20187 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20188 Improve vector initialization code gen for only variable case.
20190 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20192 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20194 2017-06-14 Richard Biener <rguenther@suse.de>
20196 PR tree-optimization/81083
20197 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20200 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20202 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20203 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20204 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20205 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20206 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20207 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20209 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20211 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20212 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20214 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20216 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20218 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20220 * config/rs6000/t-rtems: Don't handle SPE.
20222 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20224 * config/rs6000/t-linux: Don't handle SPE.
20226 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20228 * config/rs6000/eabispe.h: Delete file.
20230 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20232 * config/rs6000/t-spe: Delete file.
20234 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20236 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20237 (rs6000_legitimate_offset_address_p): Return false for anything in
20238 V2SImode or V2SFmode.
20240 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20242 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20243 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20244 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20246 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20247 (rs6000_legitimate_offset_address_p): Ditto.
20248 (rs6000_emit_move): Ditto.
20249 (rs6000_init_builtins): Remove V4HI_type_node.
20251 2017-06-13 Martin Liska <mliska@suse.cz>
20254 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20255 (gate_asan): Likewise.
20256 * asan.h (asan_no_sanitize_address_p): Remove the function.
20257 (sanitize_flags_p): New function.
20258 * builtins.def: Fix coding style.
20259 * common.opt: Use renamed enum value.
20260 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20261 * doc/extend.texi: Document no_sanitize attribute.
20262 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20263 to SANITIZE_UNDEFINED_NONDEFAULT.
20264 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20265 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20266 Use sanitize_flags_p.
20267 * gimplify.c (gimplify_function_tree): Likewise.
20268 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20269 * opts.c (parse_no_sanitize_attribute): New function.
20270 (common_handle_option): Use renamed enum value.
20271 * opts.h (parse_no_sanitize_attribute): Declare.
20272 * tree.c (sanitize_flags_p): New function.
20273 * tree.h: Declared here.
20274 * tsan.c: Use sanitize_flags_p.
20275 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20276 (instrument_mem_ref): Likewise.
20277 (instrument_bool_enum_load): Likewise.
20278 (do_ubsan_in_current_function): Remove the function.
20279 (pass_ubsan::execute): Use sanitize_flags_p.
20280 * ubsan.h: Remove do_ubsan_in_current_function
20281 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20282 (dump_function_to_file): Use it here.
20284 2017-06-13 Martin Jambor <mjambor@suse.cz>
20286 PR tree-optimization/80803
20287 PR tree-optimization/81063
20288 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20289 (propagate_subaccesses_across_link): Enqueue subtree whenever
20290 necessary instead of relying on the caller.
20292 2017-06-13 Martin Jambor <mjambor@suse.cz>
20294 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20295 that have a first_link.
20296 (sort_and_splice_var_accesses): Do not check first_link before
20298 (subtree_mark_written_and_enqueue): Likewise.
20299 (propagate_all_subaccesses): Likewise and do not stop at first
20300 parent with a first_link.
20302 2017-06-13 Martin Jambor <mjambor@suse.cz>
20304 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20307 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20309 * match.pd: New pattern.
20311 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20313 * tree-vrp.c (is_masked_range_test): New function.
20314 (register_edge_assert_for): Determine ranges for
20317 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20319 PR tree-optimization/67328
20320 * fold-const.c (maskable_range_p): New function.
20321 (build_range_check): Generate bittests if possible.
20323 2017-06-13 Martin Liska <mliska@suse.cz>
20325 * gimple-pretty-print.c (dump_probability): Add new argument.
20326 (dump_edge_probability): Dump both probability and count.
20327 (dump_gimple_label): Likewise.
20328 (dump_gimple_bb_header): Likewise.
20330 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20333 * config/avr/avr-devices.c: Fix indentation.
20334 * config/avr/gen-avr-mmcu-specs.c: Dito.
20336 2017-06-13 Richard Biener <rguenther@suse.de>
20338 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20339 instead get vector type from stmt_info.
20340 (vectorizable_reduction): Adjust. Remove dead code.
20342 2017-06-13 Richard Biener <rguenther@suse.de>
20344 PR middle-end/81065
20345 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20346 case of C * (x * C2 + C3).
20347 (fold_addr_of_array_ref_difference): Properly fold index difference.
20349 2017-06-12 David S. Miller <davem@davemloft.net>
20352 * config/sparc/sparc.md (return expander): Emit frame blockage if
20353 function uses alloca.
20355 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20357 * combine.c (make_field_assignment): Check len rather than the mode
20358 precision when calling force_to_mode.
20360 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20362 Support multilibs and devices that see flash in RAM address range.
20365 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20366 (avr_mcu_t) <flash_pm_offset>: New field.
20367 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20368 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20369 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20370 (AVR_TINY_PM_OFFSET): Remove macro.
20371 * config/avr/avr.opt (-mshort-calls): New option.
20372 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20373 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20374 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20375 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20376 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20377 instead of avr_arch->have_jmp_call.
20378 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20379 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20380 avr_arch->flash_pm_offset to define.
20381 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20382 new field flash_pm_offset. Add entry for avrxmega3.
20383 (avr_texinfo): Add entry for avrxmega3.
20384 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20385 attiny212, attiny214,
20386 attiny412, attiny414, attiny416, attiny417,
20387 attiny814, attiny816, attiny817,
20388 attiny1614, attiny1616, attiny1617,
20389 attiny3214, attiny3216, attiny3217.
20390 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20391 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20392 (avr_print_operand_address) [AVR_TINY]: Same.
20393 (avr_asm_init_sections) <readonly_data_section>: Only patch
20394 callback if avr_arch->flash_pm_offset = 0.
20395 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20396 for rodata if avr_arch->flash_pm_offset != 0.
20397 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20398 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20399 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20400 (m_options): Append opt_rcall.
20401 (m_dirnames): Append dir_rcall.
20402 * config/avr/t-multilib: Regenerate.
20404 * configure.ac [target=avr]: Check whether avrxmega3 default
20405 linker description file works as needed.
20406 * configure: Regenerate.
20407 * doc/avr-mmcu.texi: Regenerate.
20408 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20409 <__AVR_ARCH__>: Document avrxmega3 and 103.
20410 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20411 <__AVR_SHORT_CALLS__>: Document it.
20412 <__AVR_PM_BASE_ADDRESS__>: Document it.
20413 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20414 (AVR Variable Attributes) <progmem>: Document this is
20415 not needed for avrxmega3.
20416 (AVR Named Address Spaces) <__flash>: Dito.
20418 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20420 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20422 2017-06-12 Doug Rupp <rupp@adacore.com>
20424 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20425 Append vxworks-stdint.h to the tm_file list.
20426 * config/vxworks-stdint.h: New file.
20428 2017-06-12 Martin Liska <mliska@suse.cz>
20430 PR tree-optimization/81041
20431 * tree-profile.c (gimple_gen_ic_func_profiler):
20432 Create an extra BB in profile-generate
20433 (gimple_gen_time_profiler): Likewise.
20435 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20437 PR tree-optimization/81003
20438 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20439 (update_range_test): Use it instead of force_gimple_operand_gsi.
20441 2017-06-12 Richard Biener <rguenther@suse.de>
20443 PR tree-optimization/81053
20444 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20445 with backedge value not defined in loop. Simplify def stmt
20448 2017-06-11 Tom de Vries <tom@codesourcery.com>
20451 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20453 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20454 nvptx_cannot_force_const_mem.
20456 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20458 * opts.c (finish_options): Move test for flag_split_stack after
20459 it has been initialized.
20461 2017-06-11 Jason Merrill <jason@redhat.com>
20463 * tree.h (id_equal): New.
20464 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20465 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20466 instead of strcmp of IDENTIFIER_POINTER.
20468 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20470 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20471 (mark_all_inlined_calls_cdtor): Fix formating.
20472 (inline_transform): Rescale profile before inlining.
20474 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20476 * cgraph.h (cgraph_edge::clone): Update prototype.
20477 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20478 (cgraph_node::create_clone): Update.
20479 (cgraph_node::create_version_clone): Update.
20480 * tree-inline.c (copy_bb): Update.
20481 (expand_call_inline): Update.
20483 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20485 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20486 factored out from ...
20487 (rs6000_emit_prologue): ... here.
20489 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20491 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20492 factored out from ...
20493 (rs6000_emit_prologue): ... here.
20495 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20497 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20499 (handle_missing_profiles): Fix computation of tp_first_run.
20500 (counts_to_freqs): Do not touch freqs when count is 0.
20502 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20504 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20507 2017-06-10 Tom de Vries <tom@codesourcery.com>
20509 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20510 attributes): Document signal effective target.
20512 2017-06-10 Tom de Vries <tom@codesourcery.com>
20514 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20515 Document effective target stack_size.
20517 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20519 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20520 to the edit_context if they can be auto-applied.
20522 2017-06-9 Ian Lance Taylor <iant@golang.org>
20524 * opts.c (finish_options): If -fsplit-stack, disable implicit
20525 -forder-blocks-and-partition.
20526 * doc/invoke.texi (Optimize Options): Document that when using
20527 -fsplit-stack -forder-blocks-and-partition is not implicitly
20530 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20532 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20533 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20534 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20535 * builtins.def (abort, trap, unreachable): Declare cold.
20536 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20537 * tree-core.h (ECF_COLD): New.
20538 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20539 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20541 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20543 * predict.c (unlikely_executed_stmt_p): Cleanup.
20545 2017-06-09 Richard Biener <rguenther@suse.de>
20547 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20548 model if the ref is always written to.
20550 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20552 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20554 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20556 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20559 2017-06-09 Tom de Vries <tom@codesourcery.com>
20562 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20563 "target cannot support label values" when encountering LABEL_REF.
20565 2017-06-09 Martin Liska <mliska@suse.cz>
20567 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20568 (gimple_gen_ic_func_profiler): Emit direct comparison
20569 of __gcov_indirect_call_callee with NULL.
20570 (gimple_gen_time_profiler): Change probability from
20571 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20573 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20575 * profile.c (edge_gcov_counts): Turn to pointer.
20576 (compute_branch_probabilities, compute_branch_probabilities): Update.
20577 (branch_prob): Do not clear edge_gcov_count.
20578 * profile.h (edge_gcov_counts): Turn to pointer.
20579 (edge_gcov_count): Update.
20581 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20583 * gimple.h (gimple_check_failed): Mark cold.
20585 2017-06-09 Richard Biener <rguenther@suse.de>
20587 PR tree-optimization/66623
20588 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20589 refactor check_reduction into two parts, properly computing
20590 whether we have to check reduction validity for outer loop
20593 2017-06-09 Richard Biener <rguenther@suse.de>
20595 PR tree-optimization/79483
20596 * graphite-scop-detection.c (order): New global.
20597 (get_order): Compute bb to order mapping that satisfies code
20598 generation constraints.
20599 (cmp_pbbs): New helper.
20600 (build_scops): Start domwalk at entry block, sort generated
20603 2017-06-09 Richard Biener <rguenther@suse.de>
20605 PR middle-end/81007
20606 * ipa-polymorphic-call.c
20607 (ipa_polymorphic_call_context::restrict_to_inner_class):
20608 Skip FIELD_DECLs with error_mark_node type.
20609 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20612 2017-06-09 Martin Liska <mliska@suse.cz>
20614 * predict.c (struct branch_predictor): New struct.
20615 (test_prediction_value_range): New test.
20616 (predict_c_tests): New function.
20617 * selftest-run-tests.c (selftest::run_tests): Run the function.
20618 * selftest.h: Declare new tests.
20620 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20623 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20624 gen_add3_insn did not fail.
20625 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20626 r0, construct that number in a temporary reg and add that reg to r0.
20627 If asked to put the result in r0 as well, fail.
20629 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20631 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20632 for early expansion of vec_eqv.
20634 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20636 PR middle-end/81005
20637 * ubsan.c (instrument_null): Avoid pointless code temporary.
20638 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20640 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20644 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20647 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20650 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20652 * predict.c (unlikely_executed_edge_p): Move ahead.
20653 (probably_never_executed_edge_p): Use it.
20655 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20657 PR middle-end/79988
20658 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20659 gimple_call_builtin_p call.
20661 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20663 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20664 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20665 rtl_check_failed_type2, rtl_check_failed_code1,
20666 rtl_check_failed_code2, rtl_check_failed_code_mode,
20667 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20668 rtvec_check_failed_bounds, rtl_check_failed_flag,
20669 _fatal_insn_not_found, _fatal_insn): Likewise.
20670 * tree.h (tree_contains_struct_check_failed,
20671 tree_check_failed, tree_not_check_failed,
20672 tree_class_check_failed, tree_range_check_failed,
20673 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20674 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20675 tree_operand_check_failed, omp_clause_check_failed,
20676 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20679 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20681 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20682 flag_branch_probabilities.
20683 * ipa-inline.c (edge_badness): Likewise.
20684 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20685 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20686 * predict.c (maybe_hot_frequency_p): Likewise.
20687 (probably_never_executed): Likewise.
20688 * sched-ebb.c (schedule_ebbs): Likewise.
20689 * sched-rgn.c (find_single_block_region): Likewise.
20690 * tracer.c (tail_duplicate): Likewise.
20692 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20694 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20695 longer requires x_flag_profile_use.
20697 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20699 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20700 instead of flag_reorder_blocks_and_partition.
20701 * dbxout.c (dbxout_function_end): Likewise.
20702 * dwarf2out.c (gen_subprogram_die): Likewise.
20703 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20704 * hw-doloop.c (reorg_loops): Likewise.
20705 * varasm.c (assemble_start_function,
20706 assemble_end_function): Likewise.
20707 (decide_function_section): Do not check for
20708 flag_reorder_blocks_and_partition.
20710 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20712 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20714 (chkp_get_hard_register_fake_addr_expr): Ditto.
20715 (chkp_build_addr_expr): Add check for hard reg case.
20716 (chkp_parse_array_and_component_ref): Ditto.
20717 (chkp_find_bounds_1): Ditto.
20718 (chkp_process_stmt): Don't generate bounds store for
20721 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20723 * predict.c (maybe_hot_bb_p): Do not check profile status.
20724 (maybe_hot_edge_p): Likewise.
20725 (probably_never_executed): Check for zero counts even if profile
20727 (unlikely_executed_edge_p): New function.
20728 (unlikely_executed_stmt_p): New function.
20729 (unlikely_executed_bb_p): New function.
20730 (set_even_probabilities): Use unlikely predicates.
20731 (combine_predictions_for_bb): Likewise.
20732 (predict_paths_for_bb): Likewise.
20733 (predict_paths_leading_to_edge): Likewise.
20734 (determine_unlikely_bbs): New function.
20735 (estimate_bb_frequencies): Use it.
20736 (compute_function_frequency): Use zero counts even if profile is
20738 * profile-count.h: Fix typo.
20740 2017-08-08 Julia Koval <julia.koval@intel.com>
20742 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20743 _mm512_mask_cvtsepi16_storeu_epi8,
20744 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20745 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20746 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20747 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20748 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20749 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20750 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20751 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20752 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20753 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20754 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20755 __builtin_ia32_pmovuswb256mem_mask,
20756 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20757 __builtin_ia32_pmovwb512mem_mask): New builtins.
20759 2017-08-08 Julia Koval <julia.koval@intel.com>
20761 PR target/73350,80862
20762 * config/i386/subst.md (round): Fix round pattern.
20763 * config/i386/i386.c (ix86_erase_embedded_rounding):
20764 Fix erasing rounding for the fixed pattern.
20766 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20768 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20770 2017-06-08 Martin Liska <mliska@suse.cz>
20772 PR gcov-profile/80911
20773 * gcov.c (block_info::block_info): New constructor.
20775 2017-06-07 Carl Love <cel@us.ibm.com>
20777 * config/rs6000/rs6000-c: The return type of the following
20778 built-in functions was implemented as int not long long. Fix sign
20779 of return value for the unsigned version of vec_mulo and vec_mule.
20780 vector unsigned long long vec_bperm (vector unsigned long long,
20781 vector unsigned char)
20782 vector signed long long vec_mule (vector signed int,
20784 vector unsigned long long vec_mule (vector unsigned int,
20785 vector unsigned int)
20786 vector signed long long vec_mulo (vector signed int,
20788 vector unsigned long long vec_mulo (vector unsigned int,
20789 vector unsigned int)
20790 * doc/extend.texi: Fix the documentation for the built-in
20793 2017-06-07 Carl Love <cel@us.ibm.com>
20796 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20799 2017-06-07 Carl Love <cel@us.ibm.com>
20801 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20802 support, Generate doublehv for signed int/float for BE case only.
20804 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20806 * doc/invoke.texi (mcx16): Rewrite.
20808 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20810 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20811 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20812 *mov<mode>_softfloat, and an anonymous splitter): Use
20813 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20815 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20817 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20819 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20820 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20821 (rs6000_debug_reg_global): Adjust.
20822 (rs6000_init_hard_regno_mode_ok): Adjust.
20823 (rs6000_dbx_register_number): Adjust.
20824 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20825 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20826 Remove SPE_ACC and SPEFSCR.
20827 (REG_ALLOC_ORDER): Ditto.
20828 (FRAME_POINTER_REGNUM): Change to 111.
20829 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20830 (REG_CLASS_NAMES): Ditto.
20831 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20832 (REGISTER_NAMES): Ditto.
20833 (ADDITIONAL_REG_NAMES): Ditto.
20834 (rs6000_reg_names): Ditto.
20835 * config/rs6000/rs6000.md: Renumber some register number
20838 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20840 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20842 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20843 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20845 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20846 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20847 Delete the SPE high registers.
20848 (REG_ALLOC_ORDER): Ditto.
20849 (enum reg_class): Remove SPE_HIGH_REGS.
20850 (REG_CLASS_NAMES): Ditto.
20851 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20852 (REGISTER_NAMES): Ditto.
20853 (rs6000_reg_names): Ditto.
20854 * doc/tm.texi.in: Remove SPE as example.
20855 * doc/tm.texi: Regenerate.
20857 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20859 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20860 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20861 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20862 * config/rs6000/rs6000.md (type): Remove "brinc".
20864 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20866 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20867 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20868 * config/rs6000/linuxspe.h: Delete file.
20869 * config/rs6000/rs6000.md: Don't include spe.md.
20870 * config/rs6000/spe.h: Delete file.
20871 * config/rs6000/spe.md: Delete file.
20872 * config/rs6000/t-rs6000: Remove spe.md.
20874 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20876 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20877 (reg_or_none500mem_operand): Delete.
20878 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20879 instead of reg_or_none500mem_operand.
20881 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20883 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20884 handling of SPE flags.
20885 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20887 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20889 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20891 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20892 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20893 paired_divv2sf3): Similar.
20894 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20895 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20896 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20898 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20899 Rename the paired_* instruction patterns.
20900 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20902 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20903 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20904 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20905 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20906 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20907 PAIRED_VECTOR_MODE.
20908 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20909 (spe_func_has_64bit_regs_p): Delete.
20910 (spe_expand_predicate_builtin): Delete.
20911 (spe_expand_evsel_builtin): Delete.
20912 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20913 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20914 (invalid_e500_subreg): Delete.
20915 (rs6000_legitimize_address): Always force_reg op2 as well, for
20916 paired single memory accesses.
20917 (rs6000_member_type_forces_blk): Delete.
20918 (rs6000_spe_function_arg): Delete.
20919 (rs6000_expand_unop_builtin): Delete SPE handling.
20920 (rs6000_expand_binop_builtin): Ditto.
20921 (spe_expand_stv_builtin): Delete.
20922 (bdesc_2arg_spe): Delete.
20923 (spe_expand_builtin): Delete.
20924 (spe_expand_predicate_builtin): Delete.
20925 (spe_expand_evsel_builtin): Delete.
20926 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20927 (spe_init_builtins): Delete.
20928 (spe_func_has_64bit_regs_p): Delete.
20929 (savres_routine_name): Delete "info" parameter. Adjust callers.
20930 (rs6000_emit_stack_reset): Ditto.
20931 (rs6000_dwarf_register_span): Delete.
20932 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20933 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20934 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20936 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20938 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20939 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20940 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20941 mulv2sf3, divv2sf3): Delete expanders.
20943 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20945 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20947 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20949 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20950 * config/rs6000/rs6000.c: Ditto.
20952 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20954 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20955 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20956 comparison_operator.
20958 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20960 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20961 * config/rs6000/rs6000.opt: Ditto.
20962 * config/rs6000/t-rtems: Ditto.
20964 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20966 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20967 TARGET_E500_SINGLE by 0, simplify.
20968 * config/rs6000/rs6000.c: Ditto.
20969 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20970 (spe_build_register_parallel): Delete.
20971 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20972 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20973 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20974 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20975 (E500_CONVERT): Delete.
20976 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20978 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20980 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20981 * config/rs6000/dfp.md: Ditto.
20982 (negdd2, *negdd2_fpr): Merge.
20983 (absdd2, *absdd2_fpr): Merge.
20984 (negtd2, *negtd2_fpr): Merge.
20985 (abstd2, *abstd2_fpr): Merge.
20986 * config/rs6000/e500.h: Delete file.
20987 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20988 TARGET_FPRS by 1 and simplify.
20989 * config/rs6000/rs6000-c.c: Ditto.
20990 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20991 TARGET_DF_SPE by 0.
20992 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20994 * config/rs6000/rs6000.md: Ditto.
20995 (floatdidf2, *floatdidf2_fpr): Merge.
20996 (move_from_CR_gt_bit): Delete.
20997 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20998 (E500_CR_IOR_COMPARE): Delete.
20999 (All patterns that require !TARGET_FPRS): Delete.
21000 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21002 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21004 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21006 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21008 * graphds.c (add_edge): Intitialize edge's attached data.
21009 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21010 pointer parameter. Call pointed function on each edge during
21011 graph traversing. Skip traversing the edge when the function
21013 (graphds_dfs, graphds_scc): Ditto.
21014 (for_each_edge): New parameter. Pass the new parameter to callback
21016 * graphds.h (skip_edge_callback): New function pointer type.
21017 (graphds_dfs, graphds_scc): New function pointer parameter.
21018 (graphds_edge_callback, for_each_edge): New parameter.
21020 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21022 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21023 out code checking if runtime alias check is possible to below ...
21024 Call the new function.
21025 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21026 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21028 2017-06-07 Marek Polacek <polacek@redhat.com>
21031 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21032 TYPE_OVERFLOW_WRAPS check.
21034 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21036 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21037 if versioning is required.
21038 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21039 peeling with the check for versioning.
21041 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21043 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21044 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21045 Set true to new parameter if new ssa variable is defined.
21046 (vect_gen_vector_loop_niters): Refactor. Set range information
21047 for the new vector loop bound variable.
21048 (vect_do_peeling): Ditto.
21050 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21052 * tree-affine.c (ssa.h): Include header file.
21053 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21054 has wrapping overflow behavior.
21056 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21058 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21060 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21062 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21063 (tree_to_aff_combination): ... here.
21065 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21067 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21068 reg_pressure model function.
21069 (ivopts_global_cost_for_size): Delete.
21070 (determine_set_costs, iv_ca_recount_cost): Call new model function
21071 ivopts_estimate_reg_pressure.
21073 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21075 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21076 expensive than udiv. Remove floating point cases from mod.
21078 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21080 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21081 Increase idiv cost.
21083 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21085 * config/aarch64/aarch64.md
21086 (copysignsf3): Fix mask generation.
21088 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21090 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21092 (class dump_manager): Add register_dumps method.
21093 * dumpfile.c: Include langhooks.h.
21094 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21095 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21096 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21097 (dump_manager::dump_register): Start with 512 entries instead of 32.
21098 (dump_manager::register_dumps): New method.
21099 * toplev.c (general_init): Instead of invoking register_dumps
21100 langhook, invoke register_dumps method on the dump manager.
21101 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21104 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21106 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21107 Say that # requires an associated define_split to exist, and that
21108 the define_split must be suitable for use after register allocation.
21110 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21112 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21113 (compute_outgoing_frequencies): Also initialize zero counts.
21114 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21115 around loops; preserve more of profile when nothing changes.
21117 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21119 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21121 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21122 * config/arm/arm-cpu-cdata.h: Regenerate.
21123 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21124 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21125 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21126 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21127 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21129 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21131 * config/arm/t-rmprofile: Likewise.
21132 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21134 2017-06-06 David S. Miller <davem@davemloft.net>
21137 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21138 blockage if function uses alloca.
21140 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21142 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21143 New "uid" fields to hold pretty-print IDs of group and ref.
21144 Memory references are now identified as <group_id>:<ref_id>
21145 instead of using [random] addresses.
21146 (dump_mem_details): Simplify, no functional change.
21147 (dump_mem_ref): Simplify and make output more concise.
21148 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21149 (find_or_create_group): Initialize group uid.
21150 (record_ref): Initialize ref uid. Improve debug output.
21151 (prune_group_by_reuse, should_issue_prefetch_p,)
21152 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21153 (mark_nontemporal_store, determine_loop_nest_reuse):
21154 Improve debug output.
21156 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21158 * dbgcnt.def (prefetch): New debug counter.
21159 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21160 (schedule_prefetches): Stop issueing prefetches if debug counter
21163 2017-06-06 Tom de Vries <tom@codesourcery.com>
21165 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21166 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21168 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21170 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21171 Use aarch64_reg_or_zero predicate for operand 4.
21172 (aarch64_compare_and_swap<mode> define_insn_and_split):
21173 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21174 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21176 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21178 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21179 (arm_compute_save_core_reg_mask): This.
21180 (thumb1_compute_save_reg_mask): Rename into ...
21181 (thumb1_compute_save_core_reg_mask): This.
21182 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21183 (arm_compute_frame_layout): Likewise.
21185 2017-06-06 Richard Biener <rguenther@suse.de>
21187 PR tree-optimization/80974
21188 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21189 keep or clear leaders SSA info.
21191 2017-06-06 Tom de Vries <tom@codesourcery.com>
21193 * config/nvptx/nvptx.c (split_mode_p): New function.
21194 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21196 2017-06-06 Tom de Vries <tom@codesourcery.com>
21198 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21200 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21203 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21206 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21208 * shrink-wrap.c (handle_simple_exit): Update profile.
21209 (try_shrink_wrapping): Upate profile.
21211 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21213 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21214 (tree_guess_outgoing_edge_probabilities): New.
21215 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21216 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21218 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21220 * ipa-split.c (split_function): Initialize return bb profile.
21222 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21224 * profile.c (compute_branch_probabilities): Also initialize
21225 EXIT_BLOCK profile.
21227 2017-06-06 Richard Biener <rguenther@suse.de>
21229 PR tree-optimization/80928
21230 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21231 (vect_analyze_loop_operations): Properly guard analysis for
21233 (vect_transform_loop): Likewise.
21234 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21235 (vect_model_induction_cost): Do not cost for pure SLP.
21236 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21237 of induction in inner loop vectorization.
21238 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21239 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21240 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21242 (vect_analyze_slp_cost_1): Cost induction.
21243 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21244 (vect_get_slp_vect_defs): Likewise.
21245 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21246 (vect_transform_stmt): Handle SLP reductions.
21247 * tree-vectorizer.h (vectorizable_induction): Adjust.
21249 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21251 * config/rs6000/rs6000.c (make_resolver_func): Update
21252 init_lowered_empty_function call.
21254 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21256 * doc/invoke.texi: Document the -fprofile-abs-path option.
21257 * common.opt (fprofile-abs-path): New option.
21258 * gcov-io.h (gcov_write_filename): Declare.
21259 * gcov-io.c (gcov_write_filename): New function.
21260 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21261 * profile.c (output_location): Likewise.
21263 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21265 * shring-wrap.c: Revert accidental commit.
21267 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21269 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21271 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21273 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21275 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21276 profile in callgraph edge.
21277 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21278 (apply_scale): Likewise.
21279 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21280 Also scale profile when inlining function with zero profile.
21281 (initialize_cfun): Update exit block profile even when it is zero.
21282 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21283 when profile is read.
21285 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21287 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21288 (CLONE_*): New constants to define the processors we can generate
21289 code for with the target_clone attribute.
21290 (rs6000_clone_map): New array to identify which clone processors
21291 the current program is running on.
21292 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21293 target_clone attribute.
21294 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21295 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21296 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21297 (cpu_expand_builtin): Add support for target_clone attribute.
21298 (rs6000_valid_attribute_p): Allow "default" attribute.
21299 (get_decl_name): New debug function to simplify printing the
21300 current function name in debugging statements.
21301 (rs6000_clone_priority): New functions to support the target_clone
21302 attribute, and be able to generate code to switch between ISA 2.05
21303 through ISA 3.0 (power6 through power9).
21304 (rs6000_compare_version_priority): Likewise.
21305 (rs6000_get_function_versions_dispatcher): Likewise.
21306 (make_resolver_func): Likewise.
21307 (add_condition_to_bb): Likewise.
21308 (dispatch_function_versions): Likewise.
21309 (rs6000_generate_version_dispatcher_body): Likewise.
21310 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21311 (fusion_gpr_load_p): Fix a spacing issue.
21312 * doc/extend.texi (Common Function Attributes): Document that the
21313 PowerPC supports the target_clone attribute.
21315 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21317 * config/arm/arm.h: explain F symbol found in description of ARM
21318 register allocation in its legend.
21320 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21322 * config/mips/frame-header-opt.c: Include profile-count.h.
21323 * config/riscv/riscv.c: Include profile-count.h
21325 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21327 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21329 (sm_set_flag_if_changed): Add bbs field.
21330 (execute_sm_if_changed_flag_set): Pass BBS.
21331 (execute_sm): Update.
21333 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21335 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21338 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21340 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21341 (peephole2): New peephole2 to emit the above.
21342 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21344 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21346 * config/aarch64/aarch64.c (define_peephole2 above
21347 *sub_<shift>_<mode>): New peephole.
21349 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21351 * config/i386/i386.c (make_resolver_func): Update.
21352 * Makefile.in: Add profile-count.h and profile-count.o
21353 * auto-profile.c (afdo_indirect_call): Update to new API.
21354 (afdo_set_bb_count): Update.
21355 (afdo_propagate_edge): Update.
21356 (afdo_propagate_circuit): Update.
21357 (afdo_calculate_branch_prob): Update.
21358 (afdo_annotate_cfg): Update.
21359 * basic-block.h: Include profile-count.h
21360 (struct edge_def): Turn count to profile_count.
21361 (struct basic_block_def): Likewie.
21362 (REG_BR_PROB_BASE): Move to profile-count.h
21363 (RDIV): Move to profile-count.h
21364 * bb-reorder.c (max_entry_count): Turn to profile_count.
21365 (find_traces): Update.
21366 (rotate_loop):Update.
21367 (connect_traces):Update.
21368 (sanitize_hot_paths):Update.
21369 * bt-load.c (migrate_btr_defs): Update.
21370 * cfg.c (RDIV): Remove.
21371 (init_flow): Use alloc_block.
21372 (alloc_block): Uninitialize count.
21373 (unchecked_make_edge): Uninitialize count.
21374 (check_bb_profile): Update.
21375 (dump_edge_info): Update.
21376 (dump_bb_info): Update.
21377 (update_bb_profile_for_threading): Update.
21378 (scale_bbs_frequencies_int): Update.
21379 (scale_bbs_frequencies_gcov_type): Update.
21380 (scale_bbs_frequencies_profile_count): New.
21381 * cfg.h (update_bb_profile_for_threading): Update.
21382 (scale_bbs_frequencies_profile_count): Declare.
21383 * cfgbuild.c (compute_outgoing_frequencies): Update.
21384 (find_many_sub_basic_blocks): Update.
21385 * cfgcleanup.c (try_forward_edges): Update.
21386 (try_crossjump_to_edge): Update.
21387 * cfgexpand.c (expand_gimple_tailcall): Update.
21388 (construct_exit_block): Update.
21389 * cfghooks.c (verify_flow_info): Update.
21390 (dump_bb_for_graph): Update.
21391 (split_edge): Update.
21392 (make_forwarder_block): Update.
21393 (duplicate_block): Update.
21394 (account_profile_record): Update.
21395 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21396 (get_estimated_loop_iterations): Update.
21397 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21398 (single_likely_exit): Update.
21399 * cfgloopmanip.c (scale_loop_profile): Update.
21401 (set_zero_probability): Update.
21402 (lv_adjust_loop_entry_edge): Update.
21403 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21404 (purge_dead_edges): Update.
21405 (rtl_account_profile_record): Update.
21406 * cgraph.c (cgraph_node::create): Uninitialize count.
21407 (symbol_table::create_edge): Uninitialize count.
21408 (cgraph_update_edges_for_call_stmt_node): Update.
21409 (cgraph_edge::dump_edge_flags): Update.
21410 (cgraph_node::dump): Update.
21411 (cgraph_edge::maybe_hot_p): Update.
21412 * cgraph.h: Include profile-count.h
21413 (create_clone), create_edge, create_indirect_edge): Update.
21414 (cgraph_node): Turn count to profile_count.
21415 (cgraph_edge0: Likewise.
21416 (make_speculative, clone): Update.
21417 (create_edge): Update.
21418 (init_lowered_empty_function): Update.
21419 * cgraphclones.c (cgraph_edge::clone): Update.
21420 (duplicate_thunk_for_node): Update.
21421 (cgraph_node::create_clone): Update.
21422 * cgraphunit.c (cgraph_node::analyze): Update.
21423 (cgraph_node::expand_thunk): Update.
21424 * final.c (dump_basic_block_info): Update.
21425 * gimple-streamer-in.c (input_bb): Update.
21426 * gimple-streamer-out.c (output_bb): Update.
21427 * graphite.c (print_global_statistics): Update.
21428 (print_graphite_scop_statistics): Update.
21429 * hsa-brig.c: Include basic-block.h.
21430 * hsa-dump.c: Include basic-block.h.
21431 * hsa-gen.c (T sum_slice): Update.
21432 (convert_switch_statements):Update.
21433 * hsa-regalloc.c: Include basic-block.h.
21434 * ipa-chkp.c (chkp_produce_thunks): Update.
21435 * ipa-cp.c (struct caller_statistics): Update.
21436 (init_caller_stats): Update.
21437 (gather_caller_stats): Update.
21438 (ipcp_cloning_candidate_p): Update.
21439 (good_cloning_opportunity_p): Update.
21440 (get_info_about_necessary_edges): Update.
21441 (dump_profile_updates): Update.
21442 (update_profiling_info): Update.
21443 (update_specialized_profile): Update.
21444 (perhaps_add_new_callers): Update.
21445 (decide_about_value): Update.
21446 (ipa_cp_c_finalize): Update.
21447 * ipa-devirt.c (struct odr_type_warn_count): Update.
21448 (struct decl_warn_count): Update.
21449 (struct final_warning_record): Update.
21450 (possible_polymorphic_call_targets): Update.
21451 (ipa_devirt): Update.
21452 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21453 * ipa-icf.c (sem_function::merge): Update.
21454 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21455 * ipa-inline.c (compute_uninlined_call_time): Update.
21456 (compute_inlined_call_time): Update.
21457 (want_inline_small_function_p): Update.
21458 (want_inline_self_recursive_call_p): Update.
21459 (edge_badness): Update.
21460 (lookup_recursive_calls): Update.
21461 (recursive_inlining): Update.
21462 (inline_small_functions): Update.
21463 (dump_overall_stats): Update.
21464 (dump_inline_stats): Update.
21465 * ipa-profile.c (ipa_profile_generate_summary): Update.
21466 (ipa_propagate_frequency): Update.
21467 (ipa_profile): Update.
21468 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21469 * ipa-utils.c (ipa_merge_profiles): Update.
21470 * loop-doloop.c (doloop_modify): Update.
21471 * loop-unroll.c (report_unroll): Update.
21472 (unroll_loop_runtime_iterations): Update.
21473 * lto-cgraph.c (lto_output_edge): Update.
21474 (lto_output_node): Update.
21475 (input_node): Update.
21476 (input_edge): Update.
21477 (merge_profile_summaries): Update.
21478 * lto-streamer-in.c (input_cfg): Update.
21479 * lto-streamer-out.c (output_cfg): Update.
21480 * mcf.c (create_fixup_graph): Update.
21481 (adjust_cfg_counts): Update.
21482 (sum_edge_counts): Update.
21483 * modulo-sched.c (sms_schedule): Update.
21484 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21485 * predict.c (maybe_hot_count_p): Update.
21486 (probably_never_executed): Update.
21487 (dump_prediction): Update.
21488 (combine_predictions_for_bb): Update.
21489 (propagate_freq): Update.
21490 (handle_missing_profiles): Update.
21491 (counts_to_freqs): Update.
21492 (rebuild_frequencies): Update.
21493 (force_edge_cold): Update.
21494 * predict.h: Include profile-count.h
21495 (maybe_hot_count_p, counts_to_freqs): UPdate.
21496 * print-rtl-function.c: Do not include cfg.h
21497 * print-rtl.c: Include basic-block.h
21498 * profile-count.c: New file.
21499 * profile-count.h: New file.
21500 * profile.c (is_edge_inconsistent): Update.
21501 (correct_negative_edge_counts): Update.
21502 (is_inconsistent): Update.
21503 (set_bb_counts): Update.
21504 (read_profile_edge_counts): Update.
21505 (compute_frequency_overlap): Update.
21506 (compute_branch_probabilities): Update; Initialize and deinitialize
21508 (branch_prob): Update.
21509 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21510 (edge_gcov_count): New.
21511 (bb_gcov_count): New.
21512 * shrink-wrap.c (try_shrink_wrapping): Update.
21513 * tracer.c (better_p): Update.
21514 * trans-mem.c (expand_transaction): Update.
21515 (ipa_tm_insert_irr_call): Update.
21516 (ipa_tm_insert_gettmclone_call): Update.
21517 * tree-call-cdce.c: Update.
21518 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21519 (gimple_duplicate_sese_tail): Update.
21520 (gimple_account_profile_record): Update.
21521 (execute_fixup_cfg): Update.
21522 * tree-inline.c (copy_bb): Update.
21523 (copy_edges_for_bb): Update.
21524 (initialize_cfun): Update.
21525 (freqs_to_counts): Update.
21526 (copy_cfg_body): Update.
21527 (expand_call_inline): Update.
21528 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21529 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21530 (try_unroll_loop_completely): Update.
21531 (try_peel_loop): Update.
21532 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21533 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21534 * tree-ssa-loop-split.c (connect_loops): Update.
21535 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21536 * tree-ssa-reassoc.c (branch_fixup): Update.
21537 * tree-ssa-tail-merge.c (replace_block_by): Update.
21538 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21539 (compute_path_counts): Update.
21540 (update_profile): Update.
21541 (recompute_probabilities): Update.
21542 (update_joiner_offpath_counts): Update.
21543 (estimated_freqs_path): Update.
21544 (freqs_to_counts_path): Update.
21545 (clear_counts_path): Update.
21546 (ssa_fix_duplicate_block_edges): Update.
21547 (duplicate_thread_path): Update.
21548 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21549 (struct switch_conv_info): Update.
21550 * tree-tailcall.c (decrease_profile): Update.
21551 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21552 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21553 * value-prof.c (check_counter): Update.
21554 (gimple_divmod_fixed_value): Update.
21555 (gimple_mod_pow2): Update.
21556 (gimple_mod_subtract): Update.
21557 (gimple_ic_transform): Update.
21558 (gimple_stringop_fixed_value): Update.
21559 * value-prof.h (gimple_ic): Update.
21561 2017-06-02 Carl Love <cel@us.ibm.com>
21563 * config/rs6000/rs6000-c: Add support for built-in functions
21564 vector double vec_doublee (vector signed int);
21565 vector double vec_doublee (vector unsigned int);
21566 vector double vec_doublee (vector float);
21567 vector double vec_doubleh (vector signed int);
21568 vector double vec_doubleh (vector unsigned int);
21569 vector double vec_doubleh (vector float);
21570 vector double vec_doublel (vector signed int);
21571 vector double vec_doublel (vector unsigned int);
21572 vector double vec_doublel (vector float);
21573 vector double vec_doubleo (vector signed int);
21574 vector double vec_doubleo (vector unsigned int);
21575 vector double vec_doubleo (vector float);.
21576 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21577 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21579 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21580 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21581 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21583 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21584 vec_doublel, vec_doubleh.
21585 * doc/extend.texi: Update the built-in documentation file for the
21586 new built-in functions.
21588 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21591 * ipa-inline-analysis.c (free_growth_caches): Set
21592 edge_removal_hook_holder to NULL after removing it.
21594 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21596 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21597 comparision with zero.
21599 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21600 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21601 for early expansion of vec_min and vec_max builtins.
21602 (builtin_function_type): Add min/max unsigned variants to those
21603 identified as having unsigned arguments.
21605 2017-06-02 Olivier Hainque <hainque@adacore.com>
21607 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21609 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21611 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21612 Use VALL_F16 iterator rather than VALL.
21614 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21616 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21617 Emit CBNZ inside loop when doing a strong exchange and comparing
21618 against zero. Generate the CC flags after the loop.
21620 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21622 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21623 (dl_section_ref): New.
21624 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21625 On AIX, append an expression to subtract the size of the
21626 section length to dl_section_ref.
21628 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21630 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21631 for early expansion of vector absolute builtins.
21633 2017-06-02 Richard Biener <rguenther@suse.de>
21635 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21636 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21638 2017-06-02 Richard Biener <rguenther@suse.de>
21640 PR tree-optimization/80948
21641 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21642 stmt order as well.
21644 2017-06-02 Richard Biener <rguenther@suse.de>
21646 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21648 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21649 for inductions as relevant.
21651 2017-06-02 Richard Biener <rguenther@suse.de>
21653 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21654 (vectorizable_induction): ... this. Remove dead code.
21656 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21658 * builtins. (expand_builtin_alloca): Remove second parameter and
21659 infer its value from the first parameter instead.
21660 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21662 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21664 PR rtl-optimization/80903
21665 * loop-doloop.c (add_test): Unshare sequence.
21667 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21669 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21671 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21673 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21675 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21676 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21677 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21678 (xlogue_layout::get_stub_rtx): Make static.
21679 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21680 (xlogue_layout::compute_stub_managed_regs): Rename to...
21681 (xlogue_layout::count_stub_managed_regs): ...this.
21682 (xlogue_layout::is_stub_managed_reg): New function.
21683 (xlogue_layout::m_stub_names): Rename to...
21684 (xlogue_layout::s_stub_names): ...this, make static.
21685 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21686 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21687 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21688 xlogue_layout::s_stub_names): Instantiate statics.
21689 (stub_managed_regs): Remove.
21690 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21691 (disable_call_ms2sysv_xlogues): Rename to...
21692 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21693 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21695 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21696 change after reload_completed.
21697 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21699 (ix86_expand_prologue): Likewise.
21700 (ix86_expand_epilogue): Likewise.
21701 (ix86_expand_split_stack_prologue): Likewise.
21702 (ix86_compute_frame_layout): Remove frame parameter ...
21703 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21704 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21706 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21707 (ix86_frame): Move from here ...
21708 * config/i386/i386.h (ix86_frame): ... to here.
21709 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21710 complete ix86_frame data structure instead. Remove some_ld_name.
21712 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21714 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21715 symbols that hold a DECL_VALUE_EXPR.
21717 2017-06-01 Martin Jambor <mjambor@suse.cz>
21719 PR tree-optimization/80898
21720 * tree-sra.c (process_subtree_disqualification): Removed.
21721 (disqualify_candidate): Do not acll
21722 process_subtree_disqualification.
21723 (subtree_mark_written_and_enqueue): New function.
21724 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21725 RHS has been disqualified and re-queue LHS if necessary. Apart
21726 from that, ignore disqualified RHS.
21728 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21730 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21731 address fetch for z10 or later.
21733 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21735 * config/arc/arc.md (tst_movb): Add guard when splitting.
21737 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21739 * config/arc/arc.c (arc_can_eliminate): Test against
21740 arc_frame_pointer_needed.
21742 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21744 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21745 to prevent store reordering.
21746 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21747 (type): Add block type.
21748 (stack_tie): Define special instruction to be used in
21751 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21753 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21754 constraint. It is not valid for the pattern.
21755 (noncommutative_binary_comparison): Likewise.
21757 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21759 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21762 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21764 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21765 be used by the reg-alloc.
21767 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21769 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21770 reg-alloc when having mul64 or mul32x16 instructions.
21771 (mulsidi3): Likewise.
21772 (umulsidi3): Likewise.
21773 (mulsi32x16): New pattern.
21774 (mulsi64): Likewise.
21775 (mulsidi64): Likewise.
21776 (umulsidi64): Likewise.
21777 (MUL32x16_REG): Define.
21778 (mul64_600): Use MUL32x16_REG.
21779 (mac64_600): Likewise.
21780 (umul64_600): Likewise.
21781 (umac64_600): Likewise.
21783 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21785 * config/arc/arc.md (mulsi3_700): Make it commutative.
21787 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21789 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21791 (*sign_extendsidi2_insn): Likewise for movstosw.
21793 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21795 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21796 the type of the input discriminant value. Convert the
21797 discriminant value of signedness vary.
21799 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21801 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21802 Add to -Wall section.
21804 2017-06-01 Richard Biener <rguenther@suse.de>
21806 PR middle-end/66313
21807 * fold-const.c (fold_plusminus_mult_expr): If the factored
21808 factor may be zero use a wrapping type for the inner operation.
21809 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21810 and handle moved defs.
21811 (process_assignment): Properly guard the unary op case. Return a
21812 tri-state indicating that moving the stmt before the call may allow
21813 to continue. Pass through to_move.
21814 (find_tail_calls): Handle moving unrelated defs before
21817 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21820 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21821 splitter result in the canonical way.
21823 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21825 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21826 also for 32bit target. Update insn attributes.
21827 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21829 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21831 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21832 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21833 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21834 (_mm_maskz_min_ss): New intrinsics.
21836 2017-05-31 Martin Liska <mliska@suse.cz>
21838 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21839 Change comment style to one we normally use.
21840 (vectorizable_reduction): Likewise.
21841 (vectorizable_induction): Likewise.
21842 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21843 (vectorizable_call): Likewise.
21844 (vectorizable_simd_clone_call): Likewise.
21845 (vectorizable_conversion): Likewise.
21846 (vectorizable_assignment): Likewise.
21847 (vectorizable_shift): Likewise.
21848 (vectorizable_operation): Likewise.
21849 (vectorizable_store): Likewise.
21850 (vectorizable_load): Likewise.
21851 * tree-vectorizer.h: Likewise.
21853 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21855 * passes.c (emergency_dump_function): New.
21856 * tree-pass.h (emergency_dump_function): Declare.
21857 * plugin.c (plugins_internal_error_function): Remove.
21858 * plugin.h (plugins_internal_error_function): Remove declaration.
21859 * toplev.c (internal_error_function): New static function. Use it...
21860 (general_init): ...here.
21862 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21864 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21865 (arc_rtx_costs): Add costs for new patterns.
21866 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21867 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21869 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21871 * tree-ssa-strlen.c (get_next_strinfo): New function.
21872 (get_stridx_plus_constant): Use it.
21873 (zero_length_string): Likewise.
21874 (adjust_related_strinfos): Likewise.
21875 (adjust_last_stmt): Likewise.
21877 2017-05-31 Richard Biener <rguenther@suse.de>
21880 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21881 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21883 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21885 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21886 loop_vinfo argument and use of dependence distance vectors.
21887 Check instead whether the two references differ only in their
21888 initial value and assume that they have the same alignment if the
21889 difference is a multiple of the vector alignment.
21890 (vect_analyze_data_refs_alignment): Update call accordingly.
21892 2017-05-31 Martin Liska <mliska@suse.cz>
21895 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21897 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21899 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21900 (create_intersect_range_checks): Move from ...
21901 * tree-data-ref.c (create_intersect_range_checks_index)
21902 (create_intersect_range_checks): ... to here.
21903 (create_runtime_alias_checks): New function factored from ...
21904 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21905 here. Call above function.
21906 * tree-data-ref.h (create_runtime_alias_checks): New function.
21908 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21910 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21911 segment length for dr_b and compute it in wide_int.
21913 2017-05-31 Richard Biener <rguenther@suse.de>
21915 PR tree-optimization/80906
21916 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21917 and pass through iv_map.
21918 (copy_bb_and_scalar_dependences): Adjust.
21919 (translate_pending_phi_nodes): Likewise.
21920 (copy_loop_close_phi_args): Handle code-generating IVs instead
21923 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21925 * diagnostic-color.c (color_dict): Add "type-diff".
21926 (parse_gcc_colors): Update comment.
21927 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21928 -fdiagnostics-show-template-tree and -fno-elide-type.
21929 (GCC_COLORS): Add type-diff to example.
21931 (-fdiagnostics-show-template-tree): New.
21932 (-fno-elide-type): New.
21933 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21934 the pp_format_decoder callback. Call any m_format_postprocessor's
21936 (pretty_printer::pretty_printer): Initialize
21937 m_format_postprocessor.
21938 (pretty_printer::~pretty_printer): Delete any
21939 m_format_postprocessor.
21940 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21941 (class format_postprocessor): New class.
21942 (struct pretty_printer::format_decoder): Document the new parameters.
21943 (struct pretty_printer::m_format_postprocessor): New field.
21944 * tree-diagnostic.c (default_tree_printer): Update for new
21945 bool and const char ** params.
21946 * tree-diagnostic.h (default_tree_printer): Likewise.
21948 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21950 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21951 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21952 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21953 rs6000_gen_cell_microcode code.
21954 (rs6000_final_prescan_insn): Delete.
21955 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21956 "warn-cell-microcode" entries.
21957 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21958 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21959 throughout. Change cc_reg_not_micro_cr0_operand to
21960 cc_reg_not_cr0_operand throughout.
21961 (*extendhi<mode>2_noload): Delete.
21962 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21963 (mwarn-cell-microcode): Delete.
21964 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21965 -mgen-cell-microcode and -mwarn-cell-microcode.
21967 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21970 * config/i386/constraints.md (Yd): New constraint.
21972 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21973 and (?Yd, r) alternatives. Update insn attributes.
21974 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21975 and (?*Yd, r) alternatives. Update insn attributes.
21976 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21978 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21980 * gimplify.c (gimplify_modify_expr): Don't create a
21981 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21984 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21986 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21988 2017-05-30 Richard Biener <rguenther@suse.de>
21990 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21991 and reduc_def fields.
21992 (STMT_VINFO_REDUC_TYPE): New define.
21993 (STMT_VINFO_REDUC_DEF): Likewise.
21994 (vect_force_simple_reduction): Adjust prototype.
21995 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21996 (vect_is_simple_reduction): Remove check_reduction argument.
21997 (vect_force_simple_reduction): Adjust and set
21998 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21999 (vectorizable_reduction): Do not re-do reduction analysis
22000 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22001 * tree-parloops.c (gather_scalar_reductions): Adjust.
22003 2017-05-30 Richard Biener <rguenther@suse.de>
22005 PR middle-end/80901
22006 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22009 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22011 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22012 Introduce unknown_misalignment parameter and remove vf.
22013 (vect_peeling_hash_get_lowest_cost):
22014 Pass unknown_misalignment parameter.
22015 (vect_enhance_data_refs_alignment):
22016 Fix unsupportable data ref treatment.
22018 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22020 * tree-vect-data-refs.c (vect_get_data_access_cost):
22021 Workaround for SLP handling.
22022 (vect_enhance_data_refs_alignment):
22023 Compute costs for doing no peeling at all, compare to the best
22024 peeling costs so far and avoid peeling if cheaper.
22026 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22028 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22029 Return peeling info and set costs to zero for unlimited cost
22031 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22032 with unknown misalignment. Compute and costs for unknown
22033 misalignment, compare them to the costs for known misalignment
22034 and choose the cheapest for peeling.
22036 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22038 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22039 (vect_get_peeling_costs_all_drs): Create function.
22040 (vect_peeling_hash_get_lowest_cost):
22041 Use vect_get_peeling_costs_all_drs.
22042 (vect_peeling_supportable): Create function.
22043 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22045 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22047 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22048 DR_HAS_NEGATIVE_STEP.
22049 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22050 (vect_enhance_data_refs_alignment): Use.
22051 (vect_duplicate_ssa_name_ptr_info): Use.
22052 * tree-vectorizer.h (dr_misalignment): Use.
22053 (known_alignment_for_access_p): Use.
22055 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22058 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22060 (has_section_name): New function.
22062 2017-05-30 Martin Liska <mliska@suse.cz>
22065 * auto-profile.c (get_function_decl_from_block): Fix
22068 2017-05-30 Richard Biener <rguenther@suse.de>
22070 PR middle-end/80876
22071 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22073 2017-05-30 Martin Liska <mliska@suse.cz>
22075 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22076 * dumpfile.h (struct dump_file_info): Remove ctors.
22078 2017-05-30 Martin Liska <mliska@suse.cz>
22080 * predict.def: Fix GNU coding style.
22082 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22084 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22085 Mark 'to' argument with ATTRIBUTE_UNUSED.
22087 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22089 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22090 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22091 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22094 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22096 * doc/install.texi (Options specification): Restore entry of
22097 --enable-sjlj-exceptions.
22099 2017-05-27 Michael Eager <eager@eagercon.com>
22102 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22104 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22106 * config/microblaze/microblaze.h
22107 (FIXED_REGISTERS): Update in macro.
22108 (CALL_USED_REGISTERS): Update in macro.
22110 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22112 * doc/install.texi: Add links to macOS binary distributions.
22114 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22118 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22120 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22122 2017-05-26 Martin Liska <mliska@suse.cz>
22124 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22126 2017-05-26 Martin Liska <mliska@suse.cz>
22128 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22129 always leading ';; '.
22130 (dump_bb_info): Likewise.
22131 (brief_dump_cfg): Likewise.
22132 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22133 * dumpfile.c: Remove usage of TDF_VERBOSE.
22134 * dumpfile.h (enum dump_kind): Likewise.
22135 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22136 * print-tree.c (debug_verbose): Remove.
22137 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22138 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22139 * tree-diagnostic.c (default_tree_printer): Replace
22140 TDF_DIAGNOSTIC with TDF_SLIM.
22142 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22144 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22145 in parameter loop, rather than loop_vinfo.
22146 (create_intersect_range_checks): Ditto.
22147 (vect_create_cond_for_alias_checks): Update call to above functions.
22149 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22151 PR tree-optimization/80815
22152 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22153 for merging runtime alias checks. Handle negative DR_STEPs.
22155 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22157 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22159 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22160 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22161 out code pruning runtime alias checks.
22162 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22163 factored out from above.
22164 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22166 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22168 (prune_runtime_alias_test_list): New decalaration.
22170 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22172 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22173 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22174 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22175 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22176 (operator==, comp_dr_with_seg_len_pair): Ditto.
22177 (vect_prune_runtime_alias_test_list): Ditto.
22179 2017-05-26 Martin Liska <mliska@suse.cz>
22182 * params.def: Bound partial-inlining-entry-probability param.
22184 2017-05-26 Marek Polacek <polacek@redhat.com>
22187 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22190 2017-05-26 Richard Biener <rguenther@suse.de>
22192 PR tree-optimization/80842
22193 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22196 2017-05-26 Richard Biener <rguenther@suse.de>
22198 PR tree-optimization/80844
22199 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22201 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22203 * doc/md.texi (Machine Constraints): Update x86 family
22204 machine constraints section to match 'config/i386/constraints.md'.
22206 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22208 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22210 2017-05-25 Nathan Sidwell <nathan@acm.org>
22212 * doc/invoke.texi (--enable-languages): Update documentation.
22214 2017-05-25 Martin Liska <mliska@suse.cz>
22216 * dumpfile.c: Add TDF_FOLDING.
22217 * dumpfile.h (enum dump_kind): Likewise.
22218 * genmatch.c (dt_simplify::gen_1): Use it.
22220 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22222 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22224 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22226 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22227 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22229 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22231 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22232 * match.pd (X == C): Rewrite it here.
22233 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22234 with_certain_nonzero_bits2): New predicates.
22235 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22237 2017-05-24 Nathan Sidwell <nathan@acm.org>
22239 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22242 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22245 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22247 * config/powerpcspe: New port. Files are copied from the rs6000
22248 port, with "rs6000" in filenames replaced by "powerpcspe".
22250 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22252 PR rtl-optimization/80754
22253 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22255 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22257 * config/sparc/sparc.md (length): Return the correct value for -mflat
22258 sibcalls to match output_sibcall.
22260 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22264 * config/rs6000/rs6000.c (struct machine_function): Add new field
22266 (rs6000_get_separate_components): Init that field, use it.
22267 (rs6000_components_for_bb): Use the field.
22269 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22271 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22273 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22275 PR middle-end/80823
22276 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22278 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22281 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22282 address against address_operand predicate.
22283 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22285 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22287 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22288 parameters passed indirectly.
22290 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22292 * config/i386/i386.md (*movdi_internal): Remove SSE4
22293 alternative 18 (?r, *v). Update insn attributes.
22294 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22295 Update insn attributes.
22296 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22297 Update insn attributes.
22298 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22299 alternative 1 (r, v). Remove isa attribute.
22300 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22301 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22302 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22304 2017-05-23 Tom de Vries <tom@codesourcery.com>
22306 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22309 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22311 * cgraphunit.c (symbol_table::process_new_functions): Update.
22312 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22313 (inline_generate_summary): Rename to ...
22314 (ipa_fn_summary_generate): ... this one.
22315 (inline_read_summary): Rename to ...
22316 (ipa_fn_summary_read): ... this one.
22317 (inline_write_summary): Rename to ...
22318 (ipa_fn_summary_write): ... this one.
22319 (inline_free_summary): Rename to ...
22320 (ipa_free_fn_summary): ... this one.
22321 (pass_data_local_fn_summary, pass_local_fn_summary,
22322 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22323 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22324 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22325 make_pass_ipa_fn_summary): New.
22326 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22327 inline_write_summary, inline_free_summary): Remove.
22328 (ipa_free_fn_summary) : New.
22329 * ipa-inline.c (ipa_inline): Update.
22330 (pass_ipa_inline): Do not generate summaries.
22331 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22333 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22334 and add pass_ipa_fn_summary.
22335 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22337 (make_pass_inline_parameters): Remove.
22339 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22341 * omp-low.c (struct omp_context): Remove "default_kind" member.
22344 * omp-offload.c (execute_oacc_device_lower): Remove the
22345 parallelism dimensions function attributes for unparallelized
22346 OpenACC kernels constructs.
22348 2017-05-23 Martin Liska <mliska@suse.cz>
22350 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22352 (cgraph_edge::make_speculative): Likewise.
22353 (cgraph_edge::resolve_speculation): Likewise.
22354 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22355 (cgraph_node::dump): Likewise.
22356 * cgraph.h: Likewise.
22357 * cgraphunit.c (analyze_functions): Likewise.
22358 (symbol_table::compile): Likewise.
22359 * ipa-cp.c (print_all_lattices): Likewise.
22360 (determine_versionability): Likewise.
22361 (initialize_node_lattices): Likewise.
22362 (ipcp_verify_propagated_values): Likewise.
22363 (estimate_local_effects): Likewise.
22364 (update_profiling_info): Likewise.
22365 (create_specialized_node): Likewise.
22366 (perhaps_add_new_callers): Likewise.
22367 (decide_about_value): Likewise.
22368 (decide_whether_version_node): Likewise.
22369 (identify_dead_nodes): Likewise.
22370 (ipcp_store_bits_results): Likewise.
22371 * ipa-devirt.c (dump_targets): Likewise.
22372 (ipa_devirt): Likewise.
22373 * ipa-icf.c (sem_item::dump): Likewise.
22374 (sem_function::equals): Likewise.
22375 (sem_variable::equals): Likewise.
22376 (sem_item_optimizer::read_section): Likewise.
22377 (sem_item_optimizer::execute): Likewise.
22378 (congruence_class::dump): Likewise.
22379 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22380 (dump_inline_summary): Likewise.
22381 (estimate_node_size_and_time): Likewise.
22382 (inline_analyze_function): Likewise.
22383 * ipa-inline-transform.c (inline_call): Likewise.
22384 * ipa-inline.c (report_inline_failed_reason): Likewise.
22385 (want_early_inline_function_p): Likewise.
22386 (edge_badness): Likewise.
22387 (update_edge_key): Likewise.
22388 (inline_small_functions): Likewise.
22389 * ipa-profile.c (ipa_profile): Likewise.
22390 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22391 (ipa_make_edge_direct_to_target): Likewise.
22392 (remove_described_reference): Likewise.
22393 (ipa_impossible_devirt_target): Likewise.
22394 (propagate_controlled_uses): Likewise.
22395 (ipa_print_node_params): Likewise.
22396 (ipcp_transform_function): Likewise.
22397 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22398 (propagate_pure_const): Likewise.
22399 * ipa-reference.c (generate_summary): Likewise.
22400 (read_write_all_from_decl): Likewise.
22401 (propagate): Likewise.
22402 (ipa_reference_read_optimization_summary): Likewise.
22403 * ipa-utils.c (ipa_merge_profiles): Likewise.
22404 * ipa.c (walk_polymorphic_call_targets): Likewise.
22405 (symbol_table::remove_unreachable_nodes): Likewise.
22406 (ipa_single_use): Likewise.
22407 * passes.c (execute_todo): Likewise.
22408 * predict.c (drop_profile): Likewise.
22409 * symtab.c (symtab_node::get_dump_name): New function.
22410 (symtab_node::dump_name): Likewise.
22411 (symtab_node::dump_asm_name): Likewise.
22412 (symtab_node::dump_references): Likewise.
22413 (symtab_node::dump_referring): Likewise.
22414 (symtab_node::dump_base): Likewise.
22415 (symtab_node::debug_symtab): Likewise.
22416 * tree-sra.c (convert_callers_for_node): Likewise.
22417 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22418 * value-prof.c (init_node_map): Likewise.
22420 2017-05-23 Martin Liska <mliska@suse.cz>
22422 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22423 and symtab_node::debug_symtab to symbol_table::debug.
22424 * cgraphunit.c (analyze_functions): Use the renamed function.
22425 (symbol_table::compile): Likewise.
22426 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22427 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22428 * passes.c (execute_todo): Likewise.
22429 * symtab.c (symbol_table::dump): New function.
22430 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22432 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22434 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22435 that nonconst implies exec.
22437 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22439 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22440 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22441 (inline_edge_summary_vec): Turn into ...
22442 (ipa_call_summaries): ... this one.
22443 (redirect_to_unreachable, edge_set_predicate,
22444 evaluate_properties_for_edge, inline_summary_alloc,
22445 reset_ipa_call_summary, reset_inline_summary,
22446 inline_summary_t::duplicate): Update.
22447 (inline_edge_duplication_hook): Turn to ...
22448 (ipa_call_summary_t::duplicate): ... this one.
22449 (inline_edge_removal_hook): Turn to ...
22450 (ipa_call_summary_t::remove): ... this one.
22451 (dump_inline_edge_summary): Turn to ...
22452 (dump_ipa_call_summary): ... this one.
22453 (estimate_function_body_sizes): Update.
22454 (inline_update_callee_summaries): Update.
22455 (remap_edge_change_prob): Update.
22456 (remap_edge_summaries): Update.
22457 (inline_merge_summary): Update.
22458 (do_estimate_edge_time): Update.
22459 (inline_generate_summary): Update.
22460 (inline_read_section): Update.
22461 (inline_read_summary): Update.
22462 (inline_free_summary): Update.
22463 * ipa-inline.c (can_inline_edge_p): Update.
22464 (compute_inlined_call_time): Update.
22465 (want_inline_small_function_p): Update.
22466 (edge_badness): Update.
22467 (early_inliner): Update.
22468 * ipa-inline.h (inline_edge_summary): Turn to ...
22469 (ipa_call_summary): ... this one.
22470 (ipa_call_summary_t): New class.
22471 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22472 (ipa_call_summaries): New.
22473 (inline_edge_summary): Remove.
22474 (estimate_edge_growth): Update.
22475 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22476 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22477 * ipa-split.c (execute_split_functions): Update.
22478 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22480 2017-05-23 Tom de Vries <tom@codesourcery.com>
22482 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22483 attributes): Document rdrand effective target.
22485 2017-05-23 Tom de Vries <tom@codesourcery.com>
22487 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22488 attributes): Sort alphabetically.
22490 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22492 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22494 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22497 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22498 V2DF/V2DI splat into two separate patterns, one that handles
22499 registers, and the other that only handles memory. Drop support
22500 for splatting from a GPR on ISA 2.07 and then splitting the
22501 splat into direct move and splat.
22502 (vsx_splat_<mode>_reg): Likewise.
22503 (vsx_splat_<mode>_mem): Likewise.
22505 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22507 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22509 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22511 PR middle-end/80809
22512 * omp-low.c (finish_taskreg_remap): New function.
22513 (finish_taskreg_scan): If unit size of ctx->record_type
22514 is non-constant, unshare the size expression and replace
22515 decls in it with possible outer var refs.
22517 PR middle-end/80809
22518 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22519 GOVD_SHARED rather than GOVD_PRIVATE with it.
22520 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22521 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22523 PR middle-end/80853
22524 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22525 as last argument to build_outer_var_ref for pointer bases of array
22526 section reductions.
22528 2017-05-19 Martin Sebor <msebor@redhat.com>
22530 * print-tree.c (print_node): Print DECL_READ_P flag.
22532 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22534 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22535 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22536 * cgraph.c: Likewise.
22537 * cgraphunit.c: Likewise.
22538 * gengtype.c: Likewise.
22539 * ipa-cp.c: Likewise.
22540 * ipa-devirt.c: Likewise.
22541 * ipa-icf.c: Likewise.
22542 * ipa-predicate.c: Likewise.
22543 * ipa-profile.c: Likewise.
22544 * ipa-prop.c: Likewise.
22545 * ipa-split.c: Likewise.
22547 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22548 edge_predicate_pool, dump_inline_hints,
22549 inline_summary::account_size_time, redirect_to_unreachable,
22550 edge_set_predicate, set_hint_predicate,
22551 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22552 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22553 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22554 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22555 ipa_call_summary_t::remove, initialize_growth_caches,
22556 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22557 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22558 mark_modified, unmodified_parm_1, unmodified_parm,
22559 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22560 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22561 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22562 will_be_nonconstant_predicate, record_modified_bb_info,
22563 get_minimal_bb, record_modified, param_change_prob,
22564 phi_result_unknown_predicate, predicate_for_phi_result,
22565 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22566 estimate_function_body_sizes, compute_inline_parameters,
22567 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22568 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22569 inline_update_callee_summaries, remap_edge_change_prob,
22570 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22571 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22572 inline_analyze_function, inline_summary_t::insert,
22573 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22574 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22575 inline_free_summary): Move to ipa-fnsummary.h
22576 (predicate_t): Remove.
22577 * ipa-fnsummary.c: New file.
22578 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22579 (enum inline_hints_vals, inline_hints, agg_position_info,
22580 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22581 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22582 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22583 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22584 inline_read_summary, inline_write_summary, inline_free_summary,
22585 inline_analyze_function, initialize_inline_failed,
22586 inline_merge_summary, inline_update_overall_summary,
22587 compute_inline_parameters): Move to ipa-fnsummary.h
22588 * ipa-fnsummary.h: New file.
22589 * ipa-inline-transform.h: Include ipa-inline.h.
22590 * ipa-inline.c: LIkewise.
22592 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22594 * ipa-inline.c (edge_badness): Use inlined_time instead of
22595 inline_summaries->get.
22597 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22599 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22601 2017-05-22 Nathan Sidwell <nathan@acm.org>
22603 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22604 (fdump-lang): Document 'raw' option.
22605 * dumpfile.h (TDI_tu): Delete.
22606 * dumpfile.c (dump_files): Remove translation-unit.
22607 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22609 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22611 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22612 command option from $(AWK) call.
22613 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22614 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22615 [FORMAT]: Remove handling of variable.
22616 * config/avr/t-multilib: Regenerate.
22618 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22620 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22622 (dump_inline_summary): Do not print self_time.
22623 (estimate_function_body_sizes): Do not set self_time.
22624 (compute_inline_parameters): Likewise.
22625 (inline_read_section, inline_write_summary): Do not stream self_time.
22626 * ipa-inline.h (inline_summary): Drop self_time.
22628 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22630 * ipa-inline-analysis.c (account_size_time): Rename to ...
22631 (inline_summary::account_size_time): ... this one.
22632 (reset_ipa_call_summary): Turn to ...
22633 (ipa_call_summary::reset): ... this one.
22634 (reset_inline_summary): Turn to ...
22635 (inline_summary::reset): ... this one.
22636 (inline_summary_t::remove): Update.
22637 (inline_summary_t::duplicate): Update.
22638 (ipa_call_summary_t::remove): Update.
22639 (dump_inline_summary): Update.
22640 (estimate_function_body_sizes): Update.
22641 (compute_inline_parameters): Update.
22642 (estimate_node_size_and_time): Update.
22643 (inline_merge_summary): Update.
22644 (inline_update_overall_summary): Update.
22645 (inline_read_section): Update.
22646 (inline_write_summary): Update.
22647 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22648 add account_size_time and reset member functions.
22649 (ipa_call_summary): Add reset function.
22650 * ipa-predicate.h (predicate::operator &): Constify.
22652 2017-05-22 Richard Biener <rguenther@suse.de>
22654 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22656 2017-05-19 Jason Merrill <jason@redhat.com>
22658 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22660 2017-05-19 Marek Polacek <polacek@redhat.com>
22663 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22664 TYPE_OVERFLOW_WRAPS checks.
22666 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22668 * tree-core.h (enum omp_clause_default_kind): Add
22669 "OMP_CLAUSE_DEFAULT_PRESENT".
22670 * tree-pretty-print.c (dump_omp_clause): Handle it.
22671 * gimplify.c (enum gimplify_omp_var_data): Add
22672 "GOVD_MAP_FORCE_PRESENT".
22673 (gimplify_adjust_omp_clauses_1): Map it to
22674 "GOMP_MAP_FORCE_PRESENT".
22675 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22677 * gimplify.c (oacc_default_clause): Clarify.
22679 2017-05-19 Nathan Sidwell <nathan@acm.org>
22681 LANG_HOOK_REGISTER_DUMPS
22682 * toplev.c (general_init): Call register dump lang hook.
22683 * doc/invoke.texi: Document -fdump-lang option family.
22684 * dumpfile.c (dump_files): Remove class dump here.
22685 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22686 * dumpfile.h (tree_dump_index): Remove TDI_class.
22687 * langhooks-def.h (lhd_register_dumps): Declare.
22688 (LANG_HOOKS_REGISTER_DUMPS): Define.
22689 (LANG_HOOKS_INITIALIZER): Add it.
22690 * langhooks.c (lhd_register_dumps): Define.
22691 * langhooks.h (struct lang_hooks): Add register_dumps.
22693 2017-05-19 Nathan Sidwell <nathan@acm.org>
22695 * context.h (context::set_passes): New.
22696 * context.c (context::context): Do not create pass manager.
22697 * toplev.c (general_init): Create pass manager here.
22699 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22701 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22702 use this splitter if two add or or instructions would also work for
22703 the constant we want to generate.
22705 2017-05-19 Richard Biener <rguenther@suse.de>
22708 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22709 predicate evaluation.
22711 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22713 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22715 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22718 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22720 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22721 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22722 (inline_edge_summary_vec): Turn into ...
22723 (ipa_call_summaries): ... this one.
22724 (redirect_to_unreachable, edge_set_predicate,
22725 evaluate_properties_for_edge, inline_summary_alloc,
22726 reset_ipa_call_summary, reset_inline_summary,
22727 inline_summary_t::duplicate): Update.
22728 (inline_edge_duplication_hook): Turn to ...
22729 (ipa_call_summary_t::duplicate): ... this one.
22730 (inline_edge_removal_hook): Turn to ...
22731 (ipa_call_summary_t::remove): ... this one.
22732 (dump_inline_edge_summary): Turn to ...
22733 (dump_ipa_call_summary): ... this one.
22734 (estimate_function_body_sizes): Update.
22735 (inline_update_callee_summaries): Update.
22736 (remap_edge_change_prob): Update.
22737 (remap_edge_summaries): Update.
22738 (inline_merge_summary): Update.
22739 (do_estimate_edge_time): Update.
22740 (inline_generate_summary): Update.
22741 (inline_read_section): Update.
22742 (inline_read_summary): Update.
22743 (inline_free_summary): Update.
22744 * ipa-inline.c (can_inline_edge_p): Update.
22745 (compute_inlined_call_time): Update.
22746 (want_inline_small_function_p): Update.
22747 (edge_badness): Update.
22748 (early_inliner): Update.
22749 * ipa-inline.h (inline_edge_summary): Turn to ...
22750 (ipa_call_summary): ... this one.
22751 (ipa_call_summary_t): New class.
22752 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22753 (ipa_call_summaries): New.
22754 (inline_edge_summary): Remove.
22755 (estimate_edge_growth): Update.
22756 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22757 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22758 * ipa-split.c (execute_split_functions): Update.
22759 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22761 2017-05-19 Richard Biener <rguenther@suse.de>
22763 PR middle-end/80764
22764 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22766 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22768 * config/rs6000/rs6000.c (struct machine_function): Add field
22769 fpr_is_wrapped_separately.
22770 (rs6000_get_separate_components): Use 64 components. Handle the
22771 new FPR components.
22772 (rs6000_components_for_bb): Handle the FPR components.
22773 (rs6000_emit_prologue_components): Handle the FPR components.
22774 (rs6000_emit_epilogue_components): Handle the FPR components.
22775 (rs6000_set_handled_components): Handle the FPR components.
22776 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22777 that are already separately shrink-wrapped.
22778 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22779 that are already separately shrink-wrapped.
22781 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22784 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22787 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22788 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22789 cases where the register allocator uses a move and an offsettable
22790 memory operation to/from a FPR register on ISA 2.06/2.07.
22791 (define_peephole2 for Altivec d-form store): Likewise.
22793 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22796 * config/i386/mmx.md (*mov<mode>_internal): Enable
22797 alternatives 11, 12, 13 and 14 also for 32bit targets.
22798 Remove alternatives 15, 16, 17 and 18.
22799 * config/i386/sse.md (vec_concatv2di): Change
22800 alternative (!x, *y) to (x, ?!*Yn).
22802 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22804 * dumpfile.h (enum dump_kind): Remove stray comma.
22806 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22808 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22809 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22810 predicate::num_conditions
22811 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22812 (CHANGED): turn into predicate::changed.
22813 (agg_position_info): Move to ipa-predicate.h
22814 (add_condition, predicate::add_clause, predicate::operator &=,
22815 predicate::or_with, predicate::evaluate, predicate::probability,
22816 dump_condition, dump_clause, predicate::dump,
22817 predicate::remap_after_duplication, predicate::remap_after_inlining,
22818 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22819 (evaluate_conditions_for_known_args): Update.
22820 (set_cond_stmt_execution_predicate): Update.
22821 * ipa-inline.h: Include ipa-predicate.h
22822 (condition, inline_param_summary, conditions, agg_position_info,
22823 predicate): Move to ipa-predicate.h
22824 * ipa-predicate.c: New file.
22825 * ipa-predicate.h: New file.
22827 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22829 * final.c (leaf_function_p): Check we are not in a sequence.
22831 2017-05-18 Martin Liska <mliska@suse.cz>
22833 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22834 * dumpfile.c (dump_register): Use new enum dump_kind.
22835 (get_dump_file_name): Likewise.
22836 (dump_enable_all): Likewise.
22837 (dump_switch_p_1): Likewise.
22838 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22839 * dumpfile.h (enum dump_kind): New enum type.
22840 (struct dump_file_info): Create constructor and
22841 format fields and comments.
22842 * passes.c (pass_manager::register_one_dump_file):
22844 * statistics.c (statistics_early_init): Likewise.
22845 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22846 TDF_TREE with TDF_SLIM.
22847 (gather_memory_references_ref): Likewise.
22849 2017-05-18 Martin Liska <mliska@suse.cz>
22851 * vec.h (struct vnull): Use it.
22853 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22855 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22856 (true_predicate, false_predicate, true_predicate_p,
22857 false_predicate_p): Remove.
22858 (single_cond_predicate, not_inlined_predicate): Turn to member function
22860 (add_condition): Update.
22861 (add_clause): Turn to...
22862 (predicate::add_clause): ... this one; update; allow passing NULL
22864 (and_predicates): Turn to ...
22865 (predicate::operator &=): ... this one.
22866 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22867 (or_predicates): Turn to ...
22868 (predicate::or_with): ... this one.
22869 (evaluate_predicate): Turn to ...
22870 (predicate::evaluate): ... this one.
22871 (predicate_probability): Turn to ...
22872 (predicate::probability): ... this one.
22873 (dump_condition): Update.
22874 (dump_predicate): Turn to ...
22875 (predicate::dump): ... this one.
22876 (account_size_time): Update.
22877 (edge_set_predicate): Update.
22878 (set_hint_predicate): UPdate.
22879 (evaluate_conditions_for_known_args): Update.
22880 (evaluate_properties_for_edge): Update.
22881 (remap_predicate_after_duplication): Turn to...
22882 (predicate::remap_after_duplication): ... this one.
22883 (remap_hint_predicate_after_duplication): Update.
22884 (inline_summary_t::duplicate): UPdate.
22885 (dump_inline_edge_summary): Update.
22886 (dump_inline_summary): Update.
22887 (set_cond_stmt_execution_predicate): Update.
22888 (set_switch_stmt_execution_predicate): Update.
22889 (compute_bb_predicates): Update.
22890 (will_be_nonconstant_expr_predicate): Update.
22891 (will_be_nonconstant_predicate): Update.
22892 (phi_result_unknown_predicate): Update.
22893 (predicate_for_phi_result): Update.
22894 (array_index_predicate): Update.
22895 (estimate_function_body_sizes): Update.
22896 (estimate_node_size_and_time): Update.
22897 (estimate_ipcp_clone_size_and_time): Update.
22898 (remap_predicate): Rename to ...
22899 (predicate::remap_after_inlining): ... this one.
22900 (remap_hint_predicate): Update.
22901 (inline_merge_summary): Update.
22902 (inline_update_overall_summary): Update.
22903 (estimate_size_after_inlining): Update.
22904 (read_predicate): Rename to ...
22905 (predicate::stream_in): ... this one.
22906 (read_inline_edge_summary): Update.
22907 (write_predicate): Rename to ...
22908 (predicate::stream_out): ... this one.
22909 (write_inline_edge_summary): Update.
22910 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22911 (clause_t): Turn to uint32_t
22912 (predicate): Turn to class; implement constructor and operators
22914 (size_time_entry): Update.
22915 (inline_summary): Update.
22916 (inline_edge_summary): Update.
22918 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22920 * fold-const.c (fold_binary_loc): Move transformation...
22921 * match.pd (C - X CMP X): ... here.
22923 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22925 * config/sparc/sparc.c (sparc_option_override): Set function
22926 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22927 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22929 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22931 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22933 2017-05-18 Marek Polacek <polacek@redhat.com>
22936 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22937 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22938 gimple_assign_load_p.
22940 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22942 PR middle-end/80692
22943 * real.c (do_compare): Give decimal_do_compare preference over
22944 comparing just the signs.
22946 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22948 * doc/md.texi (Canonicalization of Instructions): Describe the
22949 canonical form of instructions that inherently set a condition
22952 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22954 PR middle-end/80775
22955 * tree-cfg.c: Move deletion of unreachable case statements to after
22956 the merging of consecutive case labels.
22958 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22960 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22961 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22962 restoring of callee-saved registers.
22964 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22966 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22967 * config/visium/visium.c (single_set_and_flags): Likewise.
22968 * config/visium/visium.md (Substitutions): Likewise.
22970 2017-05-17 Martin Liska <mliska@suse.cz>
22972 * cfg.c: Introduce dump_flags_t type and
22973 use it instead of int type.
22975 * cfghooks.c: Likewise.
22976 * cfghooks.h (struct cfg_hooks): Likewise.
22977 * cfgrtl.c: Likewise.
22978 * cfgrtl.h: Likewise.
22979 * cgraph.c (cgraph_node::get_body): Likewise.
22980 * coretypes.h: Likewise.
22981 * domwalk.c: Likewise.
22982 * domwalk.h: Likewise.
22983 * dumpfile.c (struct dump_option_value_info): Likewise.
22984 (dump_enable_all): Likewise.
22985 (dump_switch_p_1): Likewise.
22986 (opt_info_switch_p): Likewise.
22987 * dumpfile.h (enum tree_dump_index): Likewise.
22988 (struct dump_file_info): Likewise.
22989 * genemit.c: Likewise.
22990 * generic-match-head.c: Likewise.
22991 * gengtype.c (open_base_files): Likewise.
22992 * gimple-pretty-print.c: Likewise.
22993 * gimple-pretty-print.h: Likewise.
22994 * graph.c (print_graph_cfg): Likewise.
22995 * graphite-scop-detection.c (dot_all_sese): Likewise.
22996 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22997 * loop-unroll.c (report_unroll): Likewise.
22998 * passes.c (pass_manager::register_one_dump_file): Likewise.
22999 * print-tree.c: Likewise.
23000 * statistics.c: Likewise.
23001 * tree-cfg.c: Likewise.
23002 * tree-cfg.h: Likewise.
23003 * tree-dfa.c: Likewise.
23004 * tree-dfa.h: Likewise.
23005 * tree-dump.c (dump_function): Likewise.
23006 * tree-dump.h (struct dump_info): Likewise.
23007 * tree-pretty-print.c: Likewise.
23008 * tree-pretty-print.h: Likewise.
23009 * tree-ssa-live.c: Likewise.
23010 * tree-ssa-live.h: Likewise.
23011 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23012 * tree-vect-loop.c: Likewise.
23013 * tree-vect-slp.c: Likewise.
23015 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23016 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23018 PR tree-optimization/80457
23019 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23020 of all arguments to a statement as scalar_to_vec operations.
23021 (vectorizable_call): Adjust call to vect_model_simple_cost for
23023 (vectorizable_conversion): Likewise.
23024 (vectorizable_assignment): Likewise.
23025 (vectorizable_shift): Likewise.
23026 (vectorizable_operation): Likewise.
23027 (vectorizable_comparison): Likewise.
23028 (vect_is_simple_cond): Record the def types for operands.
23029 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23030 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23031 for statement argument count.
23033 2017-05-16 Carl Love <cel@us.ibm.com>
23035 * config/rs6000/rs6000-c: Add support for built-in functions
23036 vector unsigned long long vec_bperm (vector unsigned long long,
23037 vector unsigned char)
23038 vector signed long long vec_mule (vector signed int,
23040 vector unsigned long long vec_mule (vector unsigned int,
23041 vector unsigned int)
23042 vector signed long long vec_mulo (vector signed int,
23044 vector unsigned long long vec_mulo (vector unsigned int,
23045 vector unsigned int)
23046 vector signed char vec_sldw (vector signed char,
23047 vector signed char,
23049 vector unsigned char vec_sldw (vector unsigned char,
23050 vector unsigned char,
23052 vector signed short vec_sldw (vector signed short,
23053 vector signed short,
23055 vector unsigned short vec_sldw (vector unsigned short,
23056 vector unsigned short,
23058 vector signed int vec_sldw (vector signed int,
23061 vector unsigned int vec_sldw (vector unsigned int,
23062 vector unsigned int,
23064 vector signed long long vec_sldw (vector signed long long,
23065 vector signed long long,
23067 vector unsigned long long vec_sldw (vector unsigned long long,
23068 vector unsigned long long,
23070 * config/rs6000/rs6000-c: Add support for built-in functions
23071 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23072 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23073 * doc/extend.texi: Update the built-in documentation for the
23074 new built-in functions.
23076 2017-05-16 Marek Polacek <polacek@redhat.com>
23080 * tree.c (save_expr): Don't fold the expression.
23082 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23084 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23085 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23086 and (?*y,m). Update insn attributes.
23088 2017-05-16 Martin Liska <mliska@suse.cz>
23090 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23091 flags argument of print_gimple_stmt, print_gimple_expr,
23092 print_generic_stmt and print_generic_expr.
23093 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23094 * coretypes.h: Likewise.
23095 * except.c (dump_eh_tree): Likewise.
23096 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23097 * gimple-pretty-print.h: Likewise.
23098 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23099 (backprop::push_to_worklist): Likewise.
23100 (backprop::pop_from_worklist): Likewise.
23101 (backprop::process_use): Likewise.
23102 (backprop::intersect_uses): Likewise.
23103 (note_replacement): Likewise.
23104 * gimple-ssa-store-merging.c
23105 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23106 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23107 (pass_store_merging::execute): Likewise.
23108 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23109 (ssa_base_cand_dump_callback): Likewise.
23110 (dump_incr_vec): Likewise.
23111 (replace_refs): Likewise.
23112 (replace_mult_candidate): Likewise.
23113 (create_add_on_incoming_edge): Likewise.
23114 (create_phi_basis): Likewise.
23115 (insert_initializers): Likewise.
23116 (all_phi_incrs_profitable): Likewise.
23117 (introduce_cast_before_cand): Likewise.
23118 (replace_one_candidate): Likewise.
23119 * gimplify.c (gimplify_expr): Likewise.
23120 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23121 (set_rename): Likewise.
23122 (rename_uses): Likewise.
23123 (copy_loop_phi_nodes): Likewise.
23124 (add_close_phis_to_merge_points): Likewise.
23125 (copy_loop_close_phi_args): Likewise.
23126 (copy_cond_phi_args): Likewise.
23127 (graphite_copy_stmts_from_block): Likewise.
23128 (translate_pending_phi_nodes): Likewise.
23129 * graphite-poly.c (print_pdr): Likewise.
23130 (dump_gbb_cases): Likewise.
23131 (dump_gbb_conditions): Likewise.
23132 (print_scop_params): Likewise.
23133 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23134 (build_cross_bb_scalars_use): Likewise.
23135 (gather_bbs::before_dom_children): Likewise.
23136 * hsa-dump.c (dump_hsa_immed): Likewise.
23137 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23138 (get_replacement_map): Likewise.
23139 * ipa-inline-analysis.c (dump_condition): Likewise.
23140 (estimate_function_body_sizes): Likewise.
23141 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23142 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23143 * ipa-prop.c (ipa_dump_param): Likewise.
23144 (ipa_print_node_jump_functions_for_edge): Likewise.
23145 (ipa_modify_call_arguments): Likewise.
23146 (ipa_modify_expr): Likewise.
23147 (ipa_dump_param_adjustments): Likewise.
23148 (ipa_dump_agg_replacement_values): Likewise.
23149 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23150 * ipa-pure-const.c (check_stmt): Likewise.
23151 (pass_nothrow::execute): Likewise.
23152 * ipa-split.c (execute_split_functions): Likewise.
23153 * omp-offload.c (dump_oacc_loop_part): Likewise.
23154 (dump_oacc_loop): Likewise.
23155 * trans-mem.c (tm_log_emit): Likewise.
23156 (tm_memopt_accumulate_memops): Likewise.
23157 (dump_tm_memopt_set): Likewise.
23158 (dump_tm_memopt_transform): Likewise.
23159 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23160 (print_loop): Likewise.
23161 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23162 (chkp_gather_checks_info): Likewise.
23163 (chkp_get_check_result): Likewise.
23164 (chkp_remove_check_if_pass): Likewise.
23165 (chkp_use_outer_bounds_if_possible): Likewise.
23166 (chkp_reduce_bounds_lifetime): Likewise.
23167 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23168 (chkp_mark_completed_bounds): Likewise.
23169 (chkp_register_incomplete_bounds): Likewise.
23170 (chkp_mark_invalid_bounds): Likewise.
23171 (chkp_maybe_copy_and_register_bounds): Likewise.
23172 (chkp_build_returned_bound): Likewise.
23173 (chkp_get_bound_for_parm): Likewise.
23174 (chkp_build_bndldx): Likewise.
23175 (chkp_get_bounds_by_definition): Likewise.
23176 (chkp_generate_extern_var_bounds): Likewise.
23177 (chkp_get_bounds_for_decl_addr): Likewise.
23178 * tree-chrec.c (chrec_apply): Likewise.
23179 * tree-data-ref.c (dump_data_reference): Likewise.
23180 (dump_subscript): Likewise.
23181 (dump_data_dependence_relation): Likewise.
23182 (analyze_overlapping_iterations): Likewise.
23183 * tree-inline.c (expand_call_inline): Likewise.
23184 (tree_function_versioning): Likewise.
23185 * tree-into-ssa.c (dump_defs_stack): Likewise.
23186 (dump_currdefs): Likewise.
23187 (dump_names_replaced_by): Likewise.
23188 (dump_update_ssa): Likewise.
23189 (update_ssa): Likewise.
23190 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23191 * tree-parloops.c (build_new_reduction): Likewise.
23192 (try_create_reduction_list): Likewise.
23193 (ref_conflicts_with_region): Likewise.
23194 (oacc_entry_exit_ok_1): Likewise.
23195 (oacc_entry_exit_single_gang): Likewise.
23196 * tree-pretty-print.h: Likewise.
23197 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23198 (get_scalar_evolution): Likewise.
23199 (add_to_evolution): Likewise.
23200 (get_loop_exit_condition): Likewise.
23201 (analyze_evolution_in_loop): Likewise.
23202 (analyze_initial_condition): Likewise.
23203 (analyze_scalar_evolution): Likewise.
23204 (instantiate_scev): Likewise.
23205 (number_of_latch_executions): Likewise.
23206 (gather_chrec_stats): Likewise.
23207 (final_value_replacement_loop): Likewise.
23208 (scev_const_prop): Likewise.
23209 * tree-sra.c (dump_access): Likewise.
23210 (disqualify_candidate): Likewise.
23211 (create_access): Likewise.
23212 (reject): Likewise.
23213 (maybe_add_sra_candidate): Likewise.
23214 (create_access_replacement): Likewise.
23215 (analyze_access_subtree): Likewise.
23216 (analyze_all_variable_accesses): Likewise.
23217 (sra_modify_assign): Likewise.
23218 (initialize_constant_pool_replacements): Likewise.
23219 (find_param_candidates): Likewise.
23220 (decide_one_param_reduction): Likewise.
23221 (replace_removed_params_ssa_names): Likewise.
23222 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23223 * tree-ssa-copy.c (dump_copy_of): Likewise.
23224 (copy_prop_visit_cond_stmt): Likewise.
23225 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23226 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23227 (record_equivalences_from_stmt): Likewise.
23228 * tree-ssa-dse.c (compute_trims): Likewise.
23229 (delete_dead_call): Likewise.
23230 (delete_dead_assignment): Likewise.
23231 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23232 (forward_propagate_into_cond): Likewise.
23233 (pass_forwprop::execute): Likewise.
23234 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23235 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23237 (move_computations_worker): Likewise.
23238 (execute_sm): Likewise.
23239 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23240 (remove_exits_and_undefined_stmts): Likewise.
23241 (remove_redundant_iv_tests): Likewise.
23242 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23243 (adjust_iv_update_pos): Likewise.
23244 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23245 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23246 (value_replacement): Likewise.
23247 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23248 * tree-ssa-pre.c (print_pre_expr): Likewise.
23249 (get_representative_for): Likewise.
23250 (create_expression_by_pieces): Likewise.
23251 (insert_into_preds_of_block): Likewise.
23252 (eliminate_insert): Likewise.
23253 (eliminate_dom_walker::before_dom_children): Likewise.
23254 (eliminate): Likewise.
23255 (remove_dead_inserted_code): Likewise.
23256 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23257 * tree-ssa-reassoc.c (get_rank): Likewise.
23258 (eliminate_duplicate_pair): Likewise.
23259 (eliminate_plus_minus_pair): Likewise.
23260 (eliminate_not_pairs): Likewise.
23261 (undistribute_ops_list): Likewise.
23262 (eliminate_redundant_comparison): Likewise.
23263 (update_range_test): Likewise.
23264 (optimize_range_tests_var_bound): Likewise.
23265 (optimize_vec_cond_expr): Likewise.
23266 (rewrite_expr_tree): Likewise.
23267 (rewrite_expr_tree_parallel): Likewise.
23268 (linearize_expr): Likewise.
23269 (break_up_subtract): Likewise.
23270 (linearize_expr_tree): Likewise.
23271 (attempt_builtin_powi): Likewise.
23272 (attempt_builtin_copysign): Likewise.
23273 (transform_stmt_to_copy): Likewise.
23274 (transform_stmt_to_multiply): Likewise.
23275 (dump_ops_vector): Likewise.
23276 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23277 (print_scc): Likewise.
23278 (set_ssa_val_to): Likewise.
23279 (visit_reference_op_store): Likewise.
23280 (visit_use): Likewise.
23281 (sccvn_dom_walker::before_dom_children): Likewise.
23282 (run_scc_vn): Likewise.
23283 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23285 (expr_hash_elt::print): Likewise.
23286 (const_and_copies::pop_to_marker): Likewise.
23287 (const_and_copies::record_const_or_copy_raw): Likewise.
23288 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23289 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23290 (dump_predicates): Likewise.
23291 (find_uninit_use): Likewise.
23292 (warn_uninitialized_phi): Likewise.
23293 (pass_late_warn_uninitialized::execute): Likewise.
23294 * tree-ssa.c (verify_vssa): Likewise.
23295 (verify_ssa): Likewise.
23296 (maybe_optimize_var): Likewise.
23297 * tree-vrp.c (dump_value_range): Likewise.
23298 (dump_all_value_ranges): Likewise.
23299 (dump_asserts_for): Likewise.
23300 (register_edge_assert_for_2): Likewise.
23301 (vrp_visit_cond_stmt): Likewise.
23302 (vrp_visit_switch_stmt): Likewise.
23303 (vrp_visit_stmt): Likewise.
23304 (vrp_visit_phi_node): Likewise.
23305 (simplify_cond_using_ranges_1): Likewise.
23306 (fold_predicate_in): Likewise.
23307 (evrp_dom_walker::before_dom_children): Likewise.
23308 (evrp_dom_walker::push_value_range): Likewise.
23309 (evrp_dom_walker::pop_value_range): Likewise.
23310 (execute_early_vrp): Likewise.
23312 2017-05-16 Richard Biener <rguenther@suse.de>
23314 * dwarf2out.c (loc_list_from_tree_1): Do not create
23315 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23317 2017-05-16 Richard Biener <rguenther@suse.de>
23319 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23321 (note_variable_value_in_expr): If we resolved the decl ref
23322 do not push to the stack.
23324 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23326 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23327 operations in fast-math mode.
23328 (vaddq_f16): Likewise.
23329 (vmul_f16): Likewise.
23330 (vmulq_f16): Likewise.
23331 (vsub_f16): Likewise.
23332 (vsubq_f16): Likewise.
23333 * config/arm/neon.md (add<mode>3): New.
23335 (fma:<VH:mode>3): New. Also remove outdated comment.
23338 2017-05-16 Martin Liska <mliska@suse.cz>
23342 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23343 (odr_types_equivalent_p): Likewise.
23345 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23347 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23349 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23352 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23353 non-interunit SSE move alternatives with '?'.
23354 (zero-extendsidi peephole2): New peephole to skip intermediate
23355 general register in SSE zero-extend sequence.
23357 2017-05-15 Jeff Law <law@redhat.com>
23359 * reorg.c (relax_delay_slots): Create a new variable to hold
23360 the temporary target rather than clobbering TARGET_LABEL.
23362 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23363 missing argument to extract_bit_field call.
23364 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23366 2017-05-15 Martin Liska <mliska@suse.cz>
23369 * gcc.c (process_command): Do not allow empty argument of -o option.
23371 2017-05-15 Renlin Li <renlin.li@arm.com>
23373 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23374 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23375 * config/aarch64/constraints.md (Usf): Add long call check.
23376 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23377 (call_value): Likewise.
23378 (sibcall): Likewise.
23379 (sibcall_value): Likewise.
23381 (call_value_insn): New.
23382 (sibcall_insn): Update rtx pattern.
23383 (sibcall_value_insn): Likewise.
23384 (call_internal): Remove.
23385 (call_value_internal): Likewise.
23386 (sibcall_internal): Likewise.
23387 (sibcall_value_internal): Likewise.
23388 (call_reg): Likewise.
23389 (call_symbol): Likewise.
23390 (call_value_reg): Likewise.
23391 (call_value_symbol): Likewise.
23393 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23396 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23398 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23400 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23401 compatible with CCGOCmode and with CCZmode.
23403 2017-05-14 Martin Sebor <msebor@redhat.com>
23405 PR middle-end/77671
23406 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23407 (gimple_fold_builtin_snprintf): Same.
23408 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23409 (gimple_fold_builtin_snprintf): Same.
23410 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23411 of character types.
23412 (is_call_safe): New function.
23413 (try_substitute_return_value): Call it.
23414 (try_simplify_call): New function.
23415 (pass_sprintf_length::handle_gimple_call): Call it.
23417 2017-05-14 Martin Sebor <msebor@redhat.com>
23419 PR middle-end/80669
23420 * builtins.c (expand_builtin_stpncpy): Simplify.
23422 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23424 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23425 * config/i386/i386.h
23426 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23427 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23428 (struct machine_function): Add new members call_ms2sysv,
23429 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23430 (struct machine_frame_state): New fields sp_realigned and
23431 sp_realigned_offset.
23432 * config/i386/i386.c
23433 (enum xlogue_stub): New enum.
23434 (enum xlogue_stub_sets): New enum.
23435 (class xlogue_layout): New class.
23436 (struct ix86_frame): New fields stack_realign_allocate_offset,
23437 stack_realign_offset and outlined_save_offset. Modify comments to
23438 detail stack layout when using out-of-line stubs.
23439 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23440 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23441 -mcall-ms2sysv-xlogues.
23442 (stub_managed_regs): New static variable.
23443 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23444 registers managed by out-of-line stub.
23445 (disable_call_ms2sysv_xlogues): New function.
23446 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23447 m->call_ms2sysv when appropriate and compute frame layout for
23449 (sp_valid_at, fp_valid_at): New inline functions.
23450 (choose_basereg): New function.
23451 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23453 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23454 Use align parameter of choose_baseaddr to generated aligned SSE movs
23456 (pro_epilogue_adjust_stack): Modify to track
23457 machine_frame_state::sp_realigned.
23458 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23459 (ix86_nsaved_sseregs): Likewise.
23460 (ix86_emit_save_regs): Likewise.
23461 (ix86_emit_save_regs_using_mov): Likewise.
23462 (ix86_emit_save_sse_regs_using_mov): Likewise.
23463 (get_scratch_register_on_entry): Likewise.
23464 (gen_frame_set): New function.
23465 (gen_frame_load): Likewise.
23466 (gen_frame_store): Likewise.
23467 (emit_outlined_ms2sysv_save): Likewise.
23468 (emit_outlined_ms2sysv_restore): Likewise.
23469 (ix86_expand_prologue): Modify stack re-alignment code and call
23470 emit_outlined_ms2sysv_save when appropriate.
23471 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23472 parameter rtx_insn *insn, which allows the function to be used to only
23473 generate the notes.
23474 (ix86_expand_epilogue): Modify validity checks of frame and stack
23475 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23476 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23477 * config/i386/predicates.md
23478 (save_multiple): New predicate.
23479 (restore_multiple): Likewise.
23480 * config/i386/sse.md
23481 (save_multiple<mode>): New pattern.
23482 (save_multiple_realign<mode>): Likewise.
23483 (restore_multiple<mode>): Likewise.
23484 (restore_multiple_and_return<mode>): Likewise.
23485 (restore_multiple_leave_return<mode>): Likewise.
23486 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23488 2017-05-14 Julia Koval <julia.koval@intel.com>
23490 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23491 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23492 (__builtin_ia32_xsetbv): New builtins.
23493 * config/i386/i386.c (ix86_expand_special_args_builtin):
23495 (ix86_expand_builtin): Special expand for new intrinsics.
23496 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23497 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23498 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23500 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23502 * cfganal.c (inverted_post_order_compute): Change argument type
23504 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23505 * df-core.c (rest_of_handle_df_initialize): Adjust.
23506 (rest_of_handle_df_finish): Likewise.
23507 (df_analyze_1): Likewise.
23508 (df_analyze): Likewise.
23509 (loop_inverted_post_order_compute): Change argument to be a vec *.
23510 (df_analyze_loop): Adjust.
23511 (df_get_n_blocks): Likewise.
23512 (df_get_postorder): Likewise.
23513 * df.h (struct df_d): Change field to be a vec.
23514 * lcm.c (compute_laterin): Adjust.
23515 (compute_available): Likewise.
23516 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23517 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23518 * tree-ssa-pre.c (compute_antic): Likewise.
23520 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23522 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23523 (depth_first_search::depth_first_search): Change structure init
23524 function to this constructor.
23525 (depth_first_search::add_bb): Rename function to this member.
23526 (depth_first_search::execute): Likewise.
23527 (flow_dfs_compute_reverse_finish): Adjust.
23529 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23531 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23532 (longest_simple_path): Likewise.
23533 * shrink-wrap.c (spread_components): Likewise.
23534 (disqualify_problematic_components): Likewise.
23535 (emit_common_heads_for_components): Likewise.
23536 (emit_common_tails_for_components): Likewise.
23537 (insert_prologue_epilogue_for_components): Likewise.
23539 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23541 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23544 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23546 * df-core.c (df_set_blocks): Start using auto_bitmap.
23547 (df_compact_blocks): Likewise.
23548 * df-problems.c (df_rd_confluence_n): Likewise.
23549 * df-scan.c (df_insn_rescan_all): Likewise.
23550 (df_process_deferred_rescans): Likewise.
23551 (df_update_entry_block_defs): Likewise.
23552 (df_update_exit_block_uses): Likewise.
23553 (df_entry_block_bitmap_verify): Likewise.
23554 (df_exit_block_bitmap_verify): Likewise.
23555 (df_scan_verify): Likewise.
23556 * lra-constraints.c (lra_constraints): Likewise.
23557 (undo_optional_reloads): Likewise.
23558 (lra_undo_inheritance): Likewise.
23559 * lra-remat.c (calculate_gen_cands): Likewise.
23560 (do_remat): Likewise.
23561 * lra-spills.c (assign_spill_hard_regs): Likewise.
23562 (spill_pseudos): Likewise.
23563 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23564 (bitmap_set_subtract_values): Likewise.
23566 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23568 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23569 management with auto_bitmap.
23570 (fix_inter_tick): Likewise.
23571 (fix_recovery_deps): Likewise.
23572 * ira.c (add_store_equivs): Likewise.
23573 (find_moveable_pseudos): Likewise.
23574 (split_live_ranges_for_shrink_wrap): Likewise.
23575 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23576 (rtx_reuse_manager::seen_def_p): Likewise.
23577 (rtx_reuse_manager::set_seen_def): Likewise.
23578 * print-rtl.h (class rtx_reuse_manager): Likewise.
23580 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23582 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23584 (migrate_btr_def): Likewise.
23585 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23586 * df-core.c (loop_post_order_compute): Likewise.
23587 (loop_inverted_post_order_compute): Likewise.
23588 * hsa-common.h: Likewise.
23589 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23590 * init-regs.c (initialize_uninitialized_regs): Likewise.
23591 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23592 (inline_small_functions): Likewise.
23593 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23594 * ira.c (combine_and_move_insns): Likewise.
23595 (build_insn_chain): Likewise.
23596 * loop-invariant.c (find_invariants): Likewise.
23597 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23598 * predict.c (tree_predict_by_opcode): Likewise.
23599 (predict_paths_leading_to): Likewise.
23600 (predict_paths_leading_to_edge): Likewise.
23601 (estimate_loops_at_level): Likewise.
23602 (estimate_loops): Likewise.
23603 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23604 (spread_components): Likewise.
23605 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23606 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23607 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23608 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23609 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23610 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23611 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23612 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23613 (mark_threaded_blocks): Likewise.
23614 (thread_through_all_blocks): Likewise.
23615 * tree-ssa.c (verify_ssa): Likewise.
23616 (execute_update_addresses_taken): Likewise.
23617 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23619 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23621 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23623 (post_order_compute): Likewise.
23624 (inverted_post_order_compute): Likewise.
23625 (pre_and_rev_post_order_compute_fn): Likewise.
23627 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23629 * genrecog.c (int_set::int_set): Explicitly construct our
23630 auto_vec base class.
23631 * vec.h (auto_vec::auto_vec): New constructor.
23633 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23635 * bitmap.h (class auto_bitmap): New constructor taking
23636 bitmap_obstack * argument.
23638 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23640 * bitmap.h (class auto_bitmap): Change type of m_bits to
23641 bitmap_head, and adjust ctor / dtor and member operators.
23643 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23645 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23646 when returned register mode doesn't match original mode.
23648 2017-05-12 Jeff Law <law@redhat.com>
23649 Jakub Jelinek <jakub@redhat.com>
23651 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23652 we look for cc setter after the compare-elim changes.
23653 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23654 within the vector to match what compare-elim now expects.
23655 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23656 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23658 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23659 after the compare-elim changes.
23660 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23661 the vector to match what compare-elim now expects.
23662 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23663 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23664 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23665 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23666 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23668 * config/visium/visium.c (single_set_and_flags): Fix where
23669 we look for cc setter after the compare-elim changes.
23670 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23671 with the vector to match what compare-elim now expects.
23672 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23673 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23674 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23675 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23676 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23677 (neg<mode>2_insn_set_overflow): Likewise.
23679 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23681 PR middle-end/79794
23682 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23683 maybe_expand_insn call, set ops[0].target. If still set after call,
23684 set alt_rtl. Add extra arg to recursive calls.
23685 (extract_bit_field): Add alt_rtl argument. Pass to
23687 * expmed.h (extract_bit_field): Fix prototype.
23688 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23689 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23690 to extract_bit_field_calls.
23691 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23692 Pass alt_rtl to extract_bit_field calls.
23693 * calls.c (store_unaligned_arguments_into_psuedos)
23694 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23695 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23697 * optabs.h (struct expand_operand): Add target bitfield.
23699 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23701 * compare-elim.c (try_eliminate_compare): Canonicalize
23702 operation with embedded compare to
23703 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23704 (set (reg) (operation)].
23706 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23708 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23711 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23712 cost of adding a carry flag for ADC instruction.
23713 [case MINUS]: Ignore the cost of subtracting a carry flag
23714 for SBB instruction.
23716 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23718 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23720 * config/rs6000/bmiintrin.h: New file.
23721 * config/rs6000/bmi2intrin.h: New file.
23722 * config/rs6000/x86intrin.h: New file.
23724 2017-05-12 Jeff Law <law@redhat.com>
23726 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23729 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23731 PR middle-end/80707
23732 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23734 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23736 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23737 early expansion of vector divide builtins.
23738 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23739 builtins identified as having unsigned arguments.
23741 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23743 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23744 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23745 expansion of vector logical operations (and, andc, or, xor,
23748 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23750 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23751 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23753 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23755 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23756 early GIMPLE expansion of vector multiplies.
23758 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23760 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23761 TARGET_HAVE_MOVT conditional.
23762 (movt splitter): Likewise.
23764 2017-05-12 Richard Biener <rguenther@suse.de>
23766 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23767 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23768 Fold all stmts not inplace.
23770 2017-05-12 Richard Biener <rguenther@suse.de>
23772 PR tree-optimization/80713
23773 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23774 inserted_exprs bit for not removed stmts.
23776 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23778 PR middle-end/69921
23779 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23780 parallelized" attribute for parallelized OpenACC kernels.
23781 * omp-offload.c (execute_oacc_device_lower): Use it.
23783 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23784 Set "oacc kernels" attribute.
23785 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23786 parameter. Adjust all users.
23787 (oacc_fn_attrib_kernels_p): Remove function.
23788 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23789 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23790 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23791 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23792 assert "oacc kernels" attribute is set.
23794 2017-05-11 Carl Love <cel@us.ibm.com>
23796 * config/rs6000/rs6000-c: Add support for built-in functions
23797 vector unsigned char vec_popcnt (vector signed char)
23798 vector unsigned char vec_popcnt (vector unsigned char)
23799 vector unsigned short vec_popcnt (vector signed short)
23800 vector unsigned short vec_popcnt (vector unsigned short)
23801 vector unsigned int vec_popcnt (vector signed int)
23802 vector unsigned int vec_popcnt (vector unsigned int)
23803 vector unsigned long long vec_popcnt (vector signed long long)
23804 vector unsigned long long vec_popcnt (vector unsigned long long)
23805 vector signed long long vec_slo (vector signed long long,
23806 vector signed char)
23807 vector signed long long vec_slo (vector signed long long,
23808 vector unsigned char)
23809 vector unsigned long long vec_slo (vector unsigned long long,
23810 vector signed char)
23811 vector unsigned long long vec_slo (vector unsigned long long,
23812 vector unsigned char)
23813 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23814 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23815 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23816 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23817 * doc/extend.texi: Update the built-in documentation file for the
23818 new built-in functions.
23820 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23822 * attribs.h (sorted_attr_string): Move machine independent
23823 functions for target clone support from the i386 port to common
23824 code. Rename ix86_function_versions to common_function_versions.
23825 Rename make_name to make_unique_name.
23826 (common_function_versions): Likewise.
23827 (make_unique_name): Likewise.
23828 (make_dispatcher_decl): Likewise.
23829 (is_function_default_version): Likewise.
23830 * attribs.c (attr_strcmp): Likewise.
23831 (sorted_attr_string): Likewise.
23832 (common_function_versions): Likewise.
23833 (make_unique_name): Likewise.
23834 (make_dispatcher_decl): Likewise.
23835 (is_function_default_version): Likewise.
23836 * config/i386/i386.c (attr_strcmp): Likewise.
23837 (sorted_attr_string): Likewise.
23838 (ix86_function_versions): Likewise.
23839 (make_name): Likewise.
23840 (make_dispatcher_decl): Likewise.
23841 (is_function_default_version): Likewise.
23842 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23844 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23847 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23848 Account for direct move costs for vec_construct of integer
23851 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23854 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23855 (UNSPEC_STX_ATOMIC): Ditto.
23856 (loaddi_via_sse): New insn.
23857 (storedi_via_sse): Ditto.
23858 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23859 Update corresponding peephole2 patterns.
23860 (atomic_storedi_fpu): Ditto.
23862 2017-05-11 Julia Koval <julia.koval@intel.com>
23864 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23865 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23867 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23868 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23869 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23871 2017-05-11 Nathan Sidwell <nathan@acm.org>
23873 * graphite-poly.c: Include dumpfile.h.
23875 * dumpfle.h (dump_function): Declare here ...
23876 * tree-dump.h (dump_function): ... not here.
23877 * dumpfile.c: #include tree-cfg.h.
23878 (dump_function): Move here from ...
23879 * tree-dump.c (dump_function): ... here.
23880 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23881 * graphite-poly.c: Don't include tree-dump.h.
23882 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23883 * print-tree.c: Likewise.
23884 * stor-layout.c: Likewise.
23885 * tree-nested.c: Likewise.
23887 * dumpfile.c (dump_start): Use TDF_FLAGS.
23888 (dump_enable_all): Fix TDF_KIND check thinko.
23890 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23892 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23893 array entries to represent two legal parameterizations of the
23894 overloaded __builtin_cmpb function, as represented by the
23895 P6_OV_BUILTIN_CMPB constant.
23896 (altivec_resolve_overloaded_builtin): Add special case handling
23897 for the __builtin_cmpb function, as represented by the
23898 P6_OV_BUILTIN_CMPB constant.
23899 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23900 (BU_P6_64BIT_2): New macro.
23901 (BU_P6_OVERLOAD_2): New macro
23902 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23903 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23904 (CMPB): Add overload support to represent both 32-bit and 64-bit
23905 compare-bytes function.
23906 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23907 support for TARGET_CMPB.
23908 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23909 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23910 documentation of the __builtin_cmpb overloaded built-in function.
23912 2017-05-11 Richard Biener <rguenther@suse.de>
23914 PR tree-optimization/80705
23915 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23916 bases are not vectorizable.
23918 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23920 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23921 when counting register pressure.
23923 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23925 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23926 (iv_ca_more_deps): Renamed to ...
23927 (iv_ca_compare_deps): ... this.
23928 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23930 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23932 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23934 (determine_group_iv_costs): ... here.
23935 (find_inv_vars_cb): Record inv var if it's not recorded before.
23937 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23939 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23940 (get_shiftadd_cost): Ditto.
23942 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23944 * tree-ssa-address.c: Include header file.
23945 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23947 (add_to_parts): Refactor.
23948 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23949 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23952 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23954 PR tree-optimization/53090
23955 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23957 (extract_cond_operands): Detect condition with IV on both sides
23958 and return COMP_IV_EXPR_2.
23959 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23960 (rewrite_use_compare): Simplify by removing call to function
23961 extract_cond_operands.
23963 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23965 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23966 (extract_cond_operands): Detect condition comparing against non-
23967 invariant bound and return appropriate enum value.
23968 (find_interesting_uses_cond): Update use of extract_cond_operands.
23969 Handle its return value accordingly.
23970 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23972 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23974 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23975 nonlinear iv_use computation in loop invariant sensitive way.
23977 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23979 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23980 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23982 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23984 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23985 (dump_cand): Support iv_cand.inv_exprs.
23986 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23988 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23991 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23993 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23995 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23996 as local function. Include necessary header files.
23997 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23999 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24001 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24003 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24005 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24006 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24007 RSHIFT_EXPR and BIT_NOT_EXPR.
24009 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24011 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24012 (adjust_setup_cost): New parameter supporting round up adjustment.
24013 (struct address_cost_data): Delete.
24014 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24015 (split_address_cost, ptr_difference_cost): Delete.
24016 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24017 (struct ainc_cost_data): New struct.
24018 (get_address_cost_ainc): New function.
24019 (get_address_cost, get_computation_cost): Reimplement.
24020 (determine_group_iv_cost_address): Record inv_expr for all uses of
24022 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24023 (iv_ca_has_deps): Reimplemented to ...
24024 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24026 (iv_ca_extend): Call iv_ca_more_deps.
24028 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24030 * tree-ssa-address.c (struct mem_address): Move to header file.
24031 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24032 * tree-ssa-address.h (struct mem_address): Move from C file.
24033 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24035 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24037 * tree-affine.h (aff_combination_type): New interface.
24038 (aff_combination_zero_p): Remove static.
24039 (aff_combination_const_p): New interface.
24040 (aff_combination_singleton_var_p): New interfaces.
24042 2017-05-11 Richard Biener <rguenther@suse.de>
24044 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24045 Skip unreachable blocks and destinations.
24046 (eliminate): Move stmt removal and fixup ...
24047 (fini_eliminate): ... here. Skip inserted exprs.
24048 (pass_pre::execute): Move fini_pre after fini_eliminate.
24049 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24050 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24051 PRE to get rid of dead code that has invalid SSA form and
24052 split critical edges again.
24054 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24056 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24058 2017-05-11 Richard Biener <rguenther@suse.de>
24060 * passes.c (execute_function_todo): Verify loops if they are
24061 said to be up-to-date.
24062 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24063 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24065 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24068 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24069 handle calling assemble_external ourself.
24072 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24073 modes with zero size. Enhance comment.
24075 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24077 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24078 built-ins for vec_xl and vec_xst with short and char pointer
24081 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24083 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24084 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24085 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24086 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24087 (_mm_maskz_min_round_ss): New intrinsics.
24088 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24089 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24090 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24091 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24092 (__builtin_ia32_minss_mask_round): New builtins.
24093 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24094 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24095 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24097 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24098 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24100 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24103 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24105 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24106 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24107 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24108 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24109 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24110 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24111 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24112 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24113 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24114 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24115 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24116 (__builtin_ia32_mulss_mask_round): New builtins.
24117 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24118 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24119 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24121 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24122 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24124 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24127 2017-05-10 Julia Koval <julia.koval@intel.com>
24129 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24130 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24131 (_mm256_setr_m128i): New intrinsics.
24133 2017-05-10 Julia Koval <julia.koval@intel.com>
24135 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24136 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24137 (_mm_maskz_rcp14_ss): New intrinsics.
24138 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24139 (__builtin_ia32_rcp14ss_mask): New builtins.
24140 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24142 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24144 PR tree-optimization/51513
24145 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24146 (assert_unreachable_fallthru_edge_p): Use it.
24147 (group_case_labels_stmt): Likewise.
24148 * tree-cfg.h: Prototype it.
24149 * stmt.c: Include cfghooks.h and tree-cfg.h.
24150 (emit_case_dispatch_table) <gap_label>: New local variable.
24151 Use it to fill dispatch table gaps.
24152 Test for default_label before updating probabilities.
24153 (expand_case) <default_label>: Remove unneeded initialization.
24154 Test for unreachable default case statement and remove its edge.
24155 Set default_label accordingly.
24156 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24158 2017-05-10 Carl Love <cel@us.ibm.com>
24160 * config/rs6000/rs6000-c: Add support for built-in functions
24161 vector signed char vec_neg (vector signed char)
24162 vector signed short int vec_neg (vector short int)
24163 vector signed int vec_neg (vector signed int)
24164 vector signed long long vec_neg (vector signed long long)
24165 vector float vec_neg (vector float)
24166 vector double vec_neg (vector double)
24167 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24169 * config/rs6000/altivec.h: Add define for vec_neg
24170 * doc/extend.texi: Update the built-in documentation for the
24171 new built-in functions.
24173 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24175 PR tree-optimization/77644
24176 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24178 2017-05-10 Nathan Sidwell <nathan@acm.org>
24180 * dumpfile.h (TDI_lang_all): New.
24181 (TDF_KIND): New. Renumber others
24182 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24184 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24186 (get_dump_file_name): Adjust suffix generation.
24187 (dump_enable_all): Use TDF_KIND.
24188 * doc/invoke.texi (-fdump-lang-all): Document.
24190 * dumpfile.h: Tabify.
24192 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24195 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24196 Move member access before delete.
24198 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24200 * tree-inline.c (expand_call_inline): Split block at stmt
24203 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24206 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24207 are now unused after splitting mov{sf,sd}_hardfloat.
24208 (f32_lr2): Likewise.
24209 (f32_lm): Likewise.
24210 (f32_lm2): Likewise.
24211 (f32_li): Likewise.
24212 (f32_li2): Likewise.
24213 (f32_lv): Likewise.
24214 (f32_sr): Likewise.
24215 (f32_sr2): Likewise.
24216 (f32_sm): Likewise.
24217 (f32_sm2): Likewise.
24218 (f32_si): Likewise.
24219 (f32_si2): Likewise.
24220 (f32_sv): Likewise.
24221 (f32_dm): Likewise.
24222 (f32_vsx): Likewise.
24223 (f32_av): Likewise.
24224 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24225 For movsf, order stores so the VSX stores occur before the GPR
24226 store which encourages the register allocator to use a traditional
24227 FPR instead of a GPR. For movsd, order the stores so that the GPR
24228 store comes before the VSX stores to allow the power6 to work.
24229 This is due to the power6 not having a 32-bit integer store
24230 instruction from a FPR.
24231 (movsf_hardfloat): Likewise.
24232 (movsd_hardfloat): Likewise.
24234 2017-05-09 Martin Sebor <msebor@redhat.com>
24236 PR translation/80280
24237 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24240 PR translation/80280
24241 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24242 data member added in r247778.
24243 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24245 2017-05-09 Nathan Sidwell <nathan@acm.org>
24247 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24249 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24252 2017-05-09 Marek Polacek <polacek@redhat.com>
24254 * doc/invoke.texi: Fix typo.
24256 2017-05-09 Richard Biener <rguenther@suse.de>
24258 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24259 (vrp_val_is_min): Likewise.
24260 (set_value_range_to_value): Likewise.
24261 (set_value_range_to_nonnegative): Likewise.
24262 (gimple_assign_nonzero_p): Likewise.
24263 (gimple_stmt_nonzero_p): Likewise.
24264 (vrp_int_const_binop): Likewise. Remove unreachable case.
24265 (adjust_range_with_scev): Adjust comments.
24266 (compare_range_with_value): Likewise.
24267 (extract_range_from_phi_node): Likewise.
24268 (test_for_singularity): Likewise.
24270 2017-05-09 Richard Biener <rguenther@suse.de>
24272 * tree-vrp.c (get_single_symbol): Add assert that we don't
24273 get overflowed constants as invariant part.
24274 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24275 checks. Use wi::cmp instead of recursing for integer constants.
24276 (compare_values): Just ignore whether we assumed undefined
24277 overflow instead of failing the compare.
24278 (extract_range_for_var_from_comparison_expr): Add comment before the
24279 TREE_NO_WARNING sets.
24280 (test_for_singularity): Likewise.
24281 (extract_range_from_comparison): Do not disable optimization
24282 when we assumed undefined overflow.
24283 (extract_range_basic): Remove init of unused var.
24285 2017-05-09 Richard Biener <rguenther@suse.de>
24287 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24288 (extract_range_from_multiplicative_op_1): Adjust.
24289 (extract_range_from_binary_expr_1): Use int_const_binop.
24291 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24294 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24295 rs6000_store_data_bypass_p in seven define_bypass directives and
24296 in several comments.
24297 * config/rs6000/rs6000-protos.h: Add prototype for
24298 rs6000_store_data_bypass_p function.
24299 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24300 function implements slightly different (rs6000-specific) semantics
24301 than store_data_bypass_p, returning false rather than aborting
24302 with assertion error when arguments do not satisfy the
24303 requirements of store data bypass.
24304 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24305 rs6000_store_data_bypass_p.
24307 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24309 * config/xtensa/xtensa-protos.h
24310 (xtensa_initial_elimination_offset): New declaration.
24311 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24312 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24313 macro definition, add case for FRAME_POINTER_REGNUM when
24314 FRAME_GROWS_DOWNWARD.
24315 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24316 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24317 xtensa_initial_elimination_offset.
24319 2017-05-08 Nathan Sidwell <nathan@acm.org>
24321 * doc/invoke.texi: Alphabetize -fdump options.
24323 2017-05-08 Martin Sebor <msebor@redhat.com>
24325 PR translation/80280
24326 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24328 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24330 * target.def (compute_frame_layout): New optional target hook.
24331 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24332 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24333 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24335 * reload1.c (verify_initial_elim_offsets): Likewise.
24336 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24337 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24338 (arm_get_frame_offsets): Split up into this ...
24339 (arm_compute_frame_layout): ... and this function.
24341 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24343 * config/aarch64/constraints.md (Usa): New constraint.
24344 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24346 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24348 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24349 with_multilib_list after it has been checked.
24351 2017-05-08 Richard Biener <rguenther@suse.de>
24353 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24354 (bitmap_set_subtract_values): Likewise.
24356 2017-05-08 Richard Biener <rguenther@suse.de>
24358 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24359 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24361 (gimple_stmt_nonzero_warnv_p): Rename to ...
24362 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24364 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24365 (extract_range_basic): Adjust, do not disable propagation on
24366 strict overflow sensitive simplification.
24367 (vrp_visit_cond_stmt): Likewise.
24369 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24371 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24372 body size unconditionally.
24374 2017-05-07 Jeff Law <law@redhat.com>
24377 2017-05-06 Jeff Law <law@redhat.com>
24378 PR tree-optimization/78496
24379 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24382 PR tree-optimization/78496
24383 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24384 (simplify_stmt_using_ranges): Call it.
24385 (vrp_dom_walker::before_dom_children): Extract equivalences
24386 from an ASSERT_EXPR with an equality comparison against a
24389 2017-05-06 Jeff Law <law@redhat.com>
24391 PR tree-optimization/78496
24392 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24395 PR tree-optimization/78496
24396 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24397 (simplify_stmt_using_ranges): Call it.
24398 (vrp_dom_walker::before_dom_children): Extract equivalences
24399 from an ASSERT_EXPR with an equality comparison against a
24402 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24404 * lra-constraints.c (lra_copy_reg_equiv): New function.
24405 (split_reg): Use it to copy equivalence information from the
24406 original register to the spill register.
24408 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24410 PR rtl-optimization/75964
24411 * simplify-rtx.c (simplify_const_relational_operation): Remove
24412 invalid handling of comparisons of integer ABS.
24414 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24416 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24417 initialize to zero.
24418 (init_regs): Remove declaration.
24419 (function_arg_advance_32): Initialize error_p as boolean variable.
24421 2017-05-05 Nathan Sidwell <nathan@acm.org>
24423 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24424 lines. Use for (;;).
24426 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24428 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24429 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24430 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24431 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24432 VF=2 that require versioning.
24434 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24436 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24439 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24441 * diagnostic.h (diagnostic_override_option_index): Convert from
24442 macro to inline function.
24444 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24446 * diagnostic.c (last_module_changed_p): New function.
24447 (set_last_module): New function.
24448 (diagnostic_report_current_module): Convert macro usage to
24449 the above functions.
24450 * diagnostic.h (diagnostic_context::last_module): Strengthen
24451 from const line_map * to const line_map_ordinary *.
24452 (diagnostic_last_module_changed): Delete macro.
24453 (diagnostic_set_last_module): Delete macro.
24455 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24457 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24458 with diagnostic_report_diagnostic.
24459 (diagnostic_n_impl_richloc): Likewise.
24460 * diagnostic.h (report_diagnostic): Delete macro.
24461 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24462 with diagnostic_report_diagnostic.
24463 * substring-locations.c (format_warning_va): Likewise.
24465 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24467 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24468 save/restor of format_spec. Move option-printing code to...
24469 (print_option_information): ...this new function, and
24470 reimplement by simply printing to the pretty_printer,
24471 rather than appending to the format string.
24473 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24475 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24476 handling logic into...
24477 (update_effective_level_from_pragmas): ...this new function.
24479 2017-05-04 Andrew Waterman <andrew@sifive.com>
24481 * config/riscv/riscv.opt (mstrict-align): New option.
24482 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24483 (SLOW_UNALIGNED_ACCESS): Define.
24484 (riscv_slow_unaligned_access): Declare.
24485 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24487 (riscv_slow_unaligned_access): New variable.
24488 (rocket_tune_info): Set slow_unaligned_access to true.
24489 (optimize_size_tune_info): Set slow_unaligned_access to false.
24490 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24491 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24492 (riscv_option_override): Set riscv_slow_unaligned_access.
24493 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24495 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24497 * config/riscv/riscv.md: Unify indentation.
24499 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24504 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24506 (extendsi<mode>2): Add support for doing sign extension via
24507 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24508 don't have ISA 3.0 instructions.
24509 (extendsi<mode>2 splitter): Likewise.
24510 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24511 generate the normal insns since SImode can now go in vector
24512 registers. Disallow the special UNSPECs needed for previous
24513 machines to hide SImode being used. Add new insns
24514 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24515 (fix_trunc<mode>si2_stfiwx): Likewise.
24516 (fix_trunc<mode>si2_internal): Likewise.
24517 (fixuns_trunc<mode>si2): Likewise.
24518 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24519 (fctiw<u>z_<mode>_smallint): Likewise.
24520 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24521 of floating point to 32-bit integer from doing a direct move to
24522 the GPR registers to do a store.
24523 (fctiwz_<mode>): Break long line.
24525 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24527 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24528 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24529 (addr_list, addr_offset_valid_p): New.
24530 (split_address_groups): Check offset validity with above function.
24531 (gt-tree-ssa-loop-ivopts.h): Include header file.
24533 2017-05-05 Nathan Sidwell <nathan@acm.org>
24535 * config.gcc (arm*-*-*): Add missing 'fi'.
24537 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24539 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24540 included in -fopt-info does not matter.
24541 * doc/optinfo.texi (-fopt-info): Fix description of default
24542 behavour. Explicitly say order of options included in -fopt-info
24545 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24547 * config.gcc: Allow combinations of aprofile and rmprofile values for
24548 --with-multilib-list.
24549 * config/arm/t-multilib: New file.
24550 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24551 variables. Remove setting of ISA and floating-point ABI in
24552 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24553 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24554 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24555 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24557 * config/arm/t-rmprofile: Likewise except for the matches changes.
24558 * doc/install.texi (--with-multilib-list): Document the combination of
24559 aprofile and rmprofile values and warn about pitfalls in doing that.
24561 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24563 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24564 (movdi_aarch64): Likewise.
24566 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24568 PR tree-optimization/80632
24569 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24571 (build_arrays): Initialize it for virtual phis.
24572 (fix_phi_nodes): Use it for virtual phis.
24574 PR tree-optimization/80558
24575 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24576 [x, y] op z into [x op, y op z] for op & or | if conditions
24579 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24580 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24583 * config/arm/arm.md (use_literal_pool): Remove.
24584 (64-bit immediate split): No longer takes cost into consideration
24585 if arm_disable_literal_pool is enabled.
24586 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24587 used when arm_disable_literal_pool is enabled.
24588 (arm_max_const_double_inline_cost): Remove use of
24589 arm_disable_literal_pool.
24590 (push_minipool_fix): Add assert.
24591 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24592 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24593 (no_literal_pool_sf_immediate): New.
24595 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24597 PR tree-optimization/80613
24598 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24599 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24601 2017-05-05 Richard Biener <rguenther@suse.de>
24603 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24605 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24607 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24608 of this flag from insn conditions due to removal from r247495.
24610 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24612 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24614 (arm_early_store_addr_dep_ptr): Likewise.
24615 * config/arm/aarch-common-protos.h
24616 (arm_early_load_addr_dep_ptr): Add prototype.
24617 (arm_early_store_addr_dep_ptr): Likewise.
24618 * config/arm/cortex-a53.md: Add new bypasses.
24620 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24622 * tree.c (next_type_uid): Change type to unsigned.
24623 (type_hash_canon): Decrement back next_type_uid if
24624 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24625 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24628 2017-05-04 Martin Sebor <msebor@redhat.com>
24630 * builtins.c: Fix a trivial typo in a comment.
24632 PR middle-end/79234
24633 * builtins.c (check_sizes): Adjust to handle reading past the end.
24634 Avoid printing excessive upper bound of ranges. Use %E to print
24635 tree nodes instead of converting them to %wu.
24636 (expand_builtin_memchr): New function.
24637 (compute_dest_size): Rename...
24638 (compute_objsize): ...to this.
24639 (expand_builtin_memcpy): Adjust.
24640 (expand_builtin_mempcpy): Adjust.
24641 (expand_builtin_strcat): Adjust.
24642 (expand_builtin_strcpy): Adjust.
24643 (check_strncat_sizes): Adjust.
24644 (expand_builtin_strncat): Adjust.
24645 (expand_builtin_strncpy): Adjust and simplify.
24646 (expand_builtin_memset): Adjust.
24647 (expand_builtin_bzero): Adjust.
24648 (expand_builtin_memcmp): Adjust.
24649 (expand_builtin): Handle memcmp.
24650 (maybe_emit_chk_warning): Check strncat just once.
24652 2017-05-04 Martin Sebor <msebor@redhat.com>
24654 PR preprocessor/79214
24655 PR middle-end/79222
24656 PR middle-end/79223
24657 * builtins.c (check_sizes): Add inlining context and issue
24658 warnings even when -Wno-system-headers is set.
24659 (check_strncat_sizes): Same.
24660 (expand_builtin_strncat): Same.
24661 (expand_builtin_memmove): New function.
24662 (expand_builtin_stpncpy): Same.
24663 (expand_builtin): Handle memmove and stpncpy.
24665 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24667 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24668 which is not used any more.
24670 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24672 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24674 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24676 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24677 (cortexa53_tunings): Likewise.
24678 (cortexa57_tunings): Likewise.
24679 (cortexa72_tunings): Likewise.
24680 (cortexa73_tunings): Likewise.
24682 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24684 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24685 Set loop alignment to 8.
24687 2017-05-04 Martin Sebor <msebor@redhat.com>
24689 PR translation/80280
24690 * builtins.c (expand_builtin_object_size): Add missing quoting to
24691 %D and like directives.
24692 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24693 (hsa_type_for_tree_type): Same.
24694 (verify_function_arguments): Same.
24695 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24696 * varasm.c (get_section): Same.
24699 2017-05-04 Martin Sebor <msebor@redhat.com>
24701 PR translation/80280
24702 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24704 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24706 * config/aarch64/aarch64.c (generic_addrcost_table):
24707 Change HI/TI mode setting.
24709 2017-05-04 Martin Jambor <mjambor@suse.cz>
24711 PR tree-optimization/80622
24712 * tree-sra.c (comes_initialized_p): New function.
24713 (build_accesses_from_assign): Only set write lazily when
24714 comes_initialized_p is false.
24715 (analyze_access_subtree): Use comes_initialized_p.
24716 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24717 instead of testing for PARM_DECL.
24719 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24721 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24722 constraint on operand 0 to allow more general addressing modes.
24723 Adjust output template.
24724 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24726 * config/aarch64/aarch64-protos.h
24727 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24728 * config/aarch64/constraints.md (Dp): New address constraint.
24729 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24732 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24734 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24735 update use of estimate_ipcp_clone_size_and_time.
24736 (estimate_local_effects): Update use of
24737 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24738 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24739 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24740 Return nonspecialized time.
24742 2017-05-04 Richard Biener <rguenther@suse.de>
24744 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24745 for the last VUSE which def dominates the PHI. Directly call
24747 (get_continuation_for_phi_1): Remove.
24749 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24751 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24752 to explain the use of truncating division. Cap the number of
24753 iterations to the maximum given by nb_iterations_upper_bound,
24756 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24758 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24759 * configure: Regenerate.
24760 * config.in: Regenerate.
24761 * config/i386/driver-mingw32.c: new file.
24762 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24763 * config.host: Link driver-mingw32.o on MinGW host.
24764 * doc/install.texi: Document new --enable-mingw-wildcard configure
24767 2017-05-04 Marek Polacek <polacek@redhat.com>
24769 PR tree-optimization/80612
24770 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24772 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24773 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24775 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24776 (movt splitter): Likewise.
24777 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24778 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24779 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24780 block for Thumb-1 with MOVT.
24781 (thumb2_legitimate_address_p): Move code block ...
24782 (can_avoid_literal_pool_for_label_p): ... into this new function.
24783 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24785 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24786 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24787 "M-profile targets with the MOVT instruction".
24789 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24791 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24792 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24793 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24795 2017-05-04 Martin Liska <mliska@suse.cz>
24797 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24798 variable cond_code.
24800 2017-05-04 Richard Biener <rguenther@suse.de>
24802 * tree.c (array_at_struct_end_p): Handle arrays at struct
24803 end with flexarrays more conservatively. Refactor and treat
24804 arrays of arrays or aggregates more strict. Fix
24805 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24806 * tree.c (array_at_struct_end_p): Adjust prototype.
24807 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24808 * gimple-fold.c (get_range_strlen): Likewise.
24809 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24811 2017-05-04 Richard Biener <rguenther@suse.de>
24813 PR tree-optimization/31130
24814 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24816 (supports_overflow_infinity): Likewise.
24817 (is_negative_overflow_infinity): Likewise.
24818 (is_positive_overflow_infinity): Likewise.
24819 (is_overflow_infinity): Likewise.
24820 (stmt_overflow_infinity): Likewise.
24821 (overflow_infinity_range_p): Likewise.
24822 (usable_range_p): Remove as always returning true.
24823 (make_overflow_infinity): Remove.
24824 (negative_overflow_infinity): Likewise.
24825 (positive_overflow_infinity): Likewise.
24826 (avoid_overflow_infinity): Likewise.
24827 (set_value_range): Adjust accordingly.
24828 (set_value_range_to_nonnegative): Likewise, remove now unused
24829 overflow_infinity arg.
24830 (vrp_operand_equal_p): Adjust.
24831 (update_value_range): Likewise.
24832 (range_int_cst_singleton_p): Likewise.
24833 (operand_less_p): Likewise.
24834 (compare_values_warnv): Likewise.
24835 (extract_range_for_var_from_comparison_expr): Likewise.
24836 (vrp_int_const_binop): Likewise.
24837 (zero_nonzero_bits_from_vr): Likewise.
24838 (extract_range_from_multiplicative_op_1): Likewise.
24839 (extract_range_from_binary_expr_1): Likewise.
24840 (extract_range_from_unary_expr): Likewise.
24841 (extract_range_from_comparison): Likewise.
24842 (extract_range_basic): Likewise.
24843 (adjust_range_with_scev): Likewise.
24844 (compare_ranges): Likewise.
24845 (compare_range_with_value): Likewise.
24846 (dump_value_range): Likewise.
24847 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24849 (simplify_cond_using_ranges): Adjust.
24851 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24853 * brig-builtins.def: Added a builtin for class_f64.
24854 * builtin-types.def: Added a builtin type needed by class_f64.
24856 2017-05-03 Jason Merrill <jason@redhat.com>
24858 * timevar.def: Add TV_CONSTEXPR.
24860 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24862 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24864 2017-05-03 Martin Jambor <mjambor@suse.cz>
24866 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24867 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24868 * ipa-cp.c (ipcp_propagate_stage): Do not call
24869 ipa_update_after_lto_read.
24870 * ipa-inline.c (ipa_inline): Likewise.
24872 2017-05-03 Martin Jambor <mjambor@suse.cz>
24874 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24875 tag. Added a default constructor and a destructor.
24876 (ipa_edge_args_sum_t): New class;
24877 (ipa_edge_args_sum): Declare.
24878 (ipa_edge_args_vector): Remove declaration.
24879 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24880 (ipa_free_edge_args_substructures): Remove declaration.
24881 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24882 (ipa_edge_args_info_available_for_edge_p): Likewise.
24883 * ipa-prop.c (ipa_edge_args_vector): Removed.
24884 (edge_removal_hook_holder): Likewise.
24885 (edge_duplication_hook_holder): Likewise.
24886 (ipa_edge_args_sum): New variable.
24887 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24888 ipa_edge_args_vector.
24889 (ipa_free_edge_args_substructures): Likewise.
24890 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24891 ipa_edge_args_vector.
24892 (ipa_edge_removal_hook): Turned into method
24893 ipa_edge_args_sum_t::remove.
24894 (ipa_edge_duplication_hook): Turned into method
24895 ipa_edge_args_sum_t::duplicate.
24896 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24897 registering edge hooks.
24898 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24899 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24900 ipa_edge_args_sum instead of ipa_edge_args_vector.
24901 * ipa-profile.c (ipa_profile): Likewise.
24903 2017-05-03 Martin Jambor <mjambor@suse.cz>
24905 * symbol-summary.h (function_summary): New method exists.
24906 (function_summary::symtab_removal): Deallocate through release.
24907 (call_summary): New class.
24908 (gt_ggc_mx): New overload.
24909 (gt_pch_nx): Likewise.
24910 (gt_pch_nx): Likewise.
24912 2017-05-03 Jeff Law <law@redhat.com>
24914 PR tree-optimization/78496
24915 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24916 from simplify_cond_using_ranges. Split off code to walk
24917 backwards through casts into ...
24918 (simplify_cond_using_ranges_2): New function.
24919 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24920 (execute_vrp): After identifying jump threads, call
24921 simplify_cond_using_ranges_2.
24923 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24926 * ipa-inline.h (inline_summary): Add ctor.
24927 (create_ggc): Do not use ggc_cleared_alloc.
24929 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24930 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24932 * gcc.c (handle_braces): Support escaping in switch matching
24934 * doc/invoke.texi (Spec Files): Document it.
24935 Remove superfluous @code markup in items.
24937 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24939 * diagnostic-show-locus.c (struct column_range): New struct.
24940 (get_affected_columns): New function.
24941 (get_printed_columns): New function.
24942 (struct correction): New struct.
24943 (correction::ensure_capacity): New function.
24944 (correction::ensure_terminated): New function.
24945 (struct line_corrections): New struct.
24946 (line_corrections::~line_corrections): New dtor.
24947 (line_corrections::add_hint): New function.
24948 (layout::print_trailing_fixits): Reimplement in terms of the new
24950 (selftest::test_overlapped_fixit_printing): New function.
24951 (selftest::diagnostic_show_locus_c_tests): Call it.
24953 2017-05-03 Nathan Sidwell <nathan@acm.org>
24955 Canonicalize canonical type hashing
24956 * tree.h (type_hash_canon_hash): Declare.
24957 * tree.c (type_hash_list, attribute_hash_list): Move into
24958 type_hash_canon_hash.
24959 (build_type_attribute_qual_variant): Break out hash code calc into
24960 type_hash_canon_hash.
24961 (type_hash_canon_hash): New. Generic type hash computation.
24962 (build_range_type_1, build_array_type_1, build_function_type,
24963 build_method_type_directly, build_offset_type, build_complex_type,
24964 make_vector_type): Call it.
24966 2017-05-03 Richard Biener <rguenther@suse.de>
24968 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24969 When all DRs have unknown misaligned do not always peel
24970 when there is a store but apply the same costing model as if
24971 there were only loads.
24973 2017-05-03 Richard Biener <rguenther@suse.de>
24976 PR tree-optimization/80492
24977 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24978 compare_base_decls returning dont-know properly.
24980 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24982 * config/arm/iterators.md (CCSI): New mode iterator.
24983 (arch): New mode attribute.
24984 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24985 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24986 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24987 code iterator for success result mode.
24988 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24989 the corresponding new insn generators.
24991 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24994 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24995 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24997 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24999 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25000 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25001 (DDR_A): Wrap DDR argument in brackets.
25002 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25003 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25004 (DDR_REVERSED_P): Likewise.
25006 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25008 PR tree-optimization/79472
25009 * tree-switch-conversion.c (struct switch_conv_info): Add
25010 contiguous_range and default_case_nonstandard fields.
25011 (collect_switch_conv_info): Compute contiguous_range and
25012 default_case_nonstandard fields, don't clear final_bb if
25013 contiguous_range and only the default case doesn't have the required
25015 (check_all_empty_except_final): Set default_case_nonstandard instead
25016 of failing if contiguous_range and the default case doesn't have empty
25018 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25019 and only the default case doesn't have the required constants. Skip
25021 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25022 if default_case_nonstandard.
25023 (build_constructors): Build constant 1 just once. Assert that default
25024 values aren't inserted in between cases if contiguous_range. Skip
25026 (build_arrays): Skip virtual phis.
25027 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25028 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25029 Handle virtual phis.
25030 (gen_inbound_check): Handle default_case_nonstandard case.
25031 (process_switch): Adjust check_final_bb caller. Call
25032 gather_default_values with the first non-default case instead of
25033 default case if default_case_nonstandard.
25035 2017-05-02 Nathan Sidwell <nathan@acm.org>
25037 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25038 check. Fix formatting.
25040 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25042 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25043 errors when comparing specialized and unspecialized times.
25045 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25047 * diagnostic-show-locus.c
25048 (layout::should_print_annotation_line_p): Make private.
25049 (layout::print_annotation_line): Make private.
25050 (layout::annotation_line_showed_range_p): Make private.
25051 (layout::show_ruler): Make private.
25052 (layout::print_source_line): Make private. Pass in line and
25053 line_width, rather than calling location_get_source_line. Drop
25055 (layout::print_leading_fixits): New method.
25056 (layout::print_any_fixits): Rename to...
25057 (layout::print_trailing_fixits): ...this, and make private.
25058 Don't print newline fixits.
25059 (diagnostic_show_locus): Move logic for printing one row into...
25060 (layout::print_line): ...this new function. Move the
25061 location_get_source_line call and error-handling from
25062 print_source_line to here. Call print_leading_fixits, and rename
25063 print_any_fixits to print_trailing_fixits.
25064 (selftest::test_fixit_insert_containing_newline): Update now that
25065 newlines are partially supported.
25066 (selftest::test_fixit_insert_containing_newline_2): New test.
25067 (selftest::test_fixit_replace_containing_newline): Update comments.
25068 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25069 * edit-context.c (class added_line): New class.
25070 (class edited_line): Describe newline handling in comment.
25071 (edited_line::actually_edited_p): New method.
25072 (edited_line::print_content): Delete redundant decl.
25073 (edited_line::m_predecessors): New field.
25074 (edited_file::print_content): Call edited_line::print_content.
25075 (edited_file::print_diff): Update to support newlines.
25076 (edited_file::print_diff_hunk): Likewise.
25077 (edited_file::print_run_of_changed_lines): New function.
25078 (edited_file::print_diff_line): Convert to...
25079 (print_diff_line): ...this.
25080 (edited_file::get_effective_line_count): New function.
25081 (edited_line::edited_line): Initialize new field m_predecessors.
25082 (edited_line::~edited_line): Clean up m_predecessors.
25083 (edited_line::apply_fixit): Handle newlines.
25084 (edited_line::get_effective_line_count): New function.
25085 (edited_line::print_content): New function.
25086 (edited_line::print_diff_lines): New function.
25087 (selftest::test_applying_fixits_insert_containing_newline): New
25089 (selftest::test_applying_fixits_replace_containing_newline): New
25091 (selftest::insert_line): New function.
25092 (selftest::test_applying_fixits_multiple_lines): Add example of
25094 (selftest::edit_context_c_tests): Call the new tests.
25096 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25098 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25099 parameter cand. Update dump information.
25100 (get_computation_cost): Update uses.
25102 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25104 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25105 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25106 (get_computation_at, rewrite_use_address): Update use of
25107 get_computation_aff.
25109 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25111 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25112 (get_computation): Delete.
25113 (get_computation_cost): Implement like get_computation_cost_at.
25114 Use get_computation_at.
25115 (get_computation_cost_at): Delete.
25116 (rewrite_use_nonlinear_expr): Use get_computation_at.
25117 (rewrite_use_compare, remove_unused_ivs): Ditto.
25119 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25121 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25123 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25125 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25126 (ivopts_global_cost_for_size): Rename parameter and update uses.
25127 (iv_ca_recount_cost): Update uses.
25128 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25129 candidates seperately in n_invs and n_cands.
25130 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25132 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25134 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25135 (find_inv_vars_cb): New.
25136 (find_depends): Renamed to ...
25137 (find_inv_vars): ... this.
25138 (add_candidate_1, force_var_cost): Call find_inv_vars.
25139 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25141 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25143 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25144 inv_vars. Add inv_exprs.
25145 (struct iv_cand): Rename depends_on to inv_vars.
25146 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25147 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25148 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25149 (dump_cand): Dump inv_vars.
25150 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25151 (record_invariant, find_depends, add_candidate_1): Ditto.
25152 (set_group_iv_cost, force_var_cost): Ditto.
25153 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25154 (get_computation_cost_at, get_computation_cost): Ditto.
25155 (determine_group_iv_cost_generic): Ditto.
25156 (determine_group_iv_cost_address): Ditto.
25157 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25158 (determine_group_iv_costs): Ditto.
25159 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25160 (iv_ca_set_remove_invariants): Renamed to ...
25161 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25162 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25163 (iv_ca_set_add_invariants): Renamed to ...
25164 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25165 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25166 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25167 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25168 (create_new_ivs): Remove useless dump.
25170 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25172 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25174 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25175 (iv_ca_set_no_cp, create_new_iv): Ditto.
25177 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25179 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25181 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25183 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25184 function tree_check2.
25186 2017-05-02 Martin Liska <mliska@suse.cz>
25188 * doc/gcov.texi: Add missing preposition.
25189 * gcov.c (function_info::function_info): Properly fill up
25190 all member variables.
25192 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25194 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25196 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25198 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25200 2017-05-02 Martin Liska <mliska@suse.cz>
25203 * lto-streamer-in.c (lto_read_tree_1): Remove
25204 LTO_STREAMER_DEBUG.
25205 * lto-streamer.c (struct tree_hash_entry): Likewise.
25206 (struct tree_entry_hasher): Likewise.
25207 (tree_entry_hasher::hash): Likewise.
25208 (tree_entry_hasher::equal): Likewise.
25209 (lto_streamer_init): Likewise.
25210 (lto_orig_address_map): Likewise.
25211 (lto_orig_address_get): Likewise.
25212 (lto_orig_address_remove): Likewise.
25213 * lto-streamer.h: Likewise.
25214 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25215 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25217 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25219 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25220 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25221 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25222 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25223 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25224 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25225 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25226 (mm_maskz_sub_ss): New intrinsics.
25227 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25228 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25229 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25230 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25231 (__builtin_ia32_subss_mask_round): New builtins.
25232 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25233 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25234 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25236 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25237 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25239 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25242 2017-05-02 Martin Jambor <mjambor@suse.cz>
25244 PR tree-optimization/78687
25245 * tree-sra.c (access): New field parent.
25246 (process_subtree_disqualification): New function.
25247 (disqualify_candidate): Call it.
25248 (build_accesses_from_assign): Reset write flag if creating an
25250 (build_access_subtree): Fill in parent field and also prpagate
25251 down grp_write flag.
25252 (create_artificial_child_access): New parameter set_grp_write, set
25253 grp_write to its value.
25254 (propagate_subaccesses_across_link): Also propagate grp_write flag
25256 (propagate_all_subaccesses): Push the closest parent back to work
25257 queue if add_access_to_work_queue returned true.
25259 2017-05-02 Richard Biener <rguenther@suse.de>
25261 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25262 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25263 -fstrict-overflow documentation.
25264 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25265 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25266 flag_strict_overflow.
25267 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25268 * lto-opts.c (lto_write_options): Do not stream it.
25269 * lto-wrapper.c (merge_and_complain): Do not handle it.
25270 * opts.c (default_options_table): Do not set -fstrict-overflow.
25271 (finish_options): Likewise do not clear it when sanitizing.
25272 * simplify-rtx.c (simplify_const_relational_operation): Do not
25273 test flag_strict_overflow.
25275 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25277 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25278 using enabled attribute.
25279 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25280 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25281 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25282 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25283 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25284 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25285 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25286 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25287 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25288 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25290 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25292 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25294 2017-05-02 Richard Biener <rguenther@suse.de>
25296 PR tree-optimization/80591
25298 2017-04-10 Richard Biener <rguenther@suse.de>
25300 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25303 2017-05-02 Richard Biener <rguenther@suse.de>
25305 PR tree-optimization/80549
25306 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25307 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25308 headers if they do not have a preheader.
25310 2017-05-02 Martin Liska <mliska@suse.cz>
25313 * common.opt: Fix typo.
25314 * doc/invoke.texi: Likewise.
25316 2017-05-01 Jan Beulich <jbeulich@suse.com>
25318 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25319 swapping, add (x,x,m,x,n) alternative.
25321 2017-05-01 Nathan Sidwell <nathan@acm.org>
25323 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25324 unnecessary unadjusted_alignment check.
25326 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25329 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25331 * gimplify.c (gimplify_cilk_detach): New function.
25332 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25333 * tree-core.h: Document EXPR_CILK_SPAWN.
25334 * tree.h (EXPR_CILK_SPAWN): Define.
25336 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25338 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25339 to use new fixit_hint representation, using the "replace" logic.
25340 (get_line_span_for_fixit_hint): Likewise.
25341 (layout::print_any_fixits): Likewise.
25342 (selftest::test_one_liner_many_fixits): Rename to...
25343 (selftest::test_one_liner_many_fixits_1): ...this, and update
25344 comment and expected output to reflect that the multiple fix-it
25345 hints are now consolidated into one insertion.
25346 (selftest::test_one_liner_many_fixits_2): New test.
25347 (selftest::test_diagnostic_show_locus_one_liner): Update for
25349 (selftest::test_fixit_consolidation): Update for fix-it API
25351 * diagnostic.c (print_parseable_fixits): Likewise.
25352 * edit-context.c (edited_line::m_line_events): Convert from
25353 auto_vec <line_event *> to auto_vec <line_event>.
25354 (class line_event): Convert from abstract base class to a concrete
25355 class, taking over the role of replace_event.
25356 (class insert_event): Delete.
25357 (class replace_event): Rename to class line_event. Convert to
25359 (edit_context::add_fixits): Reimplement.
25360 (edit_context::apply_insert): Delete.
25361 (edit_context::apply_replace): Rename to...
25362 (edit_context::apply_fixit): ...this. Convert to half-open range.
25363 (edited_file::apply_insert): Delete.
25364 (edited_file::apply_replace): Rename to...
25365 (edited_file::apply_fixit): ...this.
25366 (edited_line::~edited_line): Drop deletion of events.
25367 (edited_line::apply_insert): Delete.
25368 (edited_line::apply_replace): Rename to...
25369 (edited_line::apply_fixit): ...this. Convert to half-open range.
25370 Update for change to type of m_line_events.
25371 * edit-context.h (edit_context::apply_insert): Delete.
25372 (edit_context::apply_replace): Rename to...
25373 (edit_context::apply_fixit): ...this.
25375 2017-05-01 Martin Sebor <msebor@redhat.com>
25377 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25380 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25383 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25384 __get_cpuid_max returns 0.
25385 (__get_cpuid_count): Ditto.
25387 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25389 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25390 replacement expression is another instance of one of its arguments.
25392 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25395 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25396 check for stack push/pop autoinc.
25397 * config/i386/i386.c (ix86_agi_dependent): Return false
25398 if the only reason why modified_in_p returned true is that
25399 addr is SP based and set_insn is a push or pop.
25401 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25403 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25406 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25409 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25410 (account_size_time): Use two predicates - exec_pred and
25412 (evaluate_conditions_for_known_args): Compute both clause and
25414 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25415 (inline_summary_t::duplicate): Update.
25416 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25418 (compute_inline_parameters): Likewise.
25419 (estimate_edge_size_and_time): Update caluclation of time.
25420 (estimate_node_size_and_time): Compute both time and nonspecialized
25422 (estimate_ipcp_clone_size_and_time): Update.
25423 (inline_merge_summary): Update.
25424 (do_estimate_edge_time): Update.
25425 (do_estimate_edge_size): Update.
25426 (do_estimate_edge_hints): Update.
25427 (inline_read_section, inline_write_summary): Stream both new predicates.
25428 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25430 (compute_inlined_call_time): Cleanup.
25431 (big_speedup_p): Update.
25432 (edge_badness): Update.
25433 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25434 (size_time_entry): Replace predicate by exec_predicate and
25435 nonconst_predicate.
25436 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25437 (estimate_edge_time): Return also nonspec_time.
25438 (reset_edge_growth_cache): Update.
25440 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25442 PR rtl-optimization/80491
25443 * ifcvt.c (noce_process_if_block): When looking for x setter
25444 with missing else_bb, don't check only the insn right before
25445 cond_earliest, but look for the last insn that x is modified in
25446 within the same bb.
25448 PR rtl-optimization/80491
25449 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25451 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25453 PR tree-optimization/80487
25454 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25456 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25458 PR tree-optimization/79697
25459 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25460 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25461 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25463 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25464 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25466 2017-04-28 Martin Sebor <msebor@redhat.com>
25468 PR tree-optimization/80523
25469 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25470 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25472 (maybe_warn, format_directive, parse_directive): Use new functions.
25473 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25475 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25477 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25479 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25481 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25482 target_header_dir): Set correctly.
25483 * configure: Regenerated.
25484 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25485 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25486 instead of SYSTEM_HEADER_DIR.
25488 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25490 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25491 (estimate_local_effects): Likewise.
25492 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25493 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25494 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25495 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25496 do_estimate_edge_time, estimate_edge_time): Likewise.
25497 * ipa-inline-analysis.c (estimate_node_size_and_time,
25498 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25499 (estimate_time_after_inlining): Remove.
25501 2017-04-28 Martin Liska <mliska@suse.cz>
25503 * doc/gcov.texi: Enhance documentation of gcov.
25505 2017-04-28 Martin Liska <mliska@suse.cz>
25507 * doc/gcov.texi: Sort options in alphabetic order.
25508 * doc/gcov-dump.texi: Likewise.
25509 * doc/gcov-tool.texi: Likewise.
25510 * gcov.c (print_usage): Likewise.
25511 * gcov-dump.c (print_usage): Likewise.
25512 * gcov-tool.c (print_merge_usage_message): Likewise.
25513 (print_rewrite_usage_message): Likewise.
25514 (print_overlap_usage_message): Likewise.
25516 2017-04-28 Martin Liska <mliska@suse.cz>
25518 PR gcov-profile/53915
25519 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25521 2017-04-28 Martin Liska <mliska@suse.cz>
25523 PR gcov-profile/79891
25524 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25525 is marked by compiler as living on a line.
25526 (get_cycles_count): Remove usage of the union.
25527 (output_intermediate_file): Likewise.
25528 (find_source): Fix GNU coding style.
25529 (accumulate_line_counts): Remove old non-all block mode.
25530 (output_lines): Remove usage of the union.
25531 * profile.c (output_location): Include all BBs, even if
25532 belonging to a same line (and file) as a previous BB.
25534 2017-04-28 Martin Liska <mliska@suse.cz>
25536 * gcov.c (process_args): Handle new argument 'w'.
25537 (read_graph_file): Assign ID to BBs.
25538 (output_branch_count): Display BB # if verbose flag is set.
25539 (output_lines): Likewise for arcs.
25540 (print_usage): Add '--verbose' option help.
25541 * doc/gcov.texi: Document --verbose (-w) option.
25543 2017-04-28 Martin Liska <mliska@suse.cz>
25545 * gcov.c (struct block_location_info): New struct.
25546 (process_file): Fill up the new structure.
25547 (read_graph_file): Replace usage of encoding by the newly added
25549 (add_line_counts): Likewise.
25550 (accumulate_line_counts): Remove usage of the union.
25551 (function_info::function_info): New function.
25552 (function_info::~function_info): Likewise.
25553 (process_file): Call delete instead of release_function.
25554 (release_function): Release the function.
25555 (release_structures): Call delete instead of release_function.
25556 (solve_flow_graph): Replace usage of num_blocks.
25557 (find_exception_blocks): Likewise.
25558 (output_lines): Fix GNU coding style.
25560 2017-04-28 Martin Liska <mliska@suse.cz>
25563 * coverage.c (coverage_remove_note_file): New function.
25564 * coverage.h: Declare the function.
25565 * toplev.c (finalize): Clean if an error has been seen.
25567 2017-04-28 Martin Liska <mliska@suse.cz>
25569 PR gcov-profile/80031
25570 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25571 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25572 * gcov.c (read_graph_file): Read just number of blocks.
25573 * profile.c (branch_prob): Do not stream 0 flags per a basic
25576 2017-04-28 Martin Liska <mliska@suse.cz>
25578 * gcov-dump.c (tag_*): Add new argument to declarations.
25579 (dump_gcov_file): Likewise.
25580 (tag_blocks): Add and use new argument depth.
25581 (tag_arcs): Likewise.
25582 (tag_lines): Likewise.
25583 (tag_counters): Likewise.
25584 (tag_summary): Likewise.
25585 (dump_working_sets): Use depth to do a proper indentation.
25587 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25590 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25591 * symtab.c (symtab_node::debug_symtab): Move definition here.
25593 2017-04-28 Richard Biener <rguenther@suse.de>
25595 * lto-streamer.h (LTO_major_version): Bump to 7.
25597 2017-04-28 Richard Biener <rguenther@suse.de>
25599 * tree-vrp.c (assert_info): New struct.
25600 (add_assert_info): New helper.
25601 (register_edge_assert_for_2): Refactor to add asserts to a vector
25603 (register_edge_assert_for_1): Likewise.
25604 (register_edge_assert_for): Likewise.
25605 (finish_register_edge_assert_for): New helper actually registering
25606 asserts where live on edge.
25607 (find_conditional_asserts): Adjust.
25608 (find_switch_asserts): Likewise.
25609 (evrp_dom_walker::try_find_new_range): Generalize.
25610 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25612 2017-04-27 Marek Polacek <polacek@redhat.com>
25615 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25616 arg10 and arg11 to itype.
25618 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25620 * doc/extend.texi (Object Size Checking): Improve grammar.
25622 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25625 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25626 that the logic for permitting reciprocal estimates matches that
25629 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25632 * tree.c (type_cache_hasher::equal): Only compare
25633 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25634 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25635 non-aggregate element types.
25636 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25637 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25639 2017-04-27 Richard Biener <rguenther@suse.de>
25641 PR middle-end/80533
25642 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25643 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25644 keeping a reference to a trailing array.
25646 2017-04-27 Richard Biener <rguenther@suse.de>
25648 PR middle-end/80539
25649 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25650 being in loop-closed SSA form conservatively.
25651 (chrec_fold_multiply_poly_poly): Likewise.
25653 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25655 PR middle-end/79665
25656 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25657 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25659 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25662 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25663 (aarch64_function_arg_alignment): Return unsigned int again, but still
25664 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25665 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25666 Don't emit -Wpsabi note.
25667 (aarch64_function_arg_boundary): Likewise.
25668 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25671 2017-04-26 Nathan Sidwell <nathan@acm.org>
25673 * tree.h (crc32_unsigned_n): Declare.
25674 (crc32_unsigned, crc32_unsigned): Make inline.
25675 * tree.c (crc32_unsigned_bits): Replace with ...
25676 (crc32_unsigned_n): ... this.
25677 (crc32_unsigned, crc32_byte): Remove.
25678 (crc32_string): Remove unnecessary braces.
25680 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25682 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25683 * ipa-inline-analysis.c (MAX_TIME): Remove.
25684 (account_size_time): Use sreal for time.
25685 (dump_inline_summary): Update.
25686 (estimate_function_body_sizes): Update.
25687 (estimate_edge_size_and_time): Update.
25688 (estimate_calls_size_and_time): Update.
25689 (estimate_node_size_and_time): Update.
25690 (inline_merge_summary): Update.
25691 (inline_update_overall_summary): Update.
25692 (estimate_time_after_inlining): Update.
25693 (inline_read_section): Update.
25694 (inline_write_summary): Update.
25695 * ipa-inline.c (compute_uninlined_call_time): Update.
25696 (compute_inlined_call_time): Update.
25697 (recursive_inlining): Update.
25698 (inline_small_functions): Update.
25699 (dump_overall_stats): Update.
25700 * ipa-inline.h: Include sreal.h.
25701 (size_time_entry): Turn time to sreal.
25702 (inline_summary): Turn self_time nad time to sreal.
25704 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25706 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25708 (sreal::stream_out, sreal::stream_in): New.
25709 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25711 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25713 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25716 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25719 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25721 (dimode_scalar_chain::compute_convert_gain): Ditto.
25722 (dimode_scalar_chain::make_vector_copies): Ditto.
25723 (dimode_scalar_chain::convert_reg): Ditto.
25724 (dimode_scalar_chain::convert_insn): Ditto.
25725 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25726 (VI248_AVX512BW_1): New mode iterator.
25727 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25728 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25731 2017-04-25 Martin Sebor <msebor@redhat.com>
25733 PR tree-optimization/80497
25734 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25735 constants are representable in HOST_WIDE_INT.
25736 (parse_directive): Ditto.
25738 2017-04-25 Martin Sebor <msebor@redhat.com>
25741 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25742 (new_zero_array): Adjust signature.
25743 (dom_info::dom_init): Used unsigned rather that size_t.
25744 (dom_info::dom_info): Same.
25746 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25747 Jakub Jelinek <jakub@redhat.com>
25750 * config/arm/arm.c: Include gimple.h.
25751 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25752 returns negative, increment ncrn only if it returned positive.
25753 (arm_needs_doubleword_align): Return int instead of bool,
25754 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25755 members, but if there is any such non-FIELD_DECL
25756 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25757 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25758 returns negative, increment nregs only if it returned positive.
25759 (arm_setup_incoming_varargs): Likewise.
25760 (arm_function_arg_boundary): Emit -Wpsabi note if
25761 arm_needs_doubleword_align returns negative, return
25762 DOUBLEWORD_ALIGNMENT only if it returned positive.
25764 2017-04-25 Marek Polacek <polacek@redhat.com>
25767 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25768 first argument to type.
25770 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25773 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25774 type checks to test for compatibility instead of equality.
25776 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25777 Jakub Jelinek <jakub@redhat.com>
25780 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25782 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25783 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25784 the alignment computation, but return their maximum in warn_alignment.
25785 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25786 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25788 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25789 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25792 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25794 * config/arc/simdext.md (dmpyh): Fix typo.
25796 2017-04-25 Richard Biener <rguenther@suse.de>
25798 PR tree-optimization/80492
25799 * alias.c (compare_base_decls): Handle registers with asm
25800 specification conservatively.
25801 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25802 compare_base_decls returning dont-know properly.
25804 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25806 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25807 (legitimate_offset_address_p): New function.
25808 (arc_legitimate_address_p): Use above function.
25810 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25812 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25814 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25816 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25817 ACCH registers whenever they are available.
25819 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25821 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25822 double regs fix when not used.
25824 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25826 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25828 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25829 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25831 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25833 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25834 when emitting short ADD instructions.
25836 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25838 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25840 (cmpsi_cc_c_insn): Likewise.
25841 (cbranchsi4_scratch): Compute proper instruction length using
25842 compact_hreg_operand.
25843 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25845 2017-04-25 Richard Biener <rguenther@suse.de>
25847 PR middle-end/80509
25848 * passes.c (pass_manager::pass_manager): Initialize
25849 m_name_to_pass_map.
25851 2017-04-25 Richard Biener <rguenther@suse.de>
25853 PR tree-optimization/79201
25854 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25856 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25859 * config/s390/vector.md: Split MEM->GPR vector moves for
25860 non-s_operand addresses.
25862 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25865 * config/s390/predicates.md (reload_const_wide_int_operand): New
25867 * config/s390/s390.md ("movti"): Remove d/P alternative.
25868 ("movti_bigconst"): New pattern definition.
25870 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25873 * s390-protos.h (s390_expand_cs_hqi): Removed.
25874 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25875 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25876 modes as well as CCZ1mode and CCZmode.
25877 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25878 signature of s390_emit_compare_and_swap.
25879 (s390_expand_cs_hqi): Likewise, make static.
25880 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25881 compare-and-swap, in some cases.
25882 (s390_expand_cs): Wrapper function.
25883 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25885 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25886 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25887 patterns for small and large integers. Forbid symref memory operands.
25888 Move expander to s390.c. Require cc register.
25889 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25890 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25891 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25892 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25893 symref memory operands. Remove CC mode and call s390_match_ccmode
25895 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25897 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25899 * config/s390/s390.md (define_peephole2): New peephole to help
25900 combining the load-and-test pattern with volatile memory.
25902 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25904 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25905 with CCZmode for TARGET_Z196.
25907 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25909 PR rtl-optimization/80501
25910 * combine.c (make_compound_operation_int): Set subreg_code to SET
25911 even for AND with mask of the sign bit of mode.
25913 PR rtl-optimization/80500
25914 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25915 sum's initial value.
25917 2017-04-25 Julian Brown <julian@codesourcery.com>
25918 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25920 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25922 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25924 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25926 2017-04-25 Julian Brown <julian@codesourcery.com>
25927 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25929 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25930 (thunderx2t99_sha): New Reservation.
25932 2017-04-25 Julian Brown <julian@codesourcery.com>
25933 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25935 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25936 type for 1-element load.
25938 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25940 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25942 2017-04-24 Martin Jambor <mjambor@suse.cz>
25944 PR tree-optimization/80293
25945 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25946 char arrays not totally scalarizable if it is false.
25947 (analyze_all_variable_accesses): Pass correct value in the new
25948 parameter. Add a statistics counter.
25950 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25952 PR middle-end/79931
25953 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25955 2017-04-24 Richard Biener <rguenther@suse.de>
25957 PR tree-optimization/80494
25958 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25959 out for complex types.
25961 2017-04-24 Richard Biener <rguenther@suse.de>
25963 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25964 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25965 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25966 (DFS): Adjust and never fail.
25967 (sccvn_dom_walker::fail): Remove.
25968 (sccvn_dom_walker::before_dom_children): Adjust.
25969 (run_scc_vn): Likewise and never fail.
25970 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25971 (pass_fre::execute): Likewise.
25973 2017-04-24 Richard Biener <rguenther@suse.de>
25975 PR tree-optimization/79725
25976 * tree-ssa-sink.c (statement_sink_location): Return whether
25977 failure reason was zero uses. Move that check later.
25978 (sink_code_in_bb): Deal with zero uses by removing the stmt
25981 2017-04-24 Richard Biener <rguenther@suse.de>
25984 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25985 pointer-based references.
25987 2017-04-24 Richard Biener <rguenther@suse.de>
25990 * pass_manager.h (pass_manager::operator new): Remove.
25991 (pass_manager::operator delete): Likewise.
25992 * passes.c (pass_manager::operator new): Remove.
25993 (pass_manager::operator delete): Likewise.
25994 (pass_manager::pass_manager): Zero individual pass members.
25996 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25999 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26000 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26001 Check "XEXP (src, 1)" operand here.
26002 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26003 Check "XEXP (src, 1)" operand here.
26004 (dimode_scalar_chain::make_vector_copies): Detect count register
26005 of a shift instruction. Zero extend count register from QImode
26006 to DImode to satisfy vector shift pattern count operand predicate.
26007 Substitute vector shift count operand with a DImode copy.
26008 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26011 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26013 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26014 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26015 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26016 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26017 (UNSPEC_NOREX_MEM): Remove definition.
26019 2017-04-21 Richard Biener <rguenther@suse.de>
26021 PR tree-optimization/79547
26022 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26023 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26024 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26025 without any constraints.
26027 2017-04-21 Richard Biener <rguenther@suse.de>
26029 PR tree-optimization/78847
26030 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26032 2017-04-21 Richard Biener <rguenther@suse.de>
26034 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26035 (build_distinct_type_copy): Likewise.
26036 (build_variant_type_copy): Likewise.
26037 * tree.c (build_qualified_type): Pass down mem-stat info.
26038 (build_distinct_type_copy): Likewise.
26039 (build_variant_type_copy): Likewise.
26041 2017-04-21 Richard Biener <rguenther@suse.de>
26043 PR tree-optimization/80237
26044 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26046 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26047 for a simplified result.
26049 2016-04-21 Richard Biener <rguenther@suse.de>
26051 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26052 sth as strict as a simple_iv but a chrec without symbols and an
26053 operand defined in the loop we are peeling (and not some subloop).
26054 (propagate_constants_for_unrolling): Propagate all constants.
26056 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26059 * config/i386/i386.c (print_reg): Remove assert for disalowed
26060 regno values, call output_operand_lossage instead.
26062 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26065 * config/i386/constraints.md (Yc): New register constraint.
26066 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26067 Use Yc constraint for alternative 2 of operand 0. Remove
26068 preferred_for_speed attribute.
26070 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26072 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26073 lastprivate clauses in SIMT case.
26075 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26077 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26079 2017-04-20 Richard Biener <rguenther@suse.de>
26081 PR tree-optimization/57796
26082 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26083 as N scalar stores.
26084 (vect_model_load_cost): Cost gathers as N scalar loads.
26086 2017-04-20 Richard Biener <rguenther@suse.de>
26088 * ggc-page.c (ggc_allocated_p): Rename to ...
26089 (safe_lookup_page_table_entry): ... this and return the lookup
26091 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26093 2017-04-20 Richard Biener <rguenther@suse.de>
26095 PR tree-optimization/80453
26096 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26097 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26098 from the conditions.
26099 (vn_phi_eq): Pass them down.
26100 (vn_phi_lookup): Record them.
26101 (vn_phi_insert): Likewise.
26103 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26105 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26106 uninitialized variable warning to avoid buffer overrun.
26108 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26111 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26112 is suppressed for '{ 0 }' in C.
26114 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26116 * BASE-VER: Set to 8.0.0.
26118 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26120 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26121 priority .init_array and .fini_array section with SECTION_NOTYPE
26124 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26126 PR middle-end/80423
26127 * tree.h (build_array_type): Add typeless_storage default argument.
26128 * tree.c (type_cache_hasher::equal): Also compare
26129 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26130 (build_array_type): Add typeless_storage argument, set
26131 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26133 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26134 (build_array_type): Likewise. Add typeless_storage argument.
26136 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26137 Jakub Jelinek <jakub@redhat.com>
26139 PR tree-optimization/80426
26140 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26141 operation on symbolic operands, also compute the overflow for the
26142 invariant part when the operation degenerates into a negation.
26144 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26147 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26148 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26151 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26153 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26156 * config/avr/avr.c (tree.h): Include it.
26157 (cgraph.h): Include it.
26158 (avr_encode_section_info): Don't warn for uninitialized progmem
26159 variable if it's just an alias.
26161 2017-04-19 Richard Biener <rguenther@suse.de>
26164 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26165 when needed by AutoPGO.
26167 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26170 * doc/lto.texi: Remove an extra 'that'.
26172 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26174 PR rtl-optimization/80429
26175 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26176 are only used in debug insns.
26178 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26179 Vladimir Makarov <vmakarov@redhat.com>
26181 * config/sparc/predicates.md (input_operand): Add comment. Return
26182 true for any memory operand when LRA is in progress.
26183 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26185 2017-04-18 Jeff Law <law@redhat.com>
26188 * mips.md ({return,simple_return}_internal): Do not overwrite
26191 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26193 PR tree-optimization/80443
26194 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26195 instead of adding 1, subtract -1 and similarly instead of subtracting
26198 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26200 PR rtl-optimization/80357
26201 * haifa-sched.c (tmp_bitmap): New variable.
26202 (model_recompute): Handle duplicate use records.
26203 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26204 (free_global_sched_pressure_data): Free it.
26206 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26209 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26210 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26211 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26212 instead of SYSTEM_HEADER_DIR.
26214 2017-04-18 Jeff Law <law@redhat.com>
26216 PR middle-end/80422
26217 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26218 predecessors after walking up the insn chain.
26220 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26223 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26224 sizetype type into debug info.
26226 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26229 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26230 unneeded test for TARGET_UPPER_REGS_SF.
26231 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26233 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26236 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26237 instead of gsi_after_labels.
26239 2017-04-18 Jeff Law <law@redhat.com>
26241 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26245 2017-04-13 Jeff Law <law@redhat.com>
26246 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26247 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26249 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26252 * config/avr/avr.c (intl.h): Include it.
26253 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26255 2017-04-18 Martin Liska <mliska@suse.cz>
26257 PR gcov-profile/78783
26258 * gcov-tool.c (gcov_output_files): Validate that destination
26259 file is either removed by the tool or by a user.
26261 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26262 Guy Benyei <guybe@mellanox.com>
26264 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26265 block, and do not negate it, the stored id is already negative.
26267 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26269 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26271 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26274 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26275 masks of options that should be turned off if the VSX vector
26276 options are turned off.
26277 (OTHER_P8_VECTOR_MASKS): Likewise.
26278 (OTHER_VSX_VECTOR_MASKS): Likewise.
26279 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26280 rs6000_disable_incompatible_switches to validate no type switches
26282 (rs6000_incompatible_switch): New function to disallow turning on
26283 other vector options if -mno-vsx, -mno-power8-vector, or
26284 -mno-power9-vector are specified.
26286 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26288 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26290 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26292 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26293 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26294 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26295 (ARG_POINTER_CFA_OFFSET): Likewise.
26297 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26299 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26300 conditions to take advantage of various optimizations.
26302 2017-04-13 Jeff Law <law@redhat.com>
26304 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26305 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26306 (zero_extendsidi2_dext): Likewise.
26308 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26311 * fold-const.c (fold_ternary_loc): Revert
26312 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26315 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26317 PR rtl-optimization/80343
26318 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26319 new scratch pseudo.
26321 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26324 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26325 to ubsan_encode_value.
26327 2017-04-13 Jeff Law <law@redhat.com>
26329 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26330 appearing in DEBUG_INSNs.
26332 2017-04-13 Martin Liska <mliska@suse.cz>
26334 PR gcov-profile/80413
26335 * gcov-io.c (gcov_write_string): Copy to buffer just when
26336 allocated size is greater than zero.
26338 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26341 * dwarf2out.c (decls_for_scope): Ignore declarations of
26342 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26344 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26347 * ipa-visibility.c (non_local_p): Fix typos.
26348 (localize_node): When localizing symbol in same comdat group,
26349 dissolve the group only when we know external symbols are going
26351 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26353 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26355 PR tree-optimization/79390
26356 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26357 order does not result in usable sequence, retry with reversed operand
26363 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26364 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26365 op0 instead of fold_convert_loc (loc, type, arg0).
26367 2017-04-12 Jeff Law <law@redhat.com>
26369 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26370 has a delay slot in the generated code.
26372 * config/cris/cris.md (cris_preferred_reload_class): Return
26373 GENNONACR_REGS rather than GENERAL_REGS.
26375 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26378 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26379 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26380 signedness of the result type.
26382 2017-04-12 Richard Biener <rguenther@suse.de>
26383 Jeff Law <law@redhat.com>
26385 PR tree-optimization/80359
26386 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26387 trim stores to TARGET_MEM_REFs.
26389 2017-04-12 Richard Biener <rguenther@suse.de>
26391 PR tree-optimization/79390
26392 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26393 threading case even more.
26395 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26398 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26399 for quad_address_p for TImode, instead of just not indexed_address.
26401 2017-04-12 Richard Biener <rguenther@suse.de>
26402 Bernd Edlinger <bernd.edlinger@hotmail.de>
26404 PR middle-end/79671
26405 * alias.c (component_uses_parent_alias_set_from): Handle
26406 TYPE_TYPELESS_STORAGE.
26407 (get_alias_set): Likewise.
26408 * tree-core.h (tree_type_common): Add typeless_storage flag.
26409 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26410 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26411 for types containing members with TYPE_TYPELESS_STORAGE.
26412 (place_field): Likewise.
26413 (layout_type): Likewise for ARRAY_TYPE.
26414 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26415 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26416 TYPE_TYPELESS_STORAGE.
26417 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26419 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26422 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26423 first argument to type.
26425 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26429 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26430 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26431 (rs6000_expand_binop_builtin): Likewise.
26432 (rs6000_expand_ternop_builtin): Likewise; also add missing
26433 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26434 vshasigma built-ins.
26435 * doc/extend.texi: Document that vec_xxpermdi's third argument
26436 must be a constant.
26438 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26440 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26441 Use shift_const cost parameter when calculating gain of STV shifts.
26443 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26445 PR rtl-optimization/70478
26446 * lra-constraints.c (process_alt_operands): Check memory for
26447 disfavoring memory insn operand.
26449 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26451 PR middle-end/80100
26452 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26453 left shift in unsigned HOST_WIDE_INT type.
26455 PR rtl-optimization/80385
26456 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26457 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26460 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26461 if they have any depend clauses.
26463 2017-04-11 Martin Liska <mliska@suse.cz>
26466 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26467 * ipa-split.c (split_function): Create a local comdat symbol
26468 if caller is in a comdat group.
26470 2017-04-11 Martin Liska <mliska@suse.cz>
26473 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26476 2017-04-11 Martin Sebor <msebor@redhat.com>
26478 PR middle-end/80364
26479 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26480 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26482 (directive::set_width, directive::set_precision, format_character):
26484 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26487 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26490 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26491 conflict, set target->arch_name instead of target->cpu_name.
26493 2017-04-11 Richard Biener <rguenther@suse.de>
26495 PR tree-optimization/80374
26496 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26497 build_zero_cst, remove fold_convertible_p check again.
26499 2017-04-11 Martin Liska <mliska@suse.cz>
26502 * ubsan.c (instrument_object_size): Do not instrument register
26505 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26508 * config/i386/i386-builtin-types.def
26509 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26510 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26511 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26512 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26513 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26514 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26515 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26516 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26517 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26518 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26519 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26520 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26521 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26522 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26523 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26524 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26525 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26526 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26527 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26528 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26529 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26530 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26531 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26532 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26533 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26534 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26535 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26536 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26537 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26538 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26539 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26540 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26541 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26542 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26543 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26544 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26545 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26546 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26547 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26548 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26549 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26550 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26551 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26552 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26553 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26554 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26556 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26557 flag to second_arg_count, handle 4 argument function type _COUNT
26558 aliases, handle second_arg_count on second argument rather than last.
26560 2017-04-10 Jeff Law <law@redhat.com>
26562 PR tree-optimization/80374
26563 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26564 record anything if we can not convert integer_zero_node to the
26567 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26570 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26571 Enhance special handling given to the TARGET_P9_MINMAX option in
26572 relation to certain other options.
26574 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26576 PR tree-optimization/80153
26577 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26578 remove POINTER_PLUS_EXPR's base part directly, rather than through
26581 2017-04-10 Richard Biener <rguenther@suse.de>
26582 Bin Cheng <bin.cheng@arm.com>
26584 PR tree-optimization/80153
26585 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26586 the first element of pointer type aff_tree. Build result expr in
26588 (add_elt_to_tree): Convert to type unconditionally. Remove other
26589 fold_convert calls.
26590 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26591 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26593 2017-04-10 Richard Biener <rguenther@suse.de>
26595 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26598 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26600 PR rtl-optimization/70478
26601 * lra-constraints.c (curr_small_class_check): New.
26602 (update_and_check_small_class_inputs): New.
26603 (process_alt_operands): Update curr_small_class_check. Disfavor
26604 alternative insn memory operands. Check available regs for small
26607 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26610 * config/mips/mips.opt (-mvirt): Update description.
26611 * doc/invoke.texi (-mvirt): Likewise.
26613 2017-04-10 Richard Biener <rguenther@suse.de>
26615 PR middle-end/80362
26616 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26617 looking for NEGATE_EXPR in -A / -B to A / B folding.
26619 2017-04-10 Martin Liska <mliska@suse.cz>
26621 PR gcov-profile/80224
26622 * gcov.c (print_usage): Fix usage string.
26623 (get_gcov_intermediate_filename): Remove.
26624 (output_gcov_file): Use both for normal and intermediate format.
26625 (generate_results): Do not initialize special file for
26626 intermediate format.
26628 2017-04-10 Richard Biener <rguenther@suse.de>
26630 PR tree-optimization/80304
26631 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26634 2017-04-10 Nathan Sidwell <nathan@acm.org>
26637 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26638 (rs6000_init_builtins): Use it.
26640 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26642 * config/arm/arm.md (<mrc>): Add mode to SET source.
26643 (<mrrc>): Likewise.
26645 2017-04-10 Richard Biener <rguenther@suse.de>
26647 PR middle-end/80344
26648 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26650 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26653 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26654 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26655 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26656 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26657 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26658 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26659 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26660 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26661 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26662 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26663 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26664 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26665 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26666 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26667 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26668 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26669 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26670 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26671 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26672 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26673 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26674 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26675 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26677 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26679 PR rtl-optimization/70478
26680 * lra-constraints.c: Reverse the last patch.
26682 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26684 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26685 Add comment for WCHAR_T.
26687 2017-04-08 Martin Liska <mliska@suse.cz>
26690 2017-04-07 Martin Liska <mliska@suse.cz>
26693 * ipa-split.c (split_function): Add function part to a same comdat
26696 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26699 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26701 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26703 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26704 (vec_pairing): ...to this.
26705 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26706 (rs6000_sched_init): Adjust for name change.
26707 (struct rs6000_sched_context): Likewise.
26708 (rs6000_init_sched_context): Likewise.
26709 (rs6000_set_sched_context): Likewise.
26711 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26717 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26719 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26720 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26721 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26723 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26725 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26727 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26729 PR rtl-optimization/70703
26730 * ira-color.c (update_conflict_hard_regno_costs): Use
26731 int64_t instead of HOST_WIDE_INT.
26733 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26735 PR rtl-optimization/70478
26736 * lra-constraints.c (process_alt_operands): Disfavor alternative
26737 insn memory operands.
26739 2017-04-07 Jeff Law <law@redhat.com>
26741 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26742 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26744 2017-04-07 Martin Liska <mliska@suse.cz>
26747 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26748 Show error message instead of an ICE.
26750 2017-04-07 Martin Liska <mliska@suse.cz>
26753 * ipa-split.c (split_function): Add function part to a same comdat
26756 2017-04-07 Richard Biener <rguenther@suse.de>
26758 PR middle-end/80341
26759 * tree.c (get_unwidened): Also handle ! for_type case for
26761 * convert.c (do_narrow): Split out from ...
26762 (convert_to_integer_1): ... here. Do not pass final truncation
26763 type to get_unwidened for TRUNC_DIV_EXPR.
26765 2017-04-07 Richard Biener <rguenther@suse.de>
26767 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26769 (aff_combination_const): Adjust.
26770 (aff_combination_scale): Likewise.
26771 (aff_combination_add_elt): Likewise.
26772 (aff_combination_add_cst): Likewise.
26773 (aff_combination_convert): Likewise.
26774 (add_elt_to_tree): Likewise. Remove unused argument.
26775 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26777 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26779 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26781 * config/arm/arm.c (arm_default_short_enums): Use
26782 ARM_DEFAULT_SHORT_ENUMS.
26783 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26785 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26788 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26789 members with redundant out-of-class redeclaration.
26791 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26794 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26795 * config/i386/i386.md (*zero_extendsidi2):
26796 Add (?*x,*x) and (?*v,*v) alternatives.
26798 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26801 * config/i386/i386.c (ix86_expand_builtin)
26802 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26803 mode from insn data. Convert operands to insn operand mode.
26804 Copy operands that don't satisfy insn predicate to a register.
26806 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26808 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26811 2017-04-06 Richard Biener <rguenther@suse.de>
26813 PR tree-optimization/80334
26814 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26815 preserve alignment of accesses.
26817 2017-04-06 Richard Biener <rguenther@suse.de>
26819 PR tree-optimization/80262
26820 * tree-sra.c (build_ref_for_offset): Preserve address-space
26822 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26823 Drop useless address-space information on MEM_REF offsets.
26825 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26827 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26829 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26831 PR rtl-optimization/70703
26832 * ira-color.c (update_conflict_hard_regno_costs): Use
26833 HOST_WIDE_INT instead of long.
26835 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26838 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26839 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26841 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26842 for x86_64 target. Handle -m3dnowa option.
26844 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26846 PR rtl-optimization/70703
26847 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26848 (update_conflict_hard_regno_costs): Use long instead of unsigned
26849 arithmetic for cost calculation.
26851 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26852 Bernd Edlinger <bernd.edlinger@hotmail.de>
26855 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26858 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26861 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26862 ptr_mode with Pmode throughout.
26863 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26864 into probe_stack_range and use DImode.
26866 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26869 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26870 call_eh_return is true.
26872 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26874 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26875 Initialize last_match_fntype_index.
26877 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26880 * tree-nvr.c: Include internal-fn.h.
26881 (pass_return_slot::execute): Ignore internal calls without
26884 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26885 Richard Biener <rguenther@suse.de>
26888 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26889 captures used multiple times, except for the last use.
26890 * generic-match-head.c: Include gimplify.h.
26892 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26894 PR tree-optimization/79390
26895 * target.h (struct noce_if_info): Declare.
26896 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26897 * target.def (noce_conversion_profitable_p): New target hook.
26898 * ifcvt.h (struct noce_if_info): New type, moved from ...
26899 * ifcvt.c (struct noce_if_info): ... here.
26900 (noce_conversion_profitable_p): Renamed to ...
26901 (default_noce_conversion_profitable_p): ... this. No longer
26903 (noce_try_store_flag_constants, noce_try_addcc,
26904 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26905 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26906 instead of noce_conversion_profitable_p.
26907 * config/i386/i386.c: Include ifcvt.h.
26908 (ix86_option_override_internal): Don't override
26909 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26910 (ix86_noce_conversion_profitable_p): New function.
26911 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26912 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26913 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26914 * doc/tm.texi: Regenerated.
26916 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26918 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26921 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26924 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26925 instructions for small multiply cores.
26927 2017-04-04 Jeff Law <law@redhat.com>
26929 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26931 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26932 that are not set by the loop over the elements.
26934 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26937 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26938 int mode, convert_modes it to mode as unsigned, otherwise use
26939 lowpart_subreg to mode rather than SImode.
26940 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26941 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26942 Use DImode instead of SImode for the shift count operand.
26943 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26946 2017-04-04 Richard Biener <rguenther@suse.de>
26948 PR middle-end/80281
26949 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26950 arithmetic done for the negate or the plus. Simplify.
26951 (A - (-B) -> A + B): Likewise.
26952 * fold-const.c (split_tree): Make sure to not negate pointers.
26954 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26956 PR rtl-optimization/60818
26957 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26958 a compare of comparisons with the thing compared if this results
26959 in a different machine mode.
26961 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26963 * alias.c (base_alias_check): Fix typo in comment.
26964 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26965 * cgraphunit.c (symbol_table::compile): Likewise.
26966 * collect2.c (maybe_run_lto_and_relink): Likewise.
26967 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26968 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26969 * config/avr/avr.c (avr_map_op_t): Likewise.
26970 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26971 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26972 * config/epiphany/epiphany.md (movcc): Likewise.
26973 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26974 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26976 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26977 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26978 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26979 * config/sh/sh.c (sh_rtx_costs): Likewise.
26980 * fold-const.c (fold_truth_andor): Likewise.
26981 * genautomata.c (collapse_flag): Likewise.
26982 * gengtype.h (struct type::u::s): Likewise.
26983 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26984 * input.c (FORMAT_AMOUNT): Likewise.
26985 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26986 (known_aggs_to_agg_replacement_list): Likewise.
26987 * ipa-inline-analysis.c: Likewise.
26988 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26989 * ipa-polymorphic-call.c
26990 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26991 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26992 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26994 * modulo-sched.c (apply_reg_moves): Likewise.
26995 * omp-expand.c (build_omp_regions_1): Likewise.
26996 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26997 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26998 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26999 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27000 * value-prof.c: Likewise.
27001 * var-tracking.c (val_reset): Likewise.
27003 2017-04-03 Richard Biener <rguenther@suse.de>
27005 PR tree-optimization/80275
27006 * fold-const.c (split_address_to_core_and_offset): Handle
27009 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27011 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27012 descriptors is at least equal to that of functions.
27014 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27016 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27018 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27021 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27022 (mov<IMOD4:mode>): New expander.
27023 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27025 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27027 PR rtl-optimization/79405
27028 * fwprop.c (propagations_left): New variable.
27029 (forward_propagate_into): Decrement it.
27030 (fwprop_init): Initialize it.
27031 (fw_prop): If the variable has reached zero, stop propagating.
27032 (fwprop_addr): Ditto.
27034 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27037 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27038 a FUNCTION_DECL, pass it as decl instead of origin to
27041 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27043 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27046 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27049 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27050 TARGET_VSX_SMALL_INTEGER.
27052 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27054 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27055 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27057 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27059 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27060 extraction from odd-numbered MSA register.
27062 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27064 PR middle-end/80173
27065 * expmed.c (store_bit_field_1): Don't attempt to create
27066 a word subreg out of hard registers wider than word if they
27067 have HARD_REGNO_NREGS of 1 for their mode.
27069 PR middle-end/80163
27070 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27071 conversions to integer types wider than word and pointer.
27074 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27075 (rtx_equal_for_cselib_p): Pass 0 to it.
27076 * cselib.c (cselib_hasher::equal): Likewise.
27077 (rtx_equal_for_cselib_1): Add depth argument. If depth
27078 is 128, don't look up VALUE locs and punt. Increment
27079 depth in recursive calls when walking VALUE locs.
27081 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27083 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27084 (make_gcov_file_name): Use the canonical path name for generating
27086 (read_line): Fix handling of files with ascii null bytes.
27088 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27090 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27091 to initialise a vector register instead
27092 of using a const_int.
27094 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27096 PR translation/80189
27097 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27098 diagnostic messages.
27100 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27103 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27104 (dfp_diex_<mode>): Update mode of operand 1.
27105 * doc/extend.texi (dxex, dxexq): Document change to return type.
27106 (diex, diexq): Document change to argument type.
27108 2017-03-30 Martin Jambor <mjambor@suse.cz>
27111 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27112 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27113 it reflects the signature changes performed at the callee side.
27114 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27115 to cgraph_build_function_type_skip_args.
27116 (build_function_decl_skip_args): Adjust call to the above function.
27118 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27121 * config/i386/sse.md
27122 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27123 register as dest whenever it is a MEM not rtx_equal_p to the
27124 corresponding dup operand, and when forcing into reg move the
27125 reg into the memory afterwards.
27126 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27127 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27128 for the force_reg mode.
27129 (avx512vl_vextractf128<mode>): Use register as dest either
27130 always when a MEM, or when it is a MEM not rtx_equal_p to the
27131 corresponding dup operand, or even not when it is a CONST_VECTOR
27132 depending on the mode and lo vs. hi.
27133 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27135 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27136 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27137 Likewise. Require that operands[2] is even.
27138 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27139 Remove extraneous parens. Require that operands[2] is a multiple
27141 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27142 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27143 disallow memory then.
27145 2017-03-30 Richard Biener <rguenther@suse.de>
27147 PR tree-optimization/77498
27148 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27149 to non-constants over backedges.
27151 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27153 PR rtl-optimization/80233
27154 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27155 as last_combined_insn. Do not test for BARRIER_P separately.
27157 2017-03-29 Andreas Schwab <schwab@suse.de>
27160 * calls.c (prepare_call_address): Convert funexp to Pmode before
27161 copying to temp reg.
27163 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27165 PR tree-optimization/80158
27166 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27167 Handle possible future case of more than one alternate
27169 (replace_rhs_if_not_dup): Likewise.
27170 (replace_one_candidate): Likewise.
27172 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27174 PR rtl-optimization/80193
27175 * ira.c (ira): Do not check allocation for LRA.
27177 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27179 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27180 (nvptx_output_simt_exit): Declare.
27181 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27182 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27183 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27184 (nvptx_declare_function_name): ...here. Emit declaration of local
27185 memory space buffer for omp_simt_enter insn.
27186 (nvptx_output_unisimt_switch): New.
27187 (nvptx_output_softstack_switch): New.
27188 (nvptx_output_simt_enter): New.
27189 (nvptx_output_simt_exit): New.
27190 * config/nvptx/nvptx.h (struct machine_function): New fields
27191 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27192 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27193 (UNSPECV_SIMT_EXIT): Ditto.
27194 (omp_simt_enter_insn): New insn.
27195 (omp_simt_enter): New expansion.
27196 (omp_simt_exit): New insn.
27197 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27199 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27200 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27201 (expand_GOMP_SIMT_EXIT): New.
27202 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27203 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27204 (GOMP_SIMT_EXIT): Ditto.
27205 * target-insns.def (omp_simt_enter): New insn.
27206 (omp_simt_exit): Ditto.
27207 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27209 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27210 (lower_rec_input_clauses): Likewise.
27211 (lower_lastprivate_clauses): Handle SIMT privatization.
27213 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27214 (ompdevlow_adjust_simt_enter): New.
27215 (find_simtpriv_var_op): New.
27216 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27217 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27219 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27220 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27221 (copy_decl_for_dup_finish): Ditto.
27223 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27225 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27228 * config/i386/i386.c (ix86_option_override_internal): Always
27229 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27231 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27233 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27235 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27237 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27238 mark new edge's irreducible flag accordign to it.
27239 (vect_do_peeling): Check loop preheader edge's irreducible flag
27240 and pass it to function slpeel_add_loop_guard.
27242 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27244 PR tree-optimization/80218
27245 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27246 Update block frequencies and counts.
27248 2017-03-28 Richard Biener <rguenther@suse.de>
27250 PR tree-optimization/78644
27251 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27252 of a simplification result we may not use it at all.
27254 2017-03-28 Richard Biener <rguenther@suse.de>
27257 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27258 without arguments, generate default definition of a SSA name.
27260 2017-03-28 Richard Biener <rguenther@suse.de>
27262 PR middle-end/80222
27263 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27264 TYPE_REF_CAN_ALIAS_ALL references.
27265 * fold-const.c (fold_indirect_ref_1): Likewise.
27267 2017-03-28 Martin Liska <mliska@suse.cz>
27270 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27271 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27273 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27274 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27276 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27277 (EXTRA_SPECS): Define.
27278 (SUBTARGET_EXTRA_SPECS): Likewise.
27279 (SUBTARGET_CPP_SPEC): Likewise.
27280 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27281 SUBTARGET_EXTRA_SPECS.
27282 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27284 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27286 * config/arc/simdext.md (vst64_insn): Update pattern.
27287 (vld32wh_insn): Likewise.
27288 (vld32wl_insn): Likewise.
27289 (vld64_insn): Likewise.
27290 (vld32_insn): Likewise.
27292 2017-03-28 Marek Polacek <polacek@redhat.com>
27295 * fold-const.c (fold_comparison): Use protected_set_expr_location
27296 instead of SET_EXPR_LOCATION.
27298 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27300 * tree.c (add_expr): Avoid name lookup warning.
27302 2017-03-27 Jeff Law <law@redhat.com>
27304 PR tree-optimization/80216
27305 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27306 function name. Limit recursion depth.
27307 (record_temporary_equivalences): Corresponding changes.
27309 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27311 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27314 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27317 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27319 * cfgcleanup.c (reg_note_cfa_p): New array.
27320 (insns_have_identical_cfa_notes): New function.
27321 (old_insns_match_p): Don't cross-jump in between /f
27322 and non-/f instructions. If both i1 and i2 are frame related,
27323 verify all CFA notes, their order and content.
27325 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27328 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27329 HImode and SImode with zero extend to DImode to one insn.
27330 (bswap<mode>2_extenddi): Likewise.
27331 (bswapsi2_extenddi): Likewise.
27332 (bswaphi2_extendsi): Likewise.
27333 (bswaphi2): Combine bswap HImode and SImode into one insn.
27334 Separate memory insns from swapping register.
27335 (bswapsi2): Likewise.
27336 (bswap<mode>2): Likewise.
27337 (bswaphi2_internal): Delete, no longer used.
27338 (bswapsi2_internal): Likewise.
27339 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27340 store, and gpr<-gpr swap insns.
27341 (bswap<mode>2_store): Likewise.
27342 (bswaphi2_reg): Register only splitter, combine with the splitter.
27343 (bswaphi2 splitter): Likewise.
27344 (bswapsi2_reg): Likewise.
27345 (bswapsi2 splitter): Likewise.
27346 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27347 the insns into load, store, and register/register insns.
27348 (bswapdi2_ldbrx): Likewise.
27349 (bswapdi2_load): Likewise.
27350 (bswapdi2_store): Likewise.
27351 (bswapdi2_reg): Likewise.
27353 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27355 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27356 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27358 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27361 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27363 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27364 special handling for target option conflicts between dform
27365 options (-mpower9-dform, -mpower9-dform-vector,
27366 -mpower9-dform-scalar) and -mno-direct-move.
27368 2017-03-27 Richard Biener <rguenther@suse.de>
27370 PR tree-optimization/80181
27371 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27373 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27375 * config/arc/predicates.md (move_double_src_operand): Replace the
27376 call to move_double_src_operand with a call to address_operand.
27378 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27380 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27381 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27382 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27384 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27386 * config/arc/predicates.md (long_immediate_loadstore_operand):
27387 Consider scaled addresses cases.
27389 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27391 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27392 restored when in interrupt.
27393 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27394 doesn't have delay slot.
27396 2017-03-27 Richard Biener <rguenther@suse.de>
27399 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27400 inlined thunk clones.
27402 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27405 * asan.c (instrument_derefs): Copy over last operand from
27406 original COMPONENT_REF to the new COMPONENT_REF with
27407 DECL_BIT_FIELD_REPRESENTATIVE.
27408 * ubsan.c (instrument_object_size): Likewise.
27410 2017-03-27 Richard Biener <rguenther@suse.de>
27412 PR tree-optimization/80170
27413 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27414 sure DR/SCEV didnt fold in constants we do not see when looking
27415 at the reference base alignment.
27417 2017-03-27 Richard Biener <rguenther@suse.de>
27419 PR middle-end/80171
27420 * gimple-fold.c (fold_ctor_reference): Properly guard against
27421 NULL return value from canonicalize_constructor_val.
27423 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27426 * config/i386/i386.c (ix86_expand_builtin)
27427 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27428 flags reg setting and flags reg using instructions.
27429 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27430 clobbering instructions to zero extend op2.
27432 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27434 * doc/install.texi (Configuration) <--with-aix-soname>:
27435 Update link to AIX ld.
27437 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27439 PR rtl-optimization/80160
27440 PR rtl-optimization/80159
27441 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27442 reg_alternate_class into account.
27444 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27447 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27448 to consider in curr_insn_transform.
27450 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27452 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27453 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27454 and emit_mode_inner.
27456 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27458 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27459 argument to the overloaded builtin variants. Use the new flag to
27460 deprecate certain builtin variants.
27461 * config/s390/s390-builtin-types.def: Add new builtin types.
27462 * config/s390/s390-builtins.h: Support new flags field for
27463 overloaded builtins.
27464 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27465 (s390_macro_to_expand): Enable vector float data type.
27466 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27467 builtins by incrementing the __VEC__ version number.
27468 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27470 (s390_resolve_overloaded_builtin): Emit error messages depending
27471 on the builtin flags.
27472 * config/s390/s390.c (s390_expand_builtin): Support additional
27473 flags argument. Change error message to match the messages
27474 emitted in s390-c.c.
27475 * config/s390/s390.md: New UNSPEC_* constants.
27476 (op_type): Add new instruction types.
27477 * config/s390/vecintrin.h: Add new builtins and test data class
27479 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27480 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27481 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27482 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27483 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27484 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27486 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27487 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27488 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27489 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27491 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27492 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27493 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27494 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27496 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27497 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27498 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27499 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27500 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27501 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27502 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27504 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27505 ("vec_scatter_element<V_HW_4:mode>_DI")
27506 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27507 ("vec_fpint<mode>", "vflls")
27508 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27509 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27510 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27511 ("*vec_cmphe<mode>_cc"): ... these.
27513 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27514 mode constant instead of magic value.
27516 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27518 * config/s390/s390.c (s390_expand_vec_compare): Support other
27519 vector floating point modes than just V2DF.
27520 (s390_expand_vcond): Likewise.
27521 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27522 (s390_cannot_change_mode_class): Prevent mode changes between TF
27523 and V1TF in vector registers.
27524 * config/s390/s390.md (DF, SF): New mode attributes.
27525 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27526 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27527 SFmode support for VRs.
27528 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27530 (VFT, VF_HW): New mode iterators.
27531 (vw, sdx): New mode attributes.
27532 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27533 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27534 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27535 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27536 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27537 also the new vector floating point modes. Renaming to ...
27539 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27540 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27541 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27542 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27543 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27544 ("vec_unordered<mode>"): ... these.
27546 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27547 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27548 ("*vec_extendv2df"): New insn definitions.
27550 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27552 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27553 ("mulditi3_2", "*muldi3_sign"): New patterns.
27554 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27555 rename the pattern definition.
27557 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27559 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27561 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27563 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27565 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27566 instruction if possible.
27567 * config/s390/vector.md (vec_halfnumelts): New mode
27569 ("*vec_vllezlf<mode>"): New pattern.
27571 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27573 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27574 ("popcountv4si2", "popcountv2di2"): Rename to ...
27575 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27576 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27578 ("popcount<mode>2_vxe"): New pattern.
27580 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27582 * common/config/s390/s390-common.c (processor_flags_table): Add
27584 * config.gcc: Add arch12.
27585 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27586 Default to arch12 for unknown CPU model numbers.
27587 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27588 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27589 PROCESSOR_max sanity check.
27590 * config/s390/s390-opts.h (enum processor_type): Add
27592 * config/s390/s390.c (processor_table): Add arch12.
27593 (s390_expand_builtin): Add check for B_VXE flag.
27594 (s390_issue_rate): Add PROCESSOR_ARCH12.
27595 (s390_get_sched_attrmask): Likewise.
27596 (s390_get_unit_mask): Likewise.
27597 (s390_sched_score): Enable z13 scheduling for arch12.
27598 (s390_sched_reorder): Likewise.
27599 (s390_sched_variable_issue): Likewise.
27600 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27602 (s390_tune_attr): Use z13 scheduling also for arch12.
27603 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27604 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27605 (TARGET_VXE_P): New macros.
27606 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27607 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27608 * config/s390/s390.opt: Add arch12 as processor_type.
27610 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27612 * config/s390/s390.md
27613 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27614 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27615 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27617 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27618 Rename expanders to ...
27620 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27621 ("fixuns_truncdddi2_emu"): ... these.
27623 ("fixuns_trunc<mode>si2_emu"): New expander.
27625 ("*fixuns_truncdfdi2_z13"): Rename to ...
27626 ("*fixuns_truncdfdi2_vx"): ... this.
27628 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27630 * config/s390/2964.md: Remove the single element vector compare
27631 instructions which are no longer used.
27632 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27634 (s390_canonicalize_comparison): Remove handling of DFmode
27636 (s390_expand_vec_compare_scalar): Remove function.
27637 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27638 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27640 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27642 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27644 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27646 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27647 will anyway by matched by mov<mode>_64dfp.
27649 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27651 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27652 vlef/vstef. Add missing operand to vleif.
27654 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27656 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27657 pair for all vector types with 64 bit elements.
27658 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27659 * config/s390/vector.md (V_HW_64): ... here.
27660 (V_128_NOSINGLE): New mode iterator.
27661 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27662 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27663 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27664 ("*vec_load_pairv2di"): Change to ...
27665 ("*vec_load_pair<mode>"): ... this one.
27667 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27669 * config/s390/constraints.md: Add comments.
27670 (jKK): Reject element sizes > 8 bytes.
27671 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27673 * config/s390/s390.md: Add the s_operand checks formerly in
27674 s390_split_ok_p to various splitters where they are still
27676 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27677 for 128 bit vectors. Plus two splitters.
27679 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27681 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27684 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27687 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27688 error if the boundary argument is not constant.
27690 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27692 PR rtl-optimization/80112
27693 * loop-doloop.c (doloop_condition_get): Don't check condition
27694 if cmp isn't SET with IF_THEN_ELSE src.
27696 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27698 PR tree-optimization/80158
27699 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27700 replacing a candidate statement, also replace it for the
27701 candidate's alternate interpretation.
27702 (replace_rhs_if_not_dup): Likewise.
27703 (replace_one_candidate): Likewise.
27705 2017-03-24 Richard Biener <rguenther@suse.de>
27707 PR tree-optimization/80167
27708 * graphite-isl-ast-to-gimple.c
27709 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27711 (translate_isl_ast_to_gimple::get_rename): Likewise.
27713 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27715 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27716 handling of certain combinations of target options, including the
27717 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27718 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27720 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27723 * config/arm/arm.md (*load_multiple): Add reload_completed to
27724 matching condition.
27726 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27727 Richard Biener <rguenth@suse.de>
27729 PR tree-optimization/79908
27730 PR tree-optimization/80136
27731 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27732 been cast away, gimplify_and_add suffices.
27734 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27736 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27738 2017-03-23 Richard Biener <rguenther@suse.de>
27740 PR tree-optimization/80032
27741 * gimplify.c (gimple_push_cleanup): Forced unconditional
27742 cleanups still have to go to the conditional_cleanups
27745 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27747 PR tree-optimization/80072
27748 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27750 (next_operand_entry_id): Change type to unsigned int.
27751 (sort_by_operand_rank): Make sure to return the right return value
27752 even if unsigned fields are bigger than INT_MAX.
27753 (struct oecount): Change cnt and id type to unsigned int.
27754 (oecount_hasher::equal): Formatting fix.
27755 (oecount_cmp): Make sure to return the right return value
27756 even if unsigned fields are bigger than INT_MAX.
27757 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27760 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27761 TREE_READONLY on result if writing it more than once.
27764 * doc/invoke.texi (-fsanitize=thread): Document that with
27765 -fnon-call-exceptions atomics are not able to throw
27769 * tsan.c: Include tree-eh.h.
27770 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27771 maybe_clean_or_replace_eh_stmt where needed.
27772 (instrument_memory_accesses): Add cfg_changed argument.
27773 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27774 if it returned true.
27775 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27777 PR rtl-optimization/63191
27778 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27779 wrapper function, moved the whole old content into ...
27780 (ix86_delegitimize_address_1): ... this. New inline function.
27781 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27782 true as last argument instead of ix86_delegitimize_address.
27784 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27786 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27787 cortexa57_branch_cost.
27789 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27791 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27793 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27796 * doc/md.texi (Constraints): Document wA constraint.
27797 * config/rs6000/constraints.md (wA): New.
27798 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27799 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27800 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27801 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27803 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27806 * gimplify.c (is_oacc_declared): New function.
27807 (oacc_default_clause): Use it to set default flags for acc declared
27808 variables inside parallel regions.
27809 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27810 declared variables.
27811 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27812 declare attribute to any decl as necessary.
27814 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27817 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27818 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27819 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27820 (arm_arch_lpae): This.
27821 * config/arm/arm.c (arm_arch7ve): Rename into ...
27822 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27823 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27826 2017-03-22 Martin Liska <mliska@suse.cz>
27829 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27830 error message instead of an ICE.
27832 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27834 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27836 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27838 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27840 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27843 2017-03-21 Martin Sebor <msebor@redhat.com>
27845 * doc/extend.texi: Use "cannot" instead of "can't."
27846 * doc/hostconfig.texi: Same.
27847 * doc/install.texi: Same.
27848 * doc/invoke.texi: Same.
27849 * doc/loop.texi: Same.
27850 * doc/md.texi: Same.
27851 * doc/objc.texi: Same.
27852 * doc/rtl.texi: Same.
27853 * doc/tm.texi: Same.
27854 * doc/tm.texi.in: Same.
27855 * doc/trouble.texi: Same.
27857 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27860 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27861 (collect_checksum_attributes): Set it.
27862 (die_checksum_ordered): Use it.
27864 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27866 PR tree-optimization/79908
27867 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27868 change: For a VA_ARG whose LHS has been cast away, use
27869 force_gimple_operand to construct the side effects.
27871 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27873 PR translation/80001
27874 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27875 more amenable to translation.
27876 (oacc_loop_auto_partitions): Likewise.
27878 2017-03-21 Marek Polacek <polacek@redhat.com>
27879 Martin Sebor <msebor@redhat.com>
27881 PR tree-optimization/80109
27882 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27883 on INTEGRAL_TYPE_P.
27885 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27886 Segher Boessenkool <segher@kernel.crashing.org>
27889 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27890 check reg_used_between_p between insn and one of succ or succ2
27891 depending on if succ is artificial insn not inserted into insn
27894 2017-03-21 Martin Liska <mliska@suse.cz>
27896 PR gcov-profile/80081
27897 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27898 * doc/gcc.texi: Include gcov-dump stuff.
27899 * doc/gcov-dump.texi: New file.
27901 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27903 PR rtl-optimization/79150
27904 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27905 conditional jump, if the jump is the last insn of the loop.
27907 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27908 Richard Biener <rguenth@suse.de>
27910 PR tree-optimization/79908
27911 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27912 been cast away, use force_gimple_operand to construct the side
27915 2017-03-21 Martin Liska <mliska@suse.cz>
27917 PR libfortran/79956
27918 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27921 2017-03-21 Brad Spengler <spender@grsecurity.net>
27924 * plugin.c (htab_hash_plugin): New function.
27925 (add_new_plugin): Use it and adjust.
27926 (parse_plugin_arg_opt): Adjust.
27927 (init_one_plugin): Likewise.
27929 2017-03-21 Richard Biener <rguenther@suse.de>
27931 PR tree-optimization/80032
27932 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27933 if set force the cleanup to happen unconditionally.
27934 (gimplify_target_expr): Push inserted clobbers with force_uncond
27935 to avoid them being removed by control-dependent DCE.
27937 2017-03-21 Richard Biener <rguenther@suse.de>
27939 PR tree-optimization/80122
27940 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27941 va_arg_pack_len when the inlined call stmt requires pack
27943 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27945 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27948 * tsan.c (instrument_builtin_call): If the memory model argument
27949 is not a constant, assume it is valid.
27952 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27955 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27957 PR rtl-optimization/79910
27958 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27959 if its dest is used by an insn before I2 (other than the combined
27960 insns themselves, which are properly handled already).
27962 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27965 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27967 * combine.c (record_used_regs): New static function.
27968 (try_combine): Handle situations where there is an additional
27969 instruction between I2 and I3 which needs to have a LOG_LINK
27973 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27975 * combine.c (try_combine): Delete redundant i1 test. Call
27976 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27978 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27981 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27982 alternatives 13/14.
27984 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27986 PR tree-optimization/80054
27987 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27988 the optimization if a PHI or any of its arguments is not dominated
27989 by the candidate's basis. Use gphi* rather than gimple* as
27991 (replace_profitable_candidates): Clean up a gimple* variable that
27992 should be a gphi* variable.
27994 2017-03-20 Martin Sebor <msebor@redhat.com>
27997 * doc/extend.texi (attribute constructor): Document present limitation.
27999 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28002 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28003 __POWER9_VECTOR__ #ifdef control, change template definition to
28004 use Power9-specific built-in function.
28005 (vec_any_eq): Likewise.
28006 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28007 to control outcomes from this test.
28008 (vector_ae_<mode>p): For VEC_F modes, likewise.
28010 2017-03-20 Ian Lance Taylor <iant@google.com>
28012 * config/i386/i386.c (ix86_function_regparm): Save an extra
28013 register for -fsplit-stack with DECL_STATIC_CHAIN.
28015 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28018 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28019 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28021 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28023 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28025 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28028 2017-03-20 Marek Polacek <polacek@redhat.com>
28031 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28033 2017-03-20 Richard Biener <rguenther@suse.de>
28035 PR tree-optimization/80113
28036 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28037 allocate extra SSA name for PHI def.
28038 (add_close_phis_to_outer_loops): Likewise.
28039 (add_close_phis_to_merge_points): Likewise.
28040 (copy_loop_close_phi_args): Likewise.
28041 (copy_cond_phi_nodes): Likewise.
28043 2017-03-20 Martin Liska <mliska@suse.cz>
28045 PR middle-end/79753
28046 * tree-chkp.c (chkp_build_returned_bound): Do not build
28047 returned bounds for a LHS that's not a BOUNDED_P type.
28049 2017-03-20 Martin Liska <mliska@suse.cz>
28053 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28054 COMPLEX_CST and VECTOR_CST.
28056 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28059 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28060 target operand. A new splitter adds the clobber statement in case
28061 the target operand is dead anyway.
28063 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28065 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28066 to age-old versions of binutils and glibc.
28068 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28070 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28072 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28074 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28076 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28078 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28079 requirement for binutils 2.13.
28081 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28083 * combine.c (try_combine): Delete redundant i1 test. Call
28084 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28086 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28088 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28089 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28091 <riscv64-*-elf>: Re-arrange section
28092 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28093 <riscv32-*-linux>: Likewise.
28094 <riscv64-*-elf>: Likewise
28095 <riscv64-*-linux>: Likewise.
28097 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28100 * aarch64.opt(verbose-cost-dump): Fix typo.
28102 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28105 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28106 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28108 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28110 * reload.c (find_reloads): When reloading a nonoffsettable address,
28111 use RELOAD_OTHER for it and its address reloads.
28113 PR rtl-optimization/79910
28114 * combine.c (record_used_regs): New static function.
28115 (try_combine): Handle situations where there is an additional
28116 instruction between I2 and I3 which needs to have a LOG_LINK
28119 2017-03-17 Jeff Law <law@redhat.com>
28121 PR tree-optimization/71437
28122 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28123 conditional in the hash table first.
28124 (vrp_dom_walker::before_dom_children): Extract condition from
28125 ASSERT_EXPR. Record condition, its inverion and any implied
28126 conditions as well.
28128 2017-03-17 Marek Polacek <polacek@redhat.com>
28129 Markus Trippelsdorf <markus@trippelsdorf.de>
28131 PR tree-optimization/80079
28132 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28135 2017-03-17 Richard Biener <rguenther@suse.de>
28137 PR middle-end/80075
28138 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28139 Properly verify the LHS before the RHS possibly claims to be
28141 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28144 2017-03-17 Martin Jambor <mjambor@suse.cz>
28146 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28147 (List of -O2 options): Likewise.
28148 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28151 2017-03-17 Tom de Vries <tom@codesourcery.com>
28153 * gcov-dump.c (print_usage): Print bug_report_url.
28155 2017-03-17 Richard Biener <rguenther@suse.de>
28157 PR middle-end/80050
28158 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28159 (parser::peek): Likewise.
28161 2017-03-17 Richard Biener <rguenther@suse.de>
28163 PR tree-optimization/80048
28164 * sese.c (free_sese_info): Properly release rename_map and
28165 copied_bb_map elements.
28167 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28169 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28170 Add linked-list forward and backlinks. Insert on
28171 construction, remove on destruction.
28172 (class pass_store_merging): Add m_stores_head field.
28173 (pass_store_merging::terminate_and_process_all_chains):
28174 Iterate over m_stores_head list.
28175 (pass_store_merging::terminate_all_aliasing_chains):
28177 (pass_store_merging::execute): Check for debug stmts first.
28178 Push new chains onto the m_stores_head stack.
28180 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28183 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28184 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28185 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28187 2017-03-16 Jeff Law <law@redhat.com>
28189 PR tree-optimization/71437
28190 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28191 member function. Implementation moved into after_dom_children
28192 member function and into the threader's thread_outgoing_edges
28194 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28195 some code into new thread_outgoing_edges.
28196 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28197 definition. Simplify marker handling (do it here). Assume we always
28198 have the available expression and the const/copies tables.
28199 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28201 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28202 * tree-vrp.c (equiv_stack): No longer file scoped.
28203 (vrp_dom_walker): New class.
28204 (vrp_dom_walker::before_dom_children): New member function.
28205 (vrp_dom_walker::after_dom_children): Likewise.
28206 (identify_jump_threads): Setup domwalker. Use it rather than
28207 walking edges in a random order by hand. Simplify setup/finalization.
28208 (finalize_jump_threads): Remove.
28209 (vrp_finalize): Do not call identify_jump_threads here.
28210 (execute_vrp): Do it here instead and call thread_through_all_blocks
28213 PR tree-optimization/71437
28214 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28216 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28218 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28219 (dom_opt_dom_walker::thread_across_edge): Remove
28220 handle_dominating_asserts argument. All callers changed.
28221 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28222 changes. Remove calls to lhs_of_dominating_assert. Other
28223 uses of handle_dominating_asserts turn into unconditional code
28224 (simplify_control_stmt_condition_1): Likewise.
28225 (simplify_control_stmt_condition): Likewise.
28226 (thread_through_normal_block, thread_across_edge): Likewise.
28227 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28228 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28229 object if it is not an SSA_NAME.
28230 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28231 before calling into the VRP specific simplifiers.
28232 (identify_jump_threads): Remove handle_dominating_asserts
28235 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28238 * tree-diagnostic.c (default_tree_printer): No longer static.
28239 * tree-diagnostic.h (default_tree_printer): New prototype.
28241 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28243 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28244 Change ins into fmov.
28246 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28248 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28249 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28250 Use h_con constraint for operand 1.
28251 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28252 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28254 2017-03-15 Jeff Law <law@redhat.com>
28256 PR tree-optimization/71437
28257 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28258 (record_temporary_equivalences): Use it.
28260 PR tree-optimization/71437
28261 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28262 tree-ssa-scopedtables.
28263 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28264 (record_conditions, record_cond, vuse_eq): Likewise.
28265 (record_edge_info): Adjust to API tweak of record_conditions.
28266 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28267 (record_temporary_equivalences, optimize_stmt): Likewise.
28268 (eliminate_redundant_computations): Likewise.
28269 (record_equivalences_from_stmt): Likewise.
28270 * tree-ssa-scopedtables.c: Include options.h and params.h.
28271 (vuse_eq): New function, moved from tree-ssa-dom.c
28272 (build_and_record_new_cond): Likewise.
28273 (record_conditions): Likewise. Accept vector of conditions rather
28274 than edge_equivalence structure for first argument.
28275 for the first argument.
28276 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28277 from tree-ssa-dom.c.
28278 (avail_exprs_stack::record_cond): Likewise.
28279 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28280 from tree-ssa-dom.c.
28281 (avail_exprs_stack): Add new member functions lookup_avail_expr
28283 (record_conditions): Declare.
28285 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28288 * lra-constraints.c (process_alt_operands): Increase reject for
28289 reloading an input/output operand.
28291 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28294 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28295 insns to convert from signed/unsigned char/short to IEEE 128-bit
28297 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28299 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28302 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28303 subreg of inner mode for values already in registers.
28305 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28307 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28308 iteration reg is used after the loop.
28310 2017-03-14 Martin Sebor <msebor@redhat.com>
28312 PR tree-optimization/79800
28313 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28314 precision in negative-positive range.
28315 (format_floating): Call non-const overload with adjusted precision.
28317 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28320 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28323 2017-03-14 Martin Sebor <msebor@redhat.com>
28325 PR middle-end/80020
28326 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28327 * builtins.def (aligned_alloc): Use it.
28330 * Makefile.in (GTFILES): Add calls.c.
28331 * calls.c: Include "gt-calls.h".
28333 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28335 PR rtl-optimization/79728
28336 * regs.h (struct target_regs): New field
28337 x_contains_allocatable_regs_of_mode.
28338 (contains_allocatable_regs_of_mode): New macro.
28339 * reginfo.c (init_reg_sets_1): Initialize it, and change
28340 contains_reg_of_mode so it includes global regs as well.
28341 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28342 rather than contains_regs_of_mode.
28344 2017-03-14 Martin Liska <mliska@suse.cz>
28346 * doc/invoke.texi: Document options that can't be combined with
28347 -fcheck-pointer-bounds.
28349 2017-03-14 Martin Liska <mliska@suse.cz>
28351 PR middle-end/79831
28352 * doc/invoke.texi (-Wchkp): Document the option.
28354 2017-03-14 Martin Liska <mliska@suse.cz>
28356 * Makefile.in: Install gcov-dump.
28358 2017-03-14 Martin Liska <mliska@suse.cz>
28360 * multiple_target.c (expand_target_clones): Bail out for
28361 an invalid attribute.
28363 2017-03-14 Richard Biener <rguenther@suse.de>
28365 * alias.c (struct alias_set_entry): Pack properly.
28366 * cfgloop.h (struct loop): Likewise.
28367 * cse.c (struct set): Likewise.
28368 * ipa-utils.c (struct searchc_env): Likewise.
28369 * loop-invariant.c (struct invariant): Likewise.
28370 * lra-remat.c (struct cand): Likewise.
28371 * recog.c (struct change_t): Likewise.
28372 * rtl.h (struct address_info): Likewise.
28373 * symbol-summary.h (function_summary): Likewise.
28374 * tree-loop-distribution.c (struct partition): Likewise.
28375 * tree-object-size.c (struct object_size_info): Likewise.
28376 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28377 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28378 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28379 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28380 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28381 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28382 (struct _stmt_vec_info): Likewise.
28384 2017-03-14 Martin Liska <mliska@suse.cz>
28387 * multiple_target.c (create_dispatcher_calls): Check that
28388 a target can create a function dispatcher.
28390 2017-03-14 Martin Liska <mliska@suse.cz>
28393 * multiple_target.c (expand_target_clones): Drop local.local
28394 flag for default implementation.
28396 2017-03-14 Richard Biener <rguenther@suse.de>
28398 PR tree-optimization/80030
28399 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28401 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28403 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28404 gcc_fallthrough() instead of __attribute__((fallthrough));
28406 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28408 * doc/gcc.texi: Remove "up" link to (DIR).
28409 * doc/gccint.texi: Ditto.
28411 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28413 * doc/install.texi (Specific) <avr>: Remove reference to
28416 2017-03-13 Jeff Law <law@redhat.com>
28418 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28419 attribute rather than comments.
28421 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28422 match_scratch operand is highest.
28424 2017-03-13 Martin Liska <mliska@suse.cz>
28426 PR middle-end/78339
28427 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28428 is a CHKP clone, use original declaration.
28430 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28432 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28433 (arc_conditional_register_usage): Use a different allocation order
28434 when optimizing for size.
28435 * common/config/arc/arc-common.c (arc_option_optimization_table):
28436 Section anchors default on when optimizing for size.
28438 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28440 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28442 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28444 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28445 * config/arc/arc.md (cpu_facility): Add cd variant.
28446 (*movqi_insn): Add code density variant.
28447 (*movhi_insn): Likewise.
28448 (*movqi_insn): Likewise.
28449 (*addsi3_mixed): Likewise.
28450 (subsi3_insn): Likewise.
28452 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28454 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28456 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28458 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28459 expressions with MINUS and UNARY ops.
28461 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28464 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28466 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28467 between vec_select and vector argument.
28468 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28469 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28470 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28471 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28472 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28473 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28475 2017-03-13 Richard Biener <rguenther@suse.de>
28478 * params.def (vect-max-peeling-for-alignment): Fix typo.
28480 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28482 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28483 issue that only occurred with binutils below 2.18.
28485 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28487 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28488 refer to binutils 2.11/2.12 minimum.
28490 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28492 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28493 ftp.kernel.org and simplify binutils requirement.
28495 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28497 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28499 (Optimize Options): Ditto. Also remove redundancy.
28501 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28503 PR translation/79848
28504 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28506 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28507 to G_ to avoid double translation.
28509 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28511 PR translation/79923
28512 * auto-profile.c (get_combined_location): Convert leading
28513 character of diagnostics to lower case and remove trailing period.
28514 (read_profile): Likewise for various diagnostics.
28515 * config/arm/arm.c (arm_option_override): Remove trailing period
28516 from various diagnostics.
28517 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28518 (msp430_expand_delay_cycles): Likewise.
28520 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28523 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28524 full command-line argument, rather than just "str".
28525 (aarch64_validate_march): Likewise.
28526 (aarch64_validate_mtune): Likewise.
28528 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28530 PR rtl-optimization/78911
28531 * lra-assigns.c (must_not_spill_p): New function.
28532 (spill_for): Use it.
28534 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28536 PR tree-optimization/79981
28537 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28538 ATOMIC_COMPARE_EXCHANGE ifn result.
28539 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28540 IFN_ATOMIC_COMPARE_EXCHANGE.
28542 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28545 * opts.c (parse_sanitizer_options): Add missing question mark to
28546 "did you mean" message.
28548 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28550 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28552 (VMULEUH_UNS): Likewise.
28553 (VMULOUB_UNS): Likewise.
28554 (VMULOUH_UNS): Likewise.
28555 * config/rs6000/rs6000.c (builtin_function_type): Remove
28556 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28558 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28561 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28562 x with result of extra_parsing_for_operand_code_0.
28563 (function_reader::extra_parsing_for_operand_code_0): Convert
28564 return type from void to rtx, returning x. When reading
28565 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28566 larger size containing struct block_symbol.
28568 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28570 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28571 -mfloat128-hardware without -m64.
28573 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28576 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28577 entries to the case statement that marks unsigned arguments to
28578 overloaded functions.
28580 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28582 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28583 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28585 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28588 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28589 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28591 2017-03-10 Martin Liska <mliska@suse.cz>
28595 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28596 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28599 2017-03-10 Olivier Hainque <hainque@adacore.com>
28601 * tree-switch-conversion (array_value_type): Start by resetting
28602 candidate type to it's main variant.
28604 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28606 PR rtl-optimization/79909
28607 * combine.c (try_combine): Use simplify_replace_rtx on individual
28608 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28609 of the whole CALL_INSN_FUNCTION_USAGE.
28611 PR tree-optimization/79972
28612 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28613 get_range_info on SSA_NAMEs. Formatting fixes.
28615 2017-03-10 Richard Biener <rguenther@suse.de>
28616 Jakub Jelinek <jakub@redhat.com>
28618 PR tree-optimization/77975
28619 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28620 edge to be constant.
28621 (get_val_for): For constant x return it. Formatting fix.
28622 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28623 would use the same bases as the current one.
28625 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28627 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28628 instead of vec_select for V1TImode.
28629 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28631 (VSX_LE_128): Add V1TI to this mode iterator.
28632 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28633 (*vsx_le_perm_store_<mode>): Likewise.
28634 (pre-reload splitter for VSX stores): Likewise.
28635 (post-reload splitter for VSX stores): Likewise.
28636 (*vsx_xxpermdi2_le_<mode>): Likewise.
28637 (*vsx_lxvd2x2_le_<mode>): Likewise.
28638 (*vsx_stxvd2x2_le_<mode>): Likewise.
28640 2017-03-09 Michael Eager <eager@eagercon.com>
28642 Correct failures with --enable-checking=yes,rtl.
28644 * config/microblaze/microblaze.c (microblaze_expand_shift):
28645 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28646 test for const0_rtx.
28647 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28648 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28650 2017-03-09 Richard Biener <rguenther@suse.de>
28652 PR tree-optimization/79977
28653 * graphite-scop-detection.c (scop_detection::merge_sese):
28654 Handle the case of extra exits to blocks dominating the entry.
28656 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28658 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28661 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28663 PR rtl-optimization/79949
28664 * lra-constraints.c (process_alt_operands): Check memory when
28665 trying to predict a cycle. Print about the overall increase.
28667 2017-03-09 Richard Biener <rguenther@suse.de>
28669 PR middle-end/79971
28670 * gimple-expr.c (useless_type_conversion_p): Preserve
28671 TYPE_SATURATING for fixed-point types.
28673 2017-03-09 Richard Biener <rguenther@suse.de>
28676 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28677 alignment of BLKmode params.
28679 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28682 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28683 (VALL_NO_V2Q): Likewise.
28685 * config/aarch64/aarch64-simd.md
28686 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28688 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28689 VALL_NO_V2Q mode iterator.
28690 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28692 2017-03-09 Martin Liska <mliska@suse.cz>
28694 PR tree-optimization/79631
28695 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28696 tree_int_cst_sign_bit just for INTEGER constants.
28698 2017-03-09 Martin Liska <mliska@suse.cz>
28702 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28705 2017-03-09 Marek Polacek <polacek@redhat.com>
28708 * tree.c (inchash::add_expr): Handle TREE_VEC.
28710 2017-03-09 Martin Liska <mliska@suse.cz>
28713 (chkp_narrow_size_and_offset): New function.
28714 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28715 (void chkp_parse_bit_field_ref): New function.
28716 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28717 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28719 2017-03-09 Martin Liska <mliska@suse.cz>
28722 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28723 (chkp_find_bounds_1): Remove gcc_unreachable.
28725 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28728 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28729 BUILT_IN_SYNC*, determine the access type from the size suffix and
28730 always build a MEM_REF with that type. Handle forgotten
28731 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28734 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28735 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28736 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28737 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28738 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28739 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28740 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28741 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28742 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28743 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28744 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28745 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28746 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28747 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28748 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28749 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28750 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28751 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28752 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28753 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28754 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28755 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28756 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28757 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28758 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28759 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28760 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28761 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28762 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28763 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28764 definitions outside of __OPTIMIZE__ guarded section.
28767 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28768 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28769 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28770 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28773 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28775 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28776 ("vfenez<mode>"): Add missing constraints.
28778 2017-03-08 Martin Sebor <msebor@redhat.com>
28781 * config/nds32/nds32.c (nds32_option_override):
28782 Fix misspelled diagnostic.
28784 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28787 * gimplify.c (gimplify_omp_for): Replace index var in outer
28788 taskloop statement with an artificial variable and add
28789 OMP_CLAUSE_PRIVATE clause for it.
28791 2017-03-08 Richard Biener <rguenther@suse.de>
28793 PR tree-optimization/79955
28794 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28795 for accesses that are completely outside of the variable.
28797 2017-03-08 Andrew Haley <aph@redhat.com>
28799 PR tree-optimization/79943
28800 * tree-ssa-loop-split.c (compute_new_first_bound): When
28801 calculating the new upper bound, (END-BEG) should be added, not
28804 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28806 * config/avr/avr.md (setmemhi): Make sure match_dup
28807 operand number comes before match_scratch.
28809 2017-03-08 Richard Biener <rguenther@suse.de>
28811 PR tree-optimization/79920
28812 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28813 with ncopies == 1 to ...
28814 (vect_transform_slp_perm_load): ... here. Properly compute
28815 all element loads by iterating VF times over the group. Do
28816 not handle ncopies (computed in a broken way) in
28817 vect_create_mask_and_perm.
28819 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28822 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28823 is a uniform vector, use uniform_vector_p return value instead of
28824 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28826 2017-03-07 Marek Polacek <polacek@redhat.com>
28828 PR middle-end/79809
28829 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28830 (alloca_call_type): Likewise.
28832 2017-03-07 Martin Liska <mliska@suse.cz>
28834 * gcov.c (process_args): Put comment to correct location.
28836 2017-03-07 Martin Liska <mliska@suse.cz>
28838 PR middle-end/68270
28839 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28840 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28841 (chkp_narrow_bounds_for_field): Likewise.
28842 (chkp_parse_array_and_component_ref): Pass one more argument to
28845 2017-03-07 Richard Biener <rguenther@suse.de>
28847 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28850 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28852 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28853 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28855 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28858 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28859 to end of description.
28860 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28862 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28864 PR rtl-optimization/79901
28865 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28867 (*avx512f_<code><mode>3<mask_name>): ... this.
28868 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28869 iterator instead of VI8_AVX2_AVX512BW.
28871 PR rtl-optimization/79901
28872 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28873 min/max expander, expand it using expand_vec_cond_expr.
28876 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28879 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28882 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28883 to void * for PCH reasons.
28884 * dwarf2out.c (output_loc_operands, output_die): Cast
28885 v.val_vec.array to unsigned char *.
28887 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28890 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28893 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28895 PR rtl-optimization/79571
28896 * lra-constraints.c (process_alt_operands): Calculate static
28897 reject and subtract it from overall when only addresses will be
28900 2017-03-06 Julia Koval <julia.koval@intel.com>
28903 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28904 incoming stack boundary to 128 for 64-bit targets.
28906 2017-03-06 Richard Biener <rguenther@suse.de>
28908 PR tree-optimization/79894
28909 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28910 to NULL after folding it.
28912 2017-03-06 Richard Biener <rguenther@suse.de>
28914 PR tree-optimization/79824
28915 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28916 check disabling peeling for gaps.
28918 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28920 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28921 attributes): Document gettimeofday.
28923 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28925 * config/s390/s390.c (s390_option_override_internal): Set
28926 PARAM_MIN_VECT_LOOP_BOUND
28928 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28930 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28931 * config/s390/s390.md: Likewise.
28933 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28936 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28937 (<avx2_avx512>_perm<mode>): Rename to ...
28938 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28940 (<avx512>_perm<mode>_mask): Rename to ...
28941 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28943 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28944 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28945 instead of VI8F_256_512.
28946 (avx512f_perm<mode>): New define_expand.
28947 (avx512f_perm<mode>_mask): Likewise.
28948 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28949 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28951 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28953 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28954 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28956 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28958 2017-03-06 Martin Liska <mliska@suse.cz>
28961 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28962 when having a SSA NAME w/o VAR_DECL assigned to it.
28964 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28966 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28967 msa_dpsub_<su>_d): Fix MODE for vec_select.
28969 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28971 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28973 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28975 2017-03-06 Richard Biener <rguenther@suse.de>
28977 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28978 * plugin.c (register_plugin_info): Likewise.
28979 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28981 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28983 * config/i386/sse.md (sse_storehps, sse_storelps,
28984 avx_<castmode><avxsizesuffix>_<castmode>,
28985 avx512f_<castmode><avxsizesuffix>_<castmode>,
28986 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28987 in condition that at least one operand is not a MEM.
28989 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28991 PR middle-end/79805
28992 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28993 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28995 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28996 gimple_call_nothrow_p flag based on whether original builtin can throw.
28997 If it can, emit following stmts on the fallthrough edge.
28998 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28999 don't create new bb if inserting just debug stmts on the edge, try to
29000 insert them on the fallthru bb or just reset debug stmts.
29002 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29005 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29006 restore recog_data (including the operand rtxes inside it) around
29007 the call to get_insn_template.
29009 2017-03-03 Martin Sebor <msebor@redhat.com>
29011 PR tree-optimization/79699
29012 * context.c (context::~context): Free MPFR caches to avoid
29013 a memory leak on program exit.
29015 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29017 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29018 Use wide_int::ulow () instead of .elt (0).
29020 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29022 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29023 (*pushxf): Limit oF constraint to 32bit targets and add oC
29024 constraint for 64bit targets.
29025 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29026 (*pushdf): Change rmF constraint to rmC.
29028 2017-03-03 Martin Liska <mliska@suse.cz>
29030 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29031 Remove unused variable.
29033 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29036 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29037 is a memory operand, increase num_memory.
29038 (ix86_expand_args_builtin): Likewise.
29040 2017-03-03 Jan Hubicka <jh@suse.cz>
29043 * ipa-devirt.c (maybe_record_node): Properly handle
29044 __cxa_pure_virtual visibility.
29046 2017-03-03 Martin Liska <mliska@suse.cz>
29048 PR tree-optimization/79803
29049 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29051 (pass_loop_prefetch::execute): Disabled optimization if an
29052 assumption about L1 cache size is not met.
29054 2017-03-03 Martin Liska <mliska@suse.cz>
29056 PR rtl-optimization/79574
29057 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29058 (hash_scan_set): Likewise.
29059 (dump_hash_table): Likewise.
29060 (hoist_code): Likewise.
29062 2017-03-03 Richard Biener <rguenther@suse.de>
29064 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29065 in place of uhigh ().
29066 (fixed_convert_from_real): Likewise.
29068 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29071 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29073 2017-03-03 Richard Biener <rguenther@suse.de>
29075 PR middle-end/79818
29076 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29077 TYPE_OVERFLOW_UNDEFINED check.
29079 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29081 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29083 (vector_ae_<mode>_p): Likewise.
29084 (vector_nez_<mode>_p): Likewise.
29085 (vector_ne_v2di_p): Likewise.
29086 (vector_ae_v2di_p): Likewise.
29087 (vector_ne_<mode>_p): Likewise.
29088 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29090 (vsx_tsqrt<mode>2_fe): Likewise.
29092 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29095 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29097 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29099 PR rtl-optimization/79780
29100 * cprop.c (one_cprop_pass): When second and further conditional trap
29101 in a single basic block is turned into an unconditional trap, turn it
29102 into a deleted note to avoid RTL verification failures.
29104 2017-03-02 Richard Biener <rguenther@suse.de>
29106 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29108 2017-03-02 Richard Biener <rguenther@suse.de>
29110 PR tree-optimization/79345
29112 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29113 param and abort the walk, returning -1 if it is hit.
29114 (walk_aliased_vdefs): Take a limit param and pass it on.
29115 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29116 defaulting to 0 and return a signed int.
29117 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29118 (check_defs): New helper.
29119 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29120 about uninitialized memory.
29121 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29122 bogus uninitialized warning.
29123 (fixed_convert_from_real): Likewise.
29125 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29127 PR tree-optimization/66768
29128 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29129 iv_use if base object can't be determined.
29131 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29133 PR tree-optimization/79345
29134 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29135 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29136 (get_pattern_stats): Initialize it.
29137 * genemit.c (gen_expand): Verify match_scratch numbers come after
29138 match_operand/match_dup numbers.
29139 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29140 match_scratch numbers.
29141 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29143 * config/s390/s390.md (trunctdsd2): Likewise.
29145 2017-03-02 Richard Biener <rguenther@suse.de>
29147 * wide-int.h (wide_int_storage::operator=): Implement in terms
29150 2017-03-02 Richard Biener <rguenther@suse.de>
29152 PR tree-optimization/79777
29153 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29154 the to insert expression to sth existing.
29156 2017-03-01 Martin Sebor <msebor@redhat.com>
29158 PR middle-end/79692
29159 * gimple-ssa-sprintf.c
29160 (directive::known_width_and_precision): New function.
29161 (format_integer): Use it.
29162 (get_mpfr_format_length): Consider the full range of precision
29163 when computing %g output with the # flag. Set the likely byte
29164 count to 3 rather than 1 when precision is indeterminate.
29165 (format_floating): Correct the lower bound of precision.
29167 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29169 * doc/invoke.texi: Document default code model for 64-bit Linux.
29171 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29174 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29175 udiv rather than div since input pattern is unsigned.
29177 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29179 * config/i386/i386.c (print_reg): Warn for values of
29180 unsupported size in integer register.
29182 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29185 * config/rs6000/predicates.md (current_file_function_operand): Do
29186 not allow self calls to be local if the function is replaceable.
29188 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29191 * config/rs6000/altivec.h (vec_ctz and others): Change the
29192 preprocessor macro that controls conditional compilation from
29193 _ARCH_PWR9 to __POWER9_VECTOR__.
29194 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29195 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29196 control (instead of _ARCH_PWR9 control) so that template
29197 definition uses power9-specific function.
29198 (vec_any_eq): Likewise.
29199 (vec_all_ne): Change macro definition to use a power9-specific
29200 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29201 _ARCH_PWR9 control).
29202 (vec_any_eq) Likewise.
29203 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29204 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29205 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29206 support for xvcmpnedp instruction.
29207 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29208 macro expansion so that Power9 implementation of vec_all_ne does
29209 not use the AltiVec predicate framework.
29210 (VCMPNEH_P): Likewise.
29211 (VCMPNEW_P): Likewise.
29212 (VCMPNED_P): Likewise.
29213 (VCMPNEFP_P): Likewise.
29214 (VCMPNEDP_P): Likewise.
29215 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29216 implementation of vec_any_eq to not use AltiVec predicate
29218 (VCMPAEH_P): Likewise.
29219 (VCMPAEW_P): Likewise.
29220 (VCMPAED_P): Likewise.
29221 (VCMPAEFP_P): Likewise.
29222 (VCMPAEDP_P): Likewise.
29223 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29224 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29225 not use the AltiVec predicate framework.
29226 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29227 of vec_any_eq to not use AltiVec predicate framework.
29228 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29229 support for predefined __POWER9_VECTOR__ macro to indicate that
29230 Power9 instruction selection is enabled.
29231 (altivec_overloaded_builtins): Remove extraneous
29232 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29233 function argument types RS6000_BTI_bool_V16QI and
29234 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29235 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29236 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29237 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29238 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29239 Power9 for implementations of vec_cmpne. Change the signature for
29240 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29241 (representing vec_all_ne) to remove the previously described first
29242 argument of type RS6000_BTI_INTSI, as this was an artifact of
29243 reliance on the AltiVec predicate framework, which is no longer
29244 used in the implementation of these functions. Add
29245 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29246 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29247 since, unlike the AltiVec predicate framework implementation, we
29248 do not share function descriptors between vec_alle and vec_anyeq.
29249 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29250 set of modes that receive special treatment even when
29251 TARGET_P9_VECTOR is true. The special treatment emits code that
29252 does not depend on Power9 instructions.
29253 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29254 define_expand to not rely on AltiVec predicate framework.
29255 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29257 (vector_ne_v2di_p): Change this define_expand to not rely on
29258 AltiVec predicate framework.
29259 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29261 (vector_ne_<mode>_p): Change this define_expand to not rely on
29262 AltiVec predicate framework.
29263 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29265 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29266 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29267 define_insn pattern.
29268 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29269 define_insn pattern because the xvcmpne<VSs>. instruction is not
29271 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29272 instruction is not supported.
29274 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29276 * config/nvptx/nvptx.c: Include intl.h.
29278 2017-03-01 Martin Jambor <mjambor@suse.cz>
29281 * ipa-prop.h (ipa_bits): Removed field known.
29282 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29283 to pointers. Adjusted their comments to warn about their sharing.
29284 (ipcp_transformation_summary): Change bits to a vector of pointers.
29285 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29286 (ipa_get_ipa_bits_for_value): Declare.
29287 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29288 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29289 (ipa_bits_hash_table): Likewise.
29290 (ipa_vr_ggc_hash_traits): Likewise.
29291 (ipa_vr_hash_table): Likewise.
29292 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29293 being pointers and vr_known being removed.
29294 (ipa_set_jf_unknown): Likewise.
29295 (ipa_get_ipa_bits_for_value): New function.
29296 (ipa_set_jfunc_bits): Likewise.
29297 (ipa_get_value_range): New overloaded functions.
29298 (ipa_set_jfunc_vr): Likewise.
29299 (ipa_compute_jump_functions_for_edge): Use the above functions to
29300 construct bits and vr parts of jump functions.
29301 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29302 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29304 (ipcp_grow_transformations_if_necessary): Also allocate
29305 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29307 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29308 them. Fix too long lines.
29309 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29310 vr_known being removed.
29311 (ipa_read_jump_function): Use new setter functions to construct bits
29312 and vr parts of jump functions or set them to NULL.
29313 (write_ipcp_transformation_info): Adjust for bits being pointers.
29314 (read_ipcp_transformation_info): Likewise.
29315 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29317 Include gt-ipa-prop.h.
29318 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29320 (ipcp_store_bits_results): Likewise.
29321 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29322 Do not write to existing jump functions but use a temporary instead.
29324 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29327 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29328 attempt to use its first operand as BIT_FIELD_REF base.
29330 2017-03-01 Richard Biener <rguenther@suse.de>
29332 PR middle-end/79721
29333 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29334 interpolating formula in wrapping arithmetic.
29335 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29337 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29339 PR tree-optimization/79734
29340 * tree-vect-generic.c (expand_vector_condition): Optimize
29341 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29342 Handle VEC_COND_EXPR where comparison has different inner width from
29343 type's inner width.
29345 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29347 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29348 markup, and similar issues. Remove @opindex entries for things
29349 that aren't options. Add missing -mmpy-option entries.
29351 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29353 PR tree-optimization/79737
29354 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29355 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29356 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29357 instead of byte_size. Formatting fix.
29358 (shift_bytes_in_array_right): Formatting fix.
29360 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29363 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29364 condition on optimize for the leaf function test.
29366 2017-02-28 Martin Liska <mliska@suse.cz>
29369 * read-rtl-function.c (function_reader::handle_unknown_directive):
29370 Bail out when one uses -flto.
29372 2017-02-28 Martin Liska <mliska@suse.cz>
29374 * common.opt: Replace space with tabular for options of <number>
29376 * config/i386/i386.opt: Show <number> value for
29377 -mlarge-data-threshold.
29378 * opts.c (print_filtered_help): Do not display number in hexadecimal
29381 2017-02-28 Martin Liska <mliska@suse.cz>
29383 * common.opt: Fix --help=option -Q for options which are of
29386 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29388 * config/i386/i386.c (print_reg): Error out for values
29389 of 8-bit size in invalid integer register.
29391 2017-02-28 Martin Sebor <msebor@redhat.com>
29393 PR tree-optimization/79691
29394 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29396 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29399 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29400 gcc_unreachable with output_operand_lossage.
29402 2017-02-28 Richard Biener <rguenther@suse.de>
29404 PR tree-optimization/79740
29405 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29407 (visit_nary_op): Insert the nary into the hashtable if we
29408 pattern-matched sth.
29409 * tree-ssa-pre.c (eliminate_insert): Robustify.
29411 2017-02-28 Richard Biener <rguenther@suse.de>
29413 PR middle-end/79731
29414 * fold-const.c (decode_field_reference): Reject out-of-bound
29417 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29419 * config/i386/i386.c: Include intl.h.
29420 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29421 instead of just cond ? "..." : "...".
29422 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29423 * coverage.c (read_counts_file): Likewise.
29424 * omp-offload.c: Include intl.h.
29425 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29426 of just cond ? "..." : "...".
29427 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29428 of just cond ? "..." : "...".
29430 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29433 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29435 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29436 'tune for' CPU name.
29437 * config/arm/arm-cpu-data.h: Regenerated.
29439 2017-02-28 Richard Biener <rguenther@suse.de>
29441 PR tree-optimization/79732
29442 * tree-inline.c (expand_call_inline): Do not shadow var.
29444 2017-02-28 Richard Biener <rguenther@suse.de>
29446 PR tree-optimization/79723
29447 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29448 address-space properly.
29450 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29452 * doc/optinfo.texi (Optimization groups): Fix option used for
29454 * doc/invoke.texi (-fopt-info): Document "omp".
29455 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29456 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29457 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29458 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29459 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29461 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29463 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29464 OPTGROUP_OMP with "omp".
29466 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29469 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29470 for arithmetic shift of unsigned V2DI.
29472 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29474 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29475 arc/linux.h headers.
29476 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29477 (LINK_SPEC): Likewise.
29478 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29479 (EXTRA_SPECS): Likewise.
29480 (STARTFILE_SPEC): Likewise.
29481 (ENDFILE_SPEC): Likewise.
29482 (LIB_SPEC): Likewise.
29483 (TARGET_SDATA_DEFAULT): Likewise.
29484 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29485 (MULTILIB_DEFAULTS): Likewise.
29486 (DWARF2_UNWIND_INFO): Likewise.
29487 * config/arc/big.h: New file.
29488 * config/arc/elf.h: Likewise.
29489 * config/arc/linux.h: Likewise.
29490 * config/arc/t-uClibc: Remove.
29492 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29494 PR tree-optimization/77536
29495 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29496 (tree_transform_and_unroll_loop): Use above function to compute the
29497 estimated niter of unrolled loop and use it when scaling profile.
29498 Also use count info rather than frequency if it's non-zero.
29499 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29500 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29501 (vect_transform_loop): Call above function.
29503 2017-02-27 Richard Biener <rguenther@suse.de>
29505 PR tree-optimization/45397
29506 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29507 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29508 (visit_nary_op): Add pattern matching for CSEing sign-changed
29509 or truncated operations with wider ones.
29511 2017-02-27 Richard Biener <rguenther@suse.de>
29513 PR tree-optimization/79690
29514 * tree-vect-stmts.c (vectorizable_store): Use vector type
29515 built from the DR with address-space.
29517 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29519 * doc/invoke.texi (Optimize Options): Refine the description
29520 of asan-use-after-return.
29522 2017-02-25 Alan Modra <amodra@gmail.com>
29524 PR rtl-optimization/79584
29525 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29526 base, not ad->base_term, the reg within base. Remove assertion
29527 that ad->base == ad->base_term. Replace gen_int_mode using
29528 bogus mode with const0_rtx.
29530 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29532 PR middle-end/79396
29533 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29534 FMA_EXPR like tcc_binary or tcc_unary.
29536 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29539 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29541 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29542 (output_loc_operands): Handle DW_OP_call_ref and
29543 DW_OP_GNU_variable_value.
29544 (struct variable_value_struct): New type.
29545 (struct variable_value_hasher): Likewise.
29546 (variable_value_hash): New variable.
29547 (string_types): Remove.
29548 (copy_loc_descr): New function.
29549 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29550 (prepend_loc_descr_to_each): New function.
29551 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29552 instead of add_loc_descr_to_each if the first argument is single
29553 location list and the second has multiple.
29554 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29555 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29556 when looking for variable value which doesn't have other location info.
29557 (loc_list_from_tree): Formatting fix.
29558 (gen_array_type_die): Simplify DW_AT_string_length handling.
29559 (adjust_string_types): Remove.
29560 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29561 string_types. Call resolve_variable_values.
29562 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29563 (resolve_addr_in_expr): Likewise. Add A argument.
29564 (copy_deref_exprloc): Remove deref argument. Adjust for the
29565 original expression being DW_OP_GNU_variable_value with optionally
29566 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29567 optionally after it.
29568 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29569 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29570 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29571 (variable_value_hasher::hash, variable_value_hasher::equal): New
29573 (resolve_variable_value_in_expr, resolve_variable_value,
29574 resolve_variable_values, note_variable_value_in_expr,
29575 note_variable_value): New functions.
29576 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29579 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29582 * opts.h (handle_generated_option): Add GENERATED_P argument.
29583 * opts-common.c (handle_option): Adjust function comment.
29584 (handle_generated_option): Add GENERATED_P argument, pass it to
29586 (control_warning_option): Pass false to handle_generated_option
29588 * opts.c (maybe_default_option): Pass true to handle_generated_option
29590 * optc-gen.awk: Likewise.
29592 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29594 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29595 a REG, look at the REG it is a SUBREG of.
29596 (splitter for cmpeqsi_t): Ditto.
29598 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29600 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29601 the special USEs with the pattern of the insn, not the insn itself.
29603 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29606 * doc/invoke.texi: Document -mload-store-pairs.
29608 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29609 Sandra Loosemore <sandra@codesourcery.com>
29611 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29612 argument isn't a CONST_INT.
29613 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29614 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29615 (nios2_validate_compare): Bypass alternate compare logic if *op2
29616 is not a CONST_INT.
29617 (ldstwm_operation_p): Return false if first_base is not a REG or
29618 if first_offset is not a CONST_INT.
29620 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29622 * config/cris/cris.md: Use correct operand in a define_peephole2.
29624 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29626 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29628 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29630 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29631 this_insn if it is an INSN or JUMP_INSN.
29632 (force_offsettable): Look at base, not at addr.
29633 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29634 on things that aren't necessarily CONST_INTs.
29636 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29638 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29639 -mfpmath=sse is the default also for x86-32 targets with SSE2
29640 instruction set when @option{-ffast-math} is enabled
29642 2017-02-24 Jeff Law <law@redhat.com>
29644 PR rtl-optimizatoin/79286
29645 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29648 2017-02-24 Richard Biener <rguenther@suse.de>
29650 PR tree-optimization/79389
29651 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29654 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29656 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29657 function comment to reflect reality.
29658 (loop_exits_before_overflow): Fix typo in function description.
29660 2017-02-24 Richard Biener <rguenther@suse.de>
29662 PR tree-optimization/79389
29663 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29664 properly that a threading opportunity exists. Detect conditional
29665 copy/constant propagation opportunities.
29667 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29669 * config/visium/visium.md (type): Add trap.
29670 (b): New mode attribute.
29671 (*btst): Rename into...
29672 (*btst<mode>): ...this and adjust.
29673 (*cbranchsi4_btst_insn): Rename into...
29674 (*cbranch<mode>4_btst_insn): ...this and adjust.
29675 (trap): New define_insn.
29677 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29679 PR tree-optimization/79389
29680 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29681 (noce_reversed_cond_code): New function.
29682 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29683 reversed_comparison_code. Formatting fix.
29684 (noce_try_store_flag): Test rev_cond != NULL in addition to
29685 reversed_comparison_code.
29686 (noce_try_store_flag_constants): Likewise.
29687 (noce_try_store_flag_mask): Likewise.
29688 (noce_try_addcc): Use rev_cond if non-NULL instead of
29689 reversed_comparison_code.
29690 (noce_try_cmove_arith): Likewise. Formatting fixes.
29691 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29692 (noce_find_if_block): Initialize rev_cond.
29693 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29694 instead of false as last argument never attempt to reverse it
29697 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29699 PR tree-optimization/79663
29700 * tree-predcom.c (combine_chains): Process refs in reverse order
29701 only for ZERO length chains, and add explaining comment.
29703 2017-02-23 Jeff Law <law@redhat.com>
29705 PR tree-optimization/79578
29706 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29707 in call to operand_equal_p.
29709 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29712 * config/i386/cpuid.h: Fix another undefined behavior.
29714 2017-02-23 Richard Biener <rguenther@suse.de>
29716 PR tree-optimization/79683
29717 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29718 vector types for data-refs.
29720 2017-02-23 Martin Liska <mliska@suse.cz>
29722 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29724 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29726 PR middle-end/79665
29727 * internal-fn.c (get_range_pos_neg): Moved to ...
29728 * tree.c (get_range_pos_neg): ... here. No longer static.
29729 * tree.h (get_range_pos_neg): New prototype.
29730 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29731 are known to be in between 0 and signed maximum inclusive, try to
29732 expand both unsigned and signed divmod and use the cheaper one from
29735 2017-02-22 Jeff Law <law@redhat.com>
29737 PR tree-optimization/79578
29738 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29739 to compare base operands.
29741 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29744 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29745 gpc_reg_operand instead of fpr_reg_operand.
29747 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29749 * config/mips/mips.c (mips_return_in_memory): Force FP
29750 vector types to be returned in memory for o32 ABI.
29752 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29754 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29755 instead of DW_TAG_member for static data member declarations and don't
29756 set no_linkage_name for static inline data members.
29757 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29760 2017-02-22 Martin Liska <mliska@suse.cz>
29762 * doc/invoke.texi: Replace inequality signs with square brackets
29765 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29768 * lra-constraints.c (simplify_operand_subreg): Handle
29769 WORD_REGISTER_OPERATIONS targets.
29771 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29774 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29775 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29776 elimination by swapping fld*.
29778 2017-02-22 Richard Biener <rguenther@suse.de>
29780 PR tree-optimization/79673
29781 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29782 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29783 irrelevant address-space qualifiers and avoiding a
29784 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29786 2017-02-22 Richard Biener <rguenther@suse.de>
29788 PR tree-optimization/79666
29789 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29790 to not symbolically negate if that may introduce undefined
29793 2017-02-22 Martin Liska <mliska@suse.cz>
29796 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29797 * data-streamer-out.c (streamer_write_gcov_count_stream):
29799 * value-prof.c (stream_out_histogram_value): Make assert more
29800 precise based on type of counter.
29802 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29805 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29806 Use nonimmediate_operand instead of memory_operand for operand 1.
29807 (float-extend standard_x87sse_constant_load splitter): Ditto.
29809 2017-02-21 Jeff Law <law@redhat.com>
29811 PR tree-optimization/79621
29812 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29813 blocks with edges to themselves.
29815 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29818 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29819 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29820 Use gimple_call_builtin_p.
29823 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29824 on temporarily removed DEBUG_INSNs.
29826 PR tree-optimization/79649
29827 * tree-loop-distribution.c (classify_partition): Give up on
29828 non-generic address space loads/stores.
29830 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29832 * doc/loop.texi (Loop manipulation): Remove nonexistent
29833 tree_ssa_loop_version from the documentation.
29834 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29836 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29839 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29840 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29841 * config/rs6000/rs6000.c: Include except.h.
29842 (rs6000_expand_split_stack_prologue): Call
29843 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29845 2017-02-21 Martin Jambor <mjambor@suse.cz>
29848 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29849 have been analyzed.
29851 2017-02-21 Martin Jambor <mjambor@suse.cz>
29853 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29854 for backward compatibility only.
29855 * doc/invoke.texi (Option Summary): Remove all references to
29856 -fipa-cp-alignment.
29858 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29862 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29864 * lra-constraints.c (curr_insn_transform): Handle
29865 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29867 2017-02-21 Martin Liska <mliska@suse.cz>
29869 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29871 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29873 PR translation/79638
29874 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29876 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29879 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29880 (arm_function_ok_for_sibcall): Return false for an indirect call by
29881 descriptor if all the argument registers are used.
29882 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29883 alignment of the function.
29885 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29887 PR tree-optimization/61441
29888 * simplify-rtx.c (simplify_const_unary_operation): For
29889 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29890 the sNaN unmodified.
29892 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29894 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29895 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29896 instead of SYSTEM_HEADER_DIR.
29898 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29899 Martin Liška <mliska@suse.cz>
29901 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29902 Fix typos and grammar, use active voice, and clarify.
29904 2017-02-20 Marek Polacek <polacek@redhat.com>
29906 PR middle-end/79537
29907 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29910 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29912 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29915 * config/i386/i386.c (ix86_expand_builtin): Handle
29916 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29917 ix86_builtins_isa[fcode].isa as a requirement of those
29918 flags and any other flag in the bitmask.
29919 (ix86_init_mmx_sse_builtins): Use 0 instead of
29920 ~OPTION_MASK_ISA_64BIT as mask.
29921 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29922 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29923 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29924 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29926 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29929 * lra-constraints.c (split_reg): Check requested split mode
29930 is supported by the register.
29932 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29934 * lra-constraints.c (simplify_operand_subreg): Remove early
29937 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29940 * lra-constraints.c (curr_insn_transform): Tighten condition
29941 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29943 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29946 * lra-constraints.c (curr_insn_transform): Handle
29947 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29949 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29952 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29954 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29956 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29959 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29962 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29964 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29967 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29969 * final.c (last_columnnum, override_columnnum): New variables.
29970 (final_start_function): Set last_columnnum, pass it to begin_prologue
29971 hook and pass 0 to dwarf2out_begin_prologue.
29972 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29973 to source_line debug hook.
29974 (notice_source_line): Compute last_columnnum and for debug_column_info
29975 return true on column changes.
29976 * debug.h (struct gcc_debug_hooks): Add column argument to
29977 source_line and begin_prologue hooks.
29978 (debug_nothing_int_charstar_int_bool): Remove prototype.
29979 (debug_nothing_int_int_charstar,
29980 debug_nothing_int_int_charstar_int_bool): New prototypes.
29981 (dwarf2out_begin_prologue): Add column argument.
29982 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29983 begin_prologue hooks.
29984 (debug_nothing_int_charstar_int_bool): Remove.
29985 (debug_nothing_int_int_charstar,
29986 debug_nothing_int_int_charstar_int_bool): New functions.
29987 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29988 through to dwarf2out_source_line.
29989 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29990 (dwarf2out_source_line): Add column argument, emit it if requested.
29991 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29993 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29994 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29995 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29996 through to dwarf2out_begin_prologue.
29997 (vmsdbgout_source_line): Add column argument, pass it through to
29998 dwarf2out_source_line.
29999 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30000 dbxout_source_line caller.
30001 (dbxout_source_line): Add column argument.
30003 * common.opt (gno-column-info, gcolumn-info): New options.
30004 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30005 (check_die): Also test for multiple DW_AT_decl_column attributes.
30006 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30007 DW_AT_decl_column if requested.
30008 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30010 (gen_variable_die): Likewise.
30011 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30012 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30015 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30016 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30017 (ix86_handle_option): Handle OPT_m3dnowa.
30018 * doc/invoke.texi (-m3dnowa): Document.
30019 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30020 -m3dnowa instead of -m3dnow -march=athlon.
30023 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30024 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30026 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30029 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30030 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30031 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30032 generator for vsx_xxpermdi_<mode>_be.
30033 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30034 force big-endian semantics.
30035 (vsx_xxpermdi_<mode>_be): New define_expand with same
30036 implementation as previous version of vsx_xxpermdi_<mode>.
30038 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30040 PR tree-optimization/79327
30041 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30042 variable, its initialization and use.
30044 2017-02-17 Julia Koval <julia.koval@intel.com>
30046 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30047 (OPTION_MASK_ISA_PKU_UNSET): New.
30048 (ix86_handle_option): Handle -mrdpid.
30049 * config/i386/cpuid.h (bit_RDPID): New.
30050 * config/i386/driver-i386.c (host_detect_local_cpu):
30051 Detect RDPID feature.
30052 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30053 * config/i386/i386-c.c (ix86_target_macros_internal):
30055 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30056 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30057 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30058 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30059 * config/i386/i386.md (define_insn "rdpid"): New.
30060 * config/i386/i386.opt Add -mrdpid.
30061 * config/i386/immintrin.h (_rdpid_u32): New.
30063 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30065 PR rtl-optimization/79541
30066 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30067 instead of transforming it into USE.
30069 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30071 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30072 If HONOR_SNANS (SFmode) force the input to a register.
30073 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30074 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30075 an frsp or similar insn.
30077 2017-02-17 Martin Liska <mliska@suse.cz>
30079 PR rtl-optimization/79577
30080 * params.def (selsched-max-sched-times): Increase minimum to 1.
30082 2017-02-17 Martin Liska <mliska@suse.cz>
30084 PR rtl-optimization/79574
30085 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30087 2017-02-17 Martin Liska <mliska@suse.cz>
30089 PR tree-optimization/79529
30090 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30091 ssa_defined_default_def_p to handle cases which are implicitly
30093 * tree-ssa.c (ssa_defined_default_def_p): New function.
30094 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30095 which are implicitly defined.
30096 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30098 2017-02-17 Richard Biener <rguenther@suse.de>
30100 PR middle-end/79576
30101 * params.def (max-ssa-name-query-depth): Limit to 10.
30103 2017-02-17 Richard Biener <rguenther@suse.de>
30105 PR tree-optimization/79552
30106 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30109 2017-02-17 Richard Biener <rguenther@suse.de>
30112 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30114 2017-02-17 Marek Polacek <polacek@redhat.com>
30116 PR middle-end/79536
30117 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30118 (fold_negate_expr): New wrapper.
30120 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30122 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30123 Correct terminology and de-emphasize pre-standard behavior.
30125 2017-02-16 Alan Modra <amodra@gmail.com>
30127 PR rtl-optimization/79286
30128 * ira.c (def_dominates_uses): New function.
30129 (update_equiv_regs): Don't create an equivalence for insns that
30130 may trap where the register def does not dominate the use.
30132 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30134 PR rtl-optimization/78127
30135 * lra.c (lra): Call lra_eliminate before finish the loop after
30138 2017-02-16 Richard Biener <rguenther@suse.de>
30140 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30142 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30143 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30144 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30145 (isl_val_int_from_wi): New function.
30146 (extract_affine_gmp): Rename to ...
30147 (extract_affine_wi): ... this, take a widest_int.
30148 (extract_affine_int): Just wrap extract_affine_wi.
30149 (add_param_constraints): Use isl_val_int_from_wi.
30150 (add_loop_constraints): Likewise, and extract_affine_wi.
30152 2017-02-15 Jeff Law <law@redhat.com>
30154 PR middle-end/79521
30155 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30156 ira_init_register_move_cost_if_necessary.
30158 2017-02-15 Martin Sebor <msebor@redhat.com>
30160 PR middle-end/32003
30161 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30162 removed in a prior commit.
30164 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30166 PR tree-optimization/79347
30167 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30168 counters during peeling.
30170 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30172 * Makefile.in (site.exp): Remove "set ISLVER".
30174 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30177 * real.c (real_from_integer): Call real_convert even for decimal.
30179 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30182 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30184 2017-02-14 Andrew Pinski <apinski@cavium.com>
30186 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30187 cores and change the partno/implementer to be correct.
30188 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30189 the 'B" as the implementer.
30190 * config/aarch64/aarch64-tune.md: Regenerate.
30192 2017-02-14 Carl Love <cel@us.ibm.com>
30194 * config/rs6000/rs6000.c: Add case statement entry to make the
30195 xvcvuxdsp built-in argument unsigned.
30196 * config/rs6000/vsx.md: Fix the source and return operand types so they
30197 match the instruction definitions from the ISA document. Fix typo
30198 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30201 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30204 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30205 member early_clobber_alts.
30206 * lra-lives.c (reg_early_clobber_p): New.
30207 (process_bb_lives): Use it.
30208 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30209 (debug_operand_data): Initialize early_clobber_alts.
30210 (setup_operand_alternative): Set up early_clobber_alts.
30211 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30212 alternatives to new_insn_reg.
30213 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30215 (lra_update_insn_regno_info): Pass the new arg.
30217 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30219 PR middle-end/79505
30220 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30221 (new_oacc_loop_raw): Don't clear already cleared fields.
30224 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30225 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30226 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30228 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30231 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30233 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30236 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30237 the extra instruction to the right place to store 128-bit constant
30240 2017-02-14 Martin Sebor <msebor@redhat.com>
30242 PR middle-end/79448
30243 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30244 warning for strings of unknown length.
30246 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30248 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30250 2017-02-14 Jeff Law <law@redhat.com>
30253 * ira-costs.c (scan_one_insn): Initialize register move costs
30254 for pseudos seen in USE/CLOBBER insns.
30256 PR tree-optimization/79095
30257 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30258 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30259 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30260 if the operands are known to be not equal, then the resulting range
30262 (intersect_ranges): If the new range is ~[0,0] and the old range is
30263 wide, then prefer ~[0,0].
30264 * tree-vrp.c (overflow_comparison_p_1): New function.
30265 (overflow_comparison_p): New function.
30266 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30267 if NAME is used in an overflow test.
30268 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30269 overflow check that can be expressed as an equality test, then adjust
30270 ops to be that equality test.
30272 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30274 * config/s390/s390-builtin-types.def: Remove flags argument.
30275 * config/s390/s390.c (s390_init_builtins): Likewise.
30277 2017-02-14 Martin Liska <mliska@suse.cz>
30279 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30280 vector. Fix trailing white spaces.
30282 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30284 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30287 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30289 PR rtl-optimization/68664
30290 * config/arm/arm.c (arm_sched_can_speculate_insn):
30291 New function. Declare prototype.
30292 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30294 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30296 PR rtl-optimization/68664
30297 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30299 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30301 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30303 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30304 max skip bytes for function, loop and jump.
30306 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30308 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30309 ABS_EXPR for gimple dump.
30311 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30314 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30316 PR tree-optimization/79408
30317 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30318 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30319 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30320 also if rhs1 is INTEGER_CST.
30322 2017-02-14 Richard Biener <rguenther@suse.de>
30324 PR middle-end/79432
30325 * tree-into-ssa.c (insert_phi_nodes): When the function can
30326 have abnormal edges rewrite SSA names with broken use-def
30327 dominance out of SSA and register them for PHI insertion.
30329 2017-02-13 Martin Sebor <msebor@redhat.com>
30331 PR middle-end/79496
30332 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30333 clearing info.nowrite flag when snprintf size argument is a range.
30335 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30337 * cprop.c (cprop_jump): Add missing space in string literal.
30338 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30339 (get_constraint_for_component_ref): Likewise.
30340 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30341 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30342 * lra-constraints.c (process_alt_operands): Likewise.
30343 * ipa-inline.c (inline_small_functions): Likewise.
30344 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30345 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30346 * trans-mem.c (diagnose_tm_1_op): Likewise.
30347 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30348 (grid_parallel_clauses_gridifiable): Likewise.
30350 * config/nvptx/mkoffload.c (process): Add space in between
30353 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30354 "MOD4_SSE_REGS" and "ALL_REGS".
30356 * spellcheck.c (test_data): Add , in between "foo" and "food".
30358 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30361 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30362 boundary crossing check and subsequent code generation agree.
30364 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30366 * config/aarch64/aarch64.c (has_memory_op): Delete.
30367 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30370 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30372 PR rtl-optimization/79388
30373 PR rtl-optimization/79450
30374 * combine.c (distribute_notes): When removing TEM_INSN for which
30375 corresponding dest has last value recorded, invalidate that last
30378 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30380 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30381 of explicit '@'. Add missing assembly comment marker on branch costs
30384 2017-02-13 Nathan Sidwell <nathan@acm.org>
30386 * gengtype-lex.l (<in_struct>): Add '/'.
30388 2017-02-13 Martin Liska <mliska@suse.cz>
30391 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30393 2017-02-13 Richard Biener <rguenther@suse.de>
30395 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30397 * configure: Re-generate.
30398 * config.in: Likewise.
30399 * graphite-dependences.c: Simplify as if
30400 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30401 * graphite-isl-ast-to-gimple.c: Likewise.
30402 * graphite-optimize-isl.c: Likewise.
30403 * graphite-poly.c: Likewise.
30404 * graphite-sese-to-poly.c: Likewise.
30405 * graphite.h: Likewise.
30406 * toplev.c: Include isl/version.h and use isl_version () for
30407 printing the ISL version.
30408 * doc/install.texi: Update ISL requirement.
30410 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30412 * doc/standards.texi (Standards): Update reference to
30415 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30417 * doc/extend.texi (Named Address Spaces): sourceware.org now
30419 * doc/install.texi (Binaries): Ditto.
30422 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30424 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30425 standard terminology "stringize"/"stringizing" throughout.
30426 * doc/cppinternals.texi: Likewise.
30428 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30430 * doc/extend.texi: Fix some spelling mistakes and typos.
30431 * doc/invoke.texi: Likewise.
30433 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30436 * params.def (inline-min-speedup) Change from 10 to 8.
30438 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30440 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30443 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30446 * ipa-inline-analysis.c (get_minimal_bb): New function.
30447 (record_modified): Use it.
30448 (remap_edge_change_prob): Handle also ancestor functions.
30450 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30452 * doc/contrib.texi (Contributors): Remove broken link into
30453 the Mauve CVS repository.
30455 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30457 PR middle-end/79454
30458 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30459 result computation whenever lhs doesn't have vector mode, not
30460 just when it has BLKmode.
30462 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30464 * doc/makefile.texi (profiledbootstrap): Refer to the
30465 installation instructions only in textual form.
30467 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30470 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30472 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30474 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30475 (Specific): Update mingw-w64 reference.
30477 (Specific): Remove broken link to Renesas RX processor.
30479 2017-02-10 Richard Biener <rguenther@suse.de>
30481 * toplev.c (process_options): Do not mention obsolete graphite
30482 options when printing sorry message about missing graphite support.
30483 Mention -floop-nest-optimize.
30485 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30487 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30488 (vtst_p16): Likewise.
30489 (vtstq_p8): Likewise.
30490 (vtstq_p16): Likewise.
30492 (vtstq_p64): Likewise.
30493 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30494 (vset_lane_p64): New.
30495 (vsetq_lane_p64): New.
30497 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30499 PR tree-optimization/79411
30500 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30501 stmt operands are SSA_NAMEs used in abnormal phis.
30502 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30505 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30508 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30511 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30513 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30515 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30517 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30521 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30522 not arbitrary TREE_CONSTANT.
30525 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30526 "omp declare target link" attribute unless is_global_var.
30527 * omp-offload.c (find_link_var_op): Likewise.
30529 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30530 Chung-Lin Tang <cltang@codesourcery.com>
30532 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30534 (gimplify_adjust_omp_clauses): Don't delete TILE.
30535 (gimplify_omp_for): Deal with TILE.
30536 * internal-fn.c (expand_GOACC_TILE): New function.
30537 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30539 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30540 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30542 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30543 avoid DIV for outermost collapse var.
30544 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30545 Remove out of date comments, fix whitespace.
30546 * omp-general.c (omp_extract_for_data): Deal with tiling.
30547 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30548 adjust OLF_DIM_BASE value.
30549 (struct omp_for_data): Add tiling field.
30550 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30551 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30552 for auto loops. Remove default auto determining, moved to
30553 oacc_loop_fixed_partitions.
30554 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30555 stmts, add e_mask field.
30556 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30557 (oacc_thread_numbers): Use oacc_dim_call.
30558 (oacc_xform_tile): New.
30559 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30560 (finish_oacc_loop): Adjust for ifns vector.
30561 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30562 add case for GOACC_TILE fns.
30563 (oacc_loop_xform_loop): Delete.
30564 (oacc_loop_process): Iterate over call list directly, and add
30565 handling for GOACC_TILE fns.
30566 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30568 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30569 vector partitioning to outer loops. Assign 2 partitions to loops
30570 when available. Add TILE handling.
30571 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30572 (execite_oacc_device_lower): Process GOACC_TILE fns,
30573 ignore unknown specs.
30574 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30575 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30576 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30578 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30580 * configure.ac (ACX_BUGURL): Update.
30581 * configure: Regenerate.
30583 2017-02-09 Richard Biener <rguenther@suse.de>
30585 PR tree-optimization/69823
30586 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30587 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30589 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30591 * config/arc/arc-c.def: Add __NPS400__ definition.
30592 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30593 (TARGET_NPS400): Define.
30595 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30597 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30599 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30600 pointer, arch_info.
30601 (arc_cpu_types): Fill the arch_info field with a pointer into the
30602 arc_arch_types table.
30603 (arc_selected_cpu): Declare.
30604 * config/arc/arc.c (arc_selected_cpu): Make global.
30605 (arc_selected_arch): Delete.
30606 (arc_base_cpu): Delete.
30607 (arc_override_options): Remove references to deleted variables,
30608 update access to arch information.
30609 (ARC_OPT): Update access to arch information.
30610 (ARC_OPTX): Likewise.
30611 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30612 (TARGET_ARC600): Update access to arch information.
30613 (TARGET_ARC601): Likewise.
30614 (TARGET_ARC700): Likewise.
30615 (TARGET_EM): Likewise.
30616 (TARGET_HS): Likewise.
30617 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30620 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30623 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30624 condition/operands for integer GE/LE/GEU/LEU operations.
30626 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30628 PR translation/79397
30629 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30632 2017-02-08 Martin Jambor <mjambor@suse.cz>
30635 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30636 whether allocation happened.
30637 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30638 nothing was allocated.
30640 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30642 PR tree-optimization/79408
30643 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30644 constant, but SSA_NAME with a known integer range, use the minimum
30645 of that range instead of op1 to determine if modulo can be replaced
30646 with its first operand.
30648 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30650 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30652 2017-02-08 Richard Biener <rguenther@suse.de>
30654 PR tree-optimization/71824
30655 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30656 Check all loops contained in the merged region.
30658 2017-02-07 Andrew Pinski <apinski@cavium.com>
30660 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30662 2017-02-07 Andrew Pinski <apinski@cavium.com>
30664 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30665 (thunderxt88): Likewise.
30666 (thunderxt81): Disable LSE and change v8.1 to v8.
30667 (thunderxt83): Likewise.
30669 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30670 Richard Biener <rguenther@suse.de>
30672 PR middle-end/79399
30673 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30674 type from int to size_t.
30675 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30677 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30679 PR rtl-optimization/79386
30680 * cprop.c (bypass_conditional_jumps): Initialize
30681 bypass_last_basic_block already before splitting bbs after
30682 unconditional traps...
30683 (bypass_conditional_jumps): ... rather than here.
30686 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30687 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30688 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30689 fix -masm=intel patterns.
30691 2017-02-07 Richard Biener <rguenther@suse.de>
30693 PR tree-optimization/79256
30694 PR middle-end/79278
30695 * builtins.c (get_object_alignment_2): Use min_align_of_type
30696 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30697 and ADJUST_FIELD_ALIGN.
30699 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30701 * doc/tm.texi: Regenerate.
30702 * stor-layout.c (layout_decl): Adjust.
30703 (update_alignment_for_field): Likewise.
30704 (place_field): Likewise.
30705 (min_align_of_type): Likewise.
30706 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30707 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30708 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30709 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30710 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30711 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30712 * config/i386/i386.c (x86_field_alignment): Likewise.
30713 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30714 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30715 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30716 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30717 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30718 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30722 2017-01-30 Richard Biener <rguenther@suse.de>
30724 PR tree-optimization/79256
30725 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30726 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30729 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30731 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30732 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30733 builtins to SImode and emit a zero-extend, if necessary.
30735 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30737 * docs/invoke.texi (RISC-V Options): Alphabetize.
30739 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30741 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30744 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30746 * config/riscv/riscv.c: New file.
30747 * common/config/riscv/riscv-common.c: Likewise.
30748 * config.gcc: Likewise.
30749 * config/riscv/constraints.md: Likewise.
30750 * config/riscv/elf.h: Likewise.
30751 * config/riscv/generic.md: Likewise.
30752 * config/riscv/linux.h: Likewise.
30753 * config/riscv/multilib-generator: Likewise.
30754 * config/riscv/peephole.md: Likewise.
30755 * config/riscv/pic.md: Likewise.
30756 * config/riscv/predicates.md: Likewise.
30757 * config/riscv/riscv-builtins.c: Likewise.
30758 * config/riscv/riscv-c.c: Likewise.
30759 * config/riscv/riscv-ftypes.def: Likewise.
30760 * config/riscv/riscv-modes.def: Likewise.
30761 * config/riscv/riscv-opts.h: Likewise.
30762 * config/riscv/riscv-protos.h: Likewise.
30763 * config/riscv/riscv.h: Likewise.
30764 * config/riscv/riscv.md: Likewise.
30765 * config/riscv/riscv.opt: Likewise.
30766 * config/riscv/sync.md: Likewise.
30767 * config/riscv/t-elf-multilib: Likewise.
30768 * config/riscv/t-linux: Likewise.
30769 * config/riscv/t-linux-multilib: Likewise.
30770 * config/riscv/t-riscv: Likewise.
30771 * configure.ac: Likewise.
30772 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30773 Waterman as RISC-V maintainers.
30774 * doc/install.texi: Add RISC-V entries.
30775 * doc/invoke.texi: Add RISC-V options section.
30776 * doc/md.texi: Add RISC-V constraints section.
30777 * configure: Regenerated.
30779 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30782 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30783 false values to be constant vectors with all 0 or all 1 bits set.
30784 (vcondu<mode><mode>): Likewise.
30785 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30787 (fpmask_comparison_operator): Update comment.
30788 (vecint_comparison_operator): New predicate.
30789 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30790 vector conditionals when the true and false values are constant
30791 vectors with all 0 bits or all 1 bits set.
30793 2017-02-06 Martin Sebor <msebor@redhat.com>
30795 PR tree-optimization/79376
30796 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30798 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30800 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30801 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30802 to simplify split condition.
30804 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30806 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30807 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30810 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30812 PR rtl-optimization/68664
30813 * target.def (can_speculate_insn): New hook.
30814 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30815 * doc/tm.texi: Regenerate.
30816 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30817 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30818 (rs6000_sched_can_speculate_insn): New function.
30820 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30822 PR tree-optimization/79284
30823 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30824 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30825 vectorizable_mask_load_store, vectorizable_operation,
30826 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30827 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30828 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30829 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30830 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30831 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30832 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30833 is_gimple_assign (stmt). Replace another such test with
30834 is_gimple_assign (stmt).
30836 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30839 * config/avr/avr.c (rtl-iter.h): Include it.
30840 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30841 (avr_legitimate_combined_insn): ...and implementation.
30843 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30845 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30846 * config/s390/s390.c (s390_const_operand_ok)
30847 (s390_canonicalize_comparison, s390_extract_part)
30848 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30849 (s390_contiguous_bitmask_p, s390_rtx_costs)
30850 (legitimize_pic_address): Likewise.
30851 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30852 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30853 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30854 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30855 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30857 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30859 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30860 REGNO($0) == REGNO($1).
30862 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30864 * config/s390/linux.h(SIZE_TYPE): Add comment.
30866 2017-02-06 Julian Brown <julian@codesourcery.com>
30867 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30868 Virendra Pathak <virendra.pathak@broadcom.com>
30870 * config/aarch64/aarch64-cores.def: Change the scheduler
30872 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30873 * config/aarch64/thunderx2t99.md: New file.
30875 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30877 * doc/standards.texi (Go Language): Update link to language
30880 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30882 * tree-eh.c (lower_resx): Sanitize profile.
30883 (cleanup_empty_eh_move_lp): Likewise.
30885 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30888 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30890 * cfgloopmanip.h (loop_version): Update prototype.
30891 * modulo-sched.c (sms_schedule): Update call of loop_version.
30892 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30893 * tree-parloops.c (gen_parallel_loop): Likewise.
30894 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30895 * tree-ssa-loop-split.c (split_loop): Likewise.
30896 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30897 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30899 2017-02-05 Martin Liska <mliska@suse.cz>
30902 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30904 (print_operand_address): Initialize a struct to zero.
30906 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30908 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30909 garbage collector only in textual form.
30911 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30913 * doc/extend.texi (x86 specific memory model extensions for
30914 transactional memory): Simplify a phrase.
30916 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30919 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30920 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30921 (atomic_storedi_1): Likewise.
30923 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30925 PR tree-optimization/79338
30926 * tree-parloops.c (gather_scalar_reductions): Don't call
30927 vect_analyze_loop_form for loop->inner before destroying loop's
30930 2017-02-03 Martin Sebor <msebor@redhat.com>
30932 PR tree-optimization/79327
30933 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30934 when precision has resulted in leading zeros.
30935 (format_integer): Adjust the likely counter to assume an unknown
30936 argument that may be zero is non-zero.
30938 2017-02-03 Jason Merrill <jason@redhat.com>
30941 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30942 avoid copying non-taken branch.
30944 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30946 PR tree-optimization/79340
30947 * tree-vect-loop.c (vectorizable_reduction): Release
30948 vec_defs elements after safe_splicing them into other vectors.
30951 PR tree-optimization/79327
30952 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30953 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30955 (format_integer): Use wide_int_to_tree instead of build_int_cst
30956 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30957 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30958 of shortest and longest sequence.
30960 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30962 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30963 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30965 2017-02-03 Walter Lee <walt@tilera.com>
30968 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30969 after initial stackframe link reg save.
30970 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30972 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30975 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30976 wu for stxssp alternative.
30978 2017-02-03 Martin Sebor <msebor@redhat.com>
30980 PR tree-optimization/79352
30981 * gimple-fold.c (get_range_strlen): Add argument.
30982 (get_range_strlen): Change return type to bool.
30983 (get_maxval_strlen): Pass in a dummy argument.
30984 * gimple-fold.h (get_range_strlen): Change return type to bool.
30985 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30986 * tree.h (array_at_struct_end_p): Add argument.
30987 * tree.c (array_at_struct_end_p): Handle it.
30989 2017-02-03 Martin Liska <mliska@suse.cz>
30992 * multiple_target.c (create_dispatcher_calls): Redirect edge
30993 from a caller of a dispatcher.
30994 (expand_target_clones): Make the clones local.
30995 (ipa_target_clone): Do both target clones and resolvers.
30996 (ipa_dispatcher_calls): Remove the pass.
30997 (pass_dispatcher_calls::gate): Likewise.
30998 (make_pass_dispatcher_calls): Likewise.
30999 * passes.def (pass_target_clone): Put as very first IPA early
31002 2017-02-03 Martin Liska <mliska@suse.cz>
31004 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31005 in case of a function with ifunc attribute.
31007 2017-02-03 Martin Liska <mliska@suse.cz>
31009 * cgraph.c (cgraph_node::dump): Dump function version info.
31010 * symtab.c (symtab_node::dump_base): Add missing new line.
31012 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31014 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31015 (ifcombine_ifandif): Use it.
31017 2017-02-03 Martin Liska <mliska@suse.cz>
31019 * doc/invoke.texi: Document default value for
31020 use-after-scope-direct-emission-threshold.
31022 2017-02-03 Martin Liska <mliska@suse.cz>
31024 PR tree-optimization/79339
31025 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31026 (format_floating): Likewise.
31028 2017-02-03 Martin Liska <mliska@suse.cz>
31031 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31033 (ipa_node_params_t::remove): Likewise.
31034 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31035 initialization from removed ipa_node_params_t::insert.
31036 (ipa_node_params::~ipa_node_params): Move from removed
31037 ipa_node_params_t::release.
31038 * symbol-summary.h (symbol_summary::m_released): New member.
31039 Do not release a summary twice. Do not allow to call finalizer
31040 for types of a summary that live in GGC memory.
31042 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31044 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31047 2017-02-02 Martin Sebor <msebor@redhat.com>
31049 PR middle-end/79275
31050 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31051 (format_string): Tighten up the range of output for non-constant
31052 strings and correct the expected range for wide non-constant strings.
31054 2017-02-02 Martin Sebor <msebor@redhat.com>
31056 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31058 PR middle-end/32003
31059 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31061 (-fdump-tree-@var): Add to index and document how to come up
31062 with pass-specific option and dump file names.
31063 (-fdump-passes): Clarify where to look for output.
31065 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31067 PR middle-end/77445
31068 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31069 statistics of the analyzed path; allow threading for speed when
31070 any of BBs along the path are optimized for speed.
31072 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31074 PR middle-end/78468
31075 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31076 settings of the virtual registers.
31079 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31081 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31082 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31085 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31087 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31088 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31090 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31092 * config/s390/s390.md: Add missing comments with the expanded
31094 * config/s390/vector.md: Likewise.
31095 * config/s390/vx-builtins.md: Likewise.
31097 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31100 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31101 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31102 conditions on a single line.
31104 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31106 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31107 __S390_VX__ to __VX__.
31109 2017-02-01 Andrew Pinski <apinski@cavium.com>
31111 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31112 stmt_info to record_stmt_cost.
31113 (vect_get_known_peeling_cost): Pass stmt_info if known to
31115 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31116 cpu_vector_cost field into
31117 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31118 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31119 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31120 splitting of scalar_stmt_cost and vec_stmt_cost.
31121 (thunderx_vector_cost): Likewise.
31122 (cortexa57_vector_cost): LIkewise.
31123 (exynosm1_vector_cost): Likewise.
31124 (xgene1_vector_cost): Likewise.
31125 (thunderx2t99_vector_cost): Improve after the splitting of the two
31127 (aarch64_builtin_vectorization_cost): Update for the splitting of
31128 scalar_stmt_cost and vec_stmt_cost.
31130 2017-02-01 Torvald Riegel <triegel@redhat.com>
31131 Richard Henderson <rth@redhat.com>
31133 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31134 conditional on existance of a fast atomic load.
31135 * optabs-query.c (can_atomic_load_p): New function.
31136 * optabs-query.h (can_atomic_load_p): Declare it.
31137 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31138 no fast atomic load is available for the particular size of access.
31139 (expand_atomic_compare_and_swap): Likewise.
31140 (expand_atomic_load): Likewise.
31141 (expand_atomic_store): Likewise.
31142 (expand_atomic_fetch_op): Likewise.
31143 * testsuite/lib/target-supports.exp
31144 (check_effective_target_sync_int_128): Remove x86 because it provides
31145 no fast atomic load.
31146 (check_effective_target_sync_int_128_runtime): Likewise.
31148 2017-02-01 Richard Biener <rguenther@suse.de>
31150 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31151 (graphite_transform_loops): Provide opt-info for optimized nests.
31152 * tree-parloop.c (parallelize_loops): Provide opt-info for
31153 parallelized loops.
31155 2017-02-01 Richard Biener <rguenther@suse.de>
31157 PR middle-end/79315
31158 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31159 was not set before.
31161 2017-02-01 Richard Biener <rguenther@suse.de>
31163 PR tree-optimization/71824
31164 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31165 Verify the loops are valid in the merged SESE region.
31166 (scop_detection::can_represent_loop_1): Check analyzing the
31167 evolution of the number of iterations in the region succeeds.
31169 2017-01-31 Ian Lance Taylor <iant@golang.org>
31171 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31172 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31174 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31176 PR preprocessor/79210
31177 * input.c (get_substring_ranges_for_loc): Replace line_width
31178 assertion with error-handling.
31180 2017-01-31 Richard Biener <rguenther@suse.de>
31182 PR tree-optimization/77318
31183 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31184 (create_pw_aff_from_tree): Take loop parameter.
31185 (add_condition_to_pbb): Pass loop of the condition to
31186 create_pw_aff_from_tree.
31188 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31190 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31191 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31193 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31197 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31199 (convert_int_to_float128): Likewise.
31200 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31201 (convert_int_to_float128): Likewise.
31202 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31203 (UNSPEC_IEEE128_CONVERT): Likewise.
31204 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31205 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31206 Use local variables for IBM extended format.
31207 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31208 (fix_trunc<mode>si2_fprs): Likewise.
31209 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31210 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31211 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31212 to know that we can now have integers of all sizes in vector
31214 (fix<uns>_<mode>di2_hw): Likewise.
31215 (float<uns>_<mode>si2_hw): Likewise.
31216 (fix_<mode>si2_hw): Likewise.
31217 (fixuns_<mode>si2_hw): Likewise.
31218 (float<uns>_<mode>di2_hw): Likewise.
31219 (float_<mode>di2_hw): Likewise.
31220 (float_<mode>si2_hw): Likewise.
31221 (floatuns_<mode>di2_hw): Likewise.
31222 (floatuns_<mode>si2_hw): Likewise.
31223 (xscvqp<su>wz_<mode>): Delete, no longer used.
31224 (xscvqp<su>dz_<mode>): Likewise.
31225 (xscv<su>dqp_<mode>): Likewise.
31226 (ieee128_mfvsrd_64bit): Likewise.
31227 (ieee128_mfvsrd_32bit): Likewise.
31228 (ieee128_mfvsrwz): Likewise.
31229 (ieee128_mtvsrw): Likewise.
31230 (ieee128_mtvsrd_64bit): Likewise.
31231 (ieee128_mtvsrd_32bit): Likewise.
31233 2017-01-31 Martin Liska <mliska@suse.cz>
31236 * ipa-prop.c (ipa_free_all_node_params): Call release method
31237 instead of ~sumbol_summary to not to trigger double times
31240 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31242 PR tree-optimization/71691
31243 * bitmap.h (class auto_bitmap): New.
31244 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31245 is_maybe_undefined instead of ssa_undefined_value_p.
31247 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31249 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31250 __S390_ARCH_LEVEL__ to __ARCH__.
31252 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31254 PR tree-optimization/79267
31255 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31256 if should_remove_lhs_p is true.
31258 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31261 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31262 (add_alignment_attribute): New.
31263 (base_type_die): Add alignment attribute.
31264 (subrange_type_die): Likewise.
31265 (modified_type_die): Likewise.
31266 (gen_array_type_die): Likewise.
31267 (gen_descr_array_type_die: Likewise.
31268 (gen_enumeration_type_die): Likewise.
31269 (gen_subprogram_die): Likewise.
31270 (gen_variable_die): Likewise.
31271 (gen_field_die): Likewise.
31272 (gen_ptr_to_mbr_type_die): Likewise.
31273 (gen_struct_or_union_type_die): Likewise.
31274 (gen_subroutine_type_die): Likewise.
31275 (gen_typedef_die): Likewise.
31276 (base_type_cmp): Compare alignment attribute.
31278 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31281 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31282 (setb_unsigned) New pattern for setb with CCUNS.
31283 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31284 subfc./subfe sequence to avoid overflow problems. Generate a
31285 shorter sequence with cmpld/setb for power9.
31286 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31287 for generating subfc. instruction.
31288 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31289 now uses this instruction.
31291 2017-01-30 Ian Lance Taylor <iant@google.com>
31294 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31295 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31297 2017-01-30 Martin Sebor <msebor@redhat.com>
31299 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31300 Move constant to the right of a relational operator.
31301 (get_mpfr_format_length, format_character, format_string): Ditto.
31302 (should_warn_p, maybe_warn): Same.
31304 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31306 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31309 * asan.c (get_translation_unit_decl): Remove function.
31310 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31312 2017-01-30 Martin Liska <mliska@suse.cz>
31314 PR gcov-profile/79259
31315 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31316 -fprofile-generate.
31318 2017-01-30 Martin Liska <mliska@suse.cz>
31321 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31322 Initialize variables with NULL value.
31324 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31327 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31329 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31331 2017-01-30 Richard Biener <rguenther@suse.de>
31333 PR tree-optimization/79276
31334 * tree-vrp.c (process_assert_insertions): Properly adjust common
31335 when removing a duplicate.
31337 2017-01-30 Richard Biener <rguenther@suse.de>
31339 PR tree-optimization/79256
31340 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31341 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31343 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31345 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31348 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31349 ("*r<noxa>sbg_<mode>_sll_bitmask")
31350 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31351 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31352 Use contiguous_bitmask_nowrap_operand.
31354 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31357 * config/rs6000/altivec.h (vec_xl): Revise #define.
31358 (vec_xst): Likewise.
31360 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31362 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31364 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31366 PR rtl-optimization/79194
31367 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31368 traps before call to bypass_conditional_jumps.
31370 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31372 PR tree-optimization/71374
31373 * lra-constraints.c (check_conflict_input_operands): New.
31374 (match_reload): Use it.
31376 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31379 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31380 account to calculate conflict_set.
31382 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31384 PR rtl-optimization/78559
31385 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31386 other_insn in combine.
31388 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31390 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31391 uint16_type_node for BT_UINT16.
31393 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31395 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31396 "RTL Tests" to menu.
31397 (GIMPLE Tests): New node.
31398 (RTL Tests): New node.
31400 2017-01-27 Richard Biener <rguenther@suse.de>
31402 PR tree-optimization/79245
31403 * tree-loop-distribution.c (distribute_loop): Apply cost
31404 modeling also to detected patterns.
31406 2017-01-27 Richard Biener <rguenther@suse.de>
31408 PR tree-optimization/71433
31409 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31410 (compare_assert_loc): New function.
31411 (process_assert_insertions): Sort and optimize assert locations
31412 to remove duplicates and push down identical assertions on
31413 edges to their destination block.
31415 2017-01-27 Richard Biener <rguenther@suse.de>
31417 PR tree-optimization/79244
31418 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31419 out SSA names even if abnormal.
31421 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31423 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31424 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31425 instead of MPFR_RNDN.
31427 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31430 * arm.c (arm_option_override): Don't call build_target_option_node
31431 until after doing all option overrides.
31432 (arm_valid_target_attribute_tree): Likewise.
31434 2017-01-27 Martin Liska <mliska@suse.cz>
31436 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31437 for {cd}tors and C++ {cd}tors.
31439 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31441 * config/s390/s390.md ("*setmem_long_and")
31442 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31444 2017-01-26 Martin Sebor <msebor@redhat.com>
31446 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31449 2017-01-26 Martin Sebor <msebor@redhat.com>
31451 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31452 HAVE_DFmode before using XFmode or DFmode.
31453 (parse_directive): Avoid using the z length modifier to avoid
31454 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31456 PR middle-end/78703
31457 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31458 to accept adjustment as an array.
31459 (get_int_range): New function.
31460 (struct directive): Make width and prec arrays.
31461 (directive::set_width, directive::set_precision): Call get_int_range.
31462 (format_integer, format_floating): Handle width and precision ranges.
31463 (format_string, parse_directive): Same.
31465 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31468 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31469 comdat_type_p set, just clone them, but keep the children in the
31473 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31474 which have direct callers with -fvar-tracking-assignments enabled
31476 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31477 inside of type units.
31479 2017-01-26 Martin Sebor <msebor@redhat.com>
31481 PR middle-end/78703
31482 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31484 (struct format_result): Replace number_chars, number_chars_min,
31485 and number_chars_max with a single member of struct result_range.
31487 (format_result::operator+=): Adjust.
31488 (struct fmtresult): Remove bounded. Handle likely and unlikely
31490 (fmtresult::adjust_for_width_or_precision): New function.
31491 (fmtresult:type_max_digits): New function.
31492 (bytes_remaining): Handle likely and unlikely counters.
31493 (min_bytes_remaining): Remove.
31494 (format_percent): Simplify.
31495 (format_integer, format_floating): Set likely and unlikely counters.
31496 (get_string_length, format_character, format_string): Same.
31497 (format_plain, should_warn_p): New function.
31498 (maybe_warn): Call should_warn_p. Update diagnostic messages
31499 and handle those for all directives, including plain strings.
31500 (format_directive): Handle likely and unlikely counters.
31501 Remove unnecessary quoting from diagnostics. Add an informational
31503 (add_bytes): Remove.
31504 (pass_sprintf_length::compute_format_length): Simplify.
31505 (try_substitute_return_value): Handle likely and unlikely counters.
31507 2017-01-26 Carl Love <cel@us.ibm.com>
31509 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31510 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31512 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31515 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31516 endianess for subregs into account.
31517 * lra-constraints.c (lra_constraints): Do risky transformations
31518 always on the first iteration.
31519 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31520 last_call_used_reg_set.
31521 (process_bb_lives): Define and use last_call_used_reg_set.
31522 * lra.c (lra): Always continue after lra_constraints on the first
31525 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31527 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31529 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31531 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31533 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31534 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31535 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31536 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31537 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31538 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31539 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31540 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31541 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31543 2017-01-26 Marek Polacek <polacek@redhat.com>
31546 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31547 for the third operand.
31549 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31551 PR middle-end/79236
31552 * omp-low.c (struct omp_context): Add simt_stmt field.
31553 (scan_omp_for): Return omp_context *.
31554 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31555 context to the _simt_ SIMD stmt.
31556 (lower_omp_for): For combined SIMD with sibling _simt_
31557 SIMD, make sure to use the same decls in _looptemp_
31558 clauses as in the sibling.
31560 2017-01-26 David Sherwood <david.sherwood@arm.com>
31562 PR middle-end/79212
31563 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31566 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31569 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31570 emit fld b; fld a; if possible.
31572 * brig-builtins.def: Update copyright years.
31573 * config/arm/arm_acle_builtins.def: Update copyright years.
31575 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31578 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31579 constraint instead of o for the stxsd instruction.
31581 2017-01-25 Carl Love <cel@us.ibm.com>
31583 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31584 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31586 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31588 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31590 2017-01-25 Richard Biener <rguenther@suse.de>
31592 PR tree-optimization/69264
31593 * target.def (vector_alignment_reachable): Improve documentation.
31594 * doc/tm.texi: Regenerate.
31595 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31597 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31598 earlier changes with respect to TYPE_USER_ALIGN.
31599 (vector_alignment_reachable_p): Likewise. Improve dumping.
31601 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31604 * config/arm/arm.md (xordi3): Force constant operand into a register
31607 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31609 * doc/invoke.texi (-fstore-merging): Correct default optimization
31610 levels at which it is enabled.
31611 (-O): Move -fstore-merging from list to...
31614 2017-01-25 Richard Biener <rguenther@suse.de>
31617 * omp-expand.c: Include debug.h.
31618 (expand_omp_taskreg): Make sure to generate early debug before
31619 outlining anything from a function.
31620 (expand_omp_target): Likewise.
31621 (grid_expand_target_grid_body): Likewise.
31623 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31626 * asan.c (get_translation_unit_decl): New function.
31627 (asan_add_global): Extract modules file name from globals
31628 TRANSLATION_UNIT_DECL name.
31630 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31633 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31634 for long calls with APCS frame and VFP.
31636 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31638 * cfg.c (original_copy_tables_initialized_p): New function.
31639 * cfg.h (original_copy_tables_initialized_p): New decl.
31640 * cfgrtl.c (relink_block_chain): Guard the call to
31641 free_original_copy_tables with a call to
31642 original_copy_tables_initialized_p.
31643 * cgraph.h (symtab_node::native_rtl_p): New decl.
31644 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31645 (symtab_node::needed_p): Don't assert for early assembly output
31646 for __RTL functions.
31647 (cgraph_node::finalize_function): Set "force_output" for __RTL
31649 (cgraph_node::analyze): Bail out early for __RTL functions.
31650 (analyze_functions): Update assertion to support __RTL functions.
31651 (cgraph_node::expand): Bail out early for __RTL functions.
31652 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31654 * function.h (struct function): Update comment for field
31656 * gimple-expr.c: Include "tree-pass.h".
31657 (gimple_has_body_p): Return false for __RTL functions.
31658 * Makefile.in (OBJS): Add run-rtl-passes.o.
31659 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31661 (gcc::pass_manager::get_clean_slate): New accessor.
31662 * passes.c: Include "insn-addr.h".
31663 (should_skip_pass_p): Add logging. Update logic for running
31664 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31665 property-provider override so it is only done for gimple passes.
31667 (skip_pass): New function.
31668 (execute_one_pass): Call skip_pass when skipping passes.
31669 * read-md.c (md_reader::read_char): Support filtering
31670 the input to a subset of line numbers.
31671 (md_reader::md_reader): Initialize fields
31672 m_first_line and m_last_line.
31673 (md_reader::read_file_fragment): New function.
31674 * read-md.h (md_reader::read_file_fragment): New decl.
31675 (md_reader::m_first_line): New field.
31676 (md_reader::m_last_line): New field.
31677 * read-rtl-function.c (function_reader::create_function): Only
31678 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31679 curr_properties. Set DECL_INITIAL to a dummy block.
31680 (read_rtl_function_body_from_file_range): New function.
31681 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31683 * run-rtl-passes.c: New file.
31684 * run-rtl-passes.h: New file.
31686 2017-01-24 Jeff Law <law@redhat.com>
31688 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31691 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31693 PR tree-optimization/79159
31694 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31695 (record_nonwrapping_iv): Improve boundary using above function if no
31696 value range information.
31698 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31699 Martin Jambor <mjambor@suse.cz>
31701 * brig-builtins.def: New file.
31702 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31703 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31704 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31705 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31706 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31707 * builtin-types.def (BT_INT8): New.
31708 (BT_INT16): Likewise.
31709 (BT_UINT8): Likewise.
31710 (BT_UINT16): Likewise.
31711 (BT_FN_ULONG): Likewise.
31712 (BT_FN_UINT_INT): Likewise.
31713 (BT_FN_UINT_ULONG): Likewise.
31714 (BT_FN_UINT_LONG): Likewise.
31715 (BT_FN_UINT_PTR): Likewise.
31716 (BT_FN_ULONG_PTR): Likewise.
31717 (BT_FN_INT8_FLOAT): Likewise.
31718 (BT_FN_INT16_FLOAT): Likewise.
31719 (BT_FN_UINT32_FLOAT): Likewise.
31720 (BT_FN_UINT16_FLOAT): Likewise.
31721 (BT_FN_UINT8_FLOAT): Likewise.
31722 (BT_FN_UINT64_FLOAT): Likewise.
31723 (BT_FN_UINT16_UINT32): Likewise.
31724 (BT_FN_UINT32_UINT16): Likewise.
31725 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31726 (BT_FN_INT_PTR_INT): Likewise.
31727 (BT_FN_UINT_PTR_UINT): Likewise.
31728 (BT_FN_LONG_PTR_LONG): Likewise.
31729 (BT_FN_ULONG_PTR_ULONG): Likewise.
31730 (BT_FN_VOID_UINT64_UINT64): Likewise.
31731 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31732 (BT_FN_INT8_INT8_INT8): Likewise.
31733 (BT_FN_INT16_INT16_INT16): Likewise.
31734 (BT_FN_INT_INT_INT): Likewise.
31735 (BT_FN_UINT_FLOAT_UINT): Likewise.
31736 (BT_FN_FLOAT_UINT_UINT): Likewise.
31737 (BT_FN_ULONG_UINT_UINT): Likewise.
31738 (BT_FN_ULONG_UINT_PTR): Likewise.
31739 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31740 (BT_FN_UINT_UINT_UINT): Likewise.
31741 (BT_FN_VOID_UINT_PTR): Likewise.
31742 (BT_FN_UINT_UINT_PTR: Likewise.
31743 (BT_FN_UINT32_UINT64_PTR): Likewise.
31744 (BT_FN_INT_INT_UINT_UINT): Likewise.
31745 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31746 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31747 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31748 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31749 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31750 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31751 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31752 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31753 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31754 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31755 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31756 * doc/frontends.texi: List BRIG FE.
31757 * doc/install.texi (Testing): Add BRIG tesring requirements.
31758 * doc/invoke.texi (Overall Options): Mention BRIG.
31759 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31761 2017-01-24 Richard Biener <rguenther@suse.de>
31763 PR translation/79208
31764 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31766 2017-01-24 Martin Jambor <mjambor@suse.cz>
31769 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31770 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31771 and known_contexts.
31773 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31775 PR middle-end/79123
31776 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31777 casts from signed to unsigned really don't have a range.
31779 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31781 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31782 GMP_RNDx for compatiblity.
31784 2017-01-24 Martin Liska <mliska@suse.cz>
31787 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31788 that would prevent us to call alloca with -1 as argument.
31790 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31792 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31793 Avoid -Wformat-security warning.
31795 2017-01-23 Andrew Pinski <apinski@cavium.com>
31797 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31800 2017-01-23 Martin Sebor <msebor@redhat.com>
31802 PR middle-end/78703
31803 * gimple-ssa-sprintf.c (warn_level): New global.
31804 (format_integer): Use it here and throughout the rest of the file.
31805 Use the same switch to compute sign as base.
31806 (maybe_warn): New function.
31807 (format_directive): Factor out warnings into maybe_warn.
31808 Add debugging output. Use warn_level.
31809 (add_bytes): Use warn_level.
31810 (pass_sprintf_length::compute_format_length): Add debugging output.
31811 (try_substitute_return_value): Same.
31812 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31814 PR middle-end/78703
31815 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31816 (struct fmtresult, format_integer, format_floating): Adjust.
31817 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31818 (get_string_length, format_string,format_directive): Same.
31819 (pass_sprintf_length::compute_format_length): Same.
31820 (try_substitute_return_value): Simplify slightly.
31822 PR middle-end/78703
31823 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31824 (fmtresult::operator+=): Outlined.
31825 (struct fmtresult): Add ctors.
31826 (struct conversion_spec): Rename...
31827 (struct directive): ...to this. Add and remove data members.
31828 (directive::set_width, directive::set_precision): New functions.
31829 (format_percent): Use fmtresult ctor.
31830 (get_width_and_precision): Remove.
31831 (format_integer): Make naming changes. Avoid computing width and
31833 (format_floating): Same. Adjust indentation.
31834 (format_character, format_none): New functions.
31835 (format_string): Moved character handling to format_character.
31836 (format_directive): Remove arguments, change return type.
31837 (parse_directive): New function.
31838 (pass_sprintf_length::compute_format_length): Move directive
31839 parsing to parse_directive.
31841 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31843 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31844 (assign_assembler_name_if_needed): ... this.
31845 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31846 (assign_assembler_name_if_needed): ... this.
31847 (free_lang_data_in_cgraph): Adjust callers.
31848 * cgraphunit.c (cgraph_node::analyze): Likewise.
31849 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31852 2017-01-23 Richard Biener <rguenther@suse.de>
31854 PR tree-optimization/79088
31855 PR tree-optimization/79188
31856 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31857 resetting loop bounds after last path deletion. Reset loop
31858 bounds of the target loop, make code match the comments.
31859 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31860 Make sure loops need no fixups.
31862 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31864 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31865 exponent support with double type for first argument.
31866 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31867 type returned by __builtin_vec_extract_sig,
31868 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31869 functions from "vector int" to "vector unsigned int" or from
31870 "vector long long int" to "vector unsigned long long int".
31871 Changed type returned by __builtin_vec_extract_exp,
31872 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31873 functions from "vector int" to "vector unsigned int" or from
31874 "vector long long int" to "vector unsigned long long int".
31875 Changed return type of __builtin_vec_test_data_class,
31876 __builtin_vec_test_data_class_sp, and
31877 __builtin_vec_test_data_class_dp from "vector int" to
31878 "vector bool int" or from "vector long long int" to "vector bool
31879 long long int" and changed second argument type from "unsigned
31880 int" to "int". Added new overloaded function forms "vector float
31881 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31882 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31883 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31884 double, vector unsigned long long int)" and "vector double
31885 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31886 long int)". Changed return type of
31887 __builtin_scalar_test_data_class and
31888 __builtin_scalar_test_data_class_sp and
31889 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31890 int" and changed second argument from "unsigned int" to "int".
31891 Changed type returned by __builtin_scalar_test_neg,
31892 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31893 from "int" to "bool int". Added new overloaded function form
31894 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31895 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31896 exponent double-precision with floating point first argument.
31897 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31898 documentation of scalar_test_data_class, scalar_test_neg,
31899 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31900 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31901 vec_test_data_class built-in functions to reflect refinements in
31902 their type signatures.
31904 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31906 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31908 (aarch64_elf_asm_destructor): Likewise.
31910 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31912 PR rtl-optimization/78634
31913 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31914 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31915 * ifcvt.c (noce_try_cmove): Add missing cost check.
31917 PR rtl-optimization/71724
31918 * combine.c (if_then_else_cond): Look for situations where it is
31919 beneficial to undo the work of one of the recursive calls.
31921 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31923 PR tree-optimization/70754
31924 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31925 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31926 combined stmt before it if not NULL.
31927 (combine_chains): Process refs reversely and compute dominance point
31930 2017-01-23 Martin Liska <mliska@suse.cz>
31932 PR tree-optimization/79196
31933 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31934 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31936 (strlen_optimize_stmt): Call the renamed function.
31938 2017-01-23 Michael Matz <matz@suse.de>
31940 PR tree-optimization/78384
31941 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31943 2017-01-23 Richard Biener <rguenther@suse.de>
31945 PR tree-optimization/79186
31946 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31947 both incoming edges before moving an assert.
31949 2017-01-23 Martin Jambor <mjambor@suse.cz>
31951 * ipa-prop.c (load_from_param_1): Removed.
31952 (load_from_unmodified_param): Bits from load_from_param_1 put back
31954 (load_from_param): Removed.
31955 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31956 with stmt. Reverted back to use of load_from_unmodified_param.
31958 2017-01-23 Martin Jambor <mjambor@suse.cz>
31961 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31962 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31963 field a pointer to garbage collected vector, mark lattices and
31964 ipcp_orig_node with GTY((skip)).
31965 (ipa_get_param_count): Adjust to descriptors being a pointer.
31966 (ipa_get_param): Likewise.
31967 (ipa_get_type): Likewise.
31968 (ipa_get_param_move_cost): Likewise.
31969 (ipa_set_param_used): Likewise.
31970 (ipa_get_controlled_uses): Likewise.
31971 (ipa_set_controlled_uses): Likewise.
31972 (ipa_is_param_used): Likewise.
31973 (ipa_node_params_t): Move into garbage collector. New methods insert
31975 (ipa_node_params_sum): Annotate wth GTY(()).
31976 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31978 (ipa_load_from_parm_agg): Adjust declaration.
31979 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31980 * ipa-profile.c (ipa_profile): Likewise.
31981 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31982 (ipa_populate_param_decls): Make descriptors parameter garbage
31984 (ipa_dump_param): Adjust to descriptors being a pointer.
31985 (ipa_alloc_node_params): Likewise.
31986 (ipa_initialize_node_params): Likewise.
31987 (load_from_param_1): Make descriptors parameter garbage collected.
31988 (load_from_unmodified_param): Likewise.
31989 (load_from_param): Likewise.
31990 (ipa_load_from_parm_agg): Likewise.
31991 (ipa_node_params::~ipa_node_params): Removed.
31992 (ipa_free_all_node_params): Remove call to delete operator.
31993 (ipa_node_params_t::insert): New.
31994 (ipa_node_params_t::remove): Likewise.
31995 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31996 copy known_csts and known_contexts vectors.
31997 (ipa_read_node_info): Adjust to descriptors being a pointer.
31998 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32000 (ipcp_transform_function): Make descriptors variable garbage
32003 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32005 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32006 * config/i386/avx512dqintrin.h: Ditto.
32007 * config/i386/avx512fintrin.h: Ditto.
32008 * config/i386/i386.c: Handle new builtins.
32009 * config/i386/i386-builtin.def: Add new builtins.
32010 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32011 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32013 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32014 Martin Liska <mliska@suse.cz>
32016 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32017 * asan.c (asan_expand_poison_ifn): Support stores and use
32018 appropriate ASAN report function.
32019 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32020 * internal-fn.def (ASAN_POISON_USE): Declare.
32021 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32022 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32023 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32024 ASAN_POISON calls w/o LHS.
32025 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32026 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32027 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32028 * gimplify.c (asan_poison_variables): Add attribute
32029 use_after_scope_memory to variables that really needs to live
32031 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32032 having the attribute.
32034 2017-01-23 Martin Liska <mliska@suse.cz>
32036 * asan.c (create_asan_shadow_var): New function.
32037 (asan_expand_poison_ifn): Likewise.
32038 * asan.h (asan_expand_poison_ifn): New declaration.
32039 * internal-fn.c (expand_ASAN_POISON): Likewise.
32040 * internal-fn.def (ASAN_POISON): New builtin.
32041 * sanopt.c (pass_sanopt::execute): Expand
32042 asan_expand_poison_ifn.
32043 * tree-inline.c (copy_decl_for_dup_finish): Make function
32045 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32046 * tree-ssa.c (is_asan_mark_p): New function.
32047 (execute_update_addresses_taken): Rewrite local variables
32048 (identified just by use-after-scope as addressable) into SSA.
32050 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32052 * doc/install.texi (Specific): opensource.apple.com uses https
32053 now. Remove trailing slash.
32055 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32057 * README.Portability: Remove note on an Irix compatibility issue.
32059 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32061 * gcov.c (INCLUDE_ALGORITHM): Define.
32062 (INCLUDE_VECTOR): Define.
32063 No longer include <vector> and <algorithm> directly.
32065 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32067 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32069 * doc/invoke.texi (Code Gen Options): Ditto.
32071 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32074 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32076 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32078 rtl-optimization/79125
32079 * cprop.c (local_cprop_pass): Handle cases where we make an
32080 unconditional trap.
32082 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32086 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32087 read from, for big endian.
32089 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32091 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32092 register pauth builtins for LP64 only.
32094 2017-01-20 Marek Polacek <polacek@redhat.com>
32097 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32100 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32102 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32104 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32105 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32106 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32108 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32111 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32112 in big-endian mode when they are not a single duplicated value.
32114 2017-01-20 Richard Biener <rguenther@suse.de>
32116 * BASE-VER: Bump to 7.0.1.
32118 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32120 * omp-low.c (omplow_simd_context): New struct. Use it...
32121 (lower_rec_simd_input_clauses): ...here and...
32122 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32123 references to idx, lane, max_vf, is_simt.
32125 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32127 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32130 2017-01-20 Martin Jambor <mjambor@suse.cz>
32132 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32133 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32135 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32136 (GTFILES): Rename hsa.c to hsa-common.c.
32137 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32138 * hsa-dump.c: Likewise.
32139 * hsa-gen.c: Likewise.
32140 * hsa-regalloc.c: Likewise.
32141 * ipa-hsa.c: Likewise.
32142 * omp-expand.c: Likewise.
32143 * omp-low.c: Likewise.
32144 * toplev.c: Likewise.
32146 2017-01-20 Marek Polacek <polacek@redhat.com>
32149 * doc/invoke.texi: Document -Wduplicated-branches.
32150 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32151 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32152 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32153 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32154 return 0 only when not OEP_LEXICOGRAPHIC.
32155 (fold_build_cleanup_point_expr): Use the expression
32156 location when building CLEANUP_POINT_EXPR.
32157 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32158 * tree.c (add_expr): Handle error_mark_node.
32160 2017-01-20 Martin Liska <mliska@suse.cz>
32163 * tree-profile.c (init_ic_make_global_vars): Do not call
32165 (gimple_init_gcov_profiler): Likewise.
32167 2017-01-20 Martin Liska <mliska@suse.cz>
32170 * cgraph.h (maybe_create_reference): Remove argument and
32172 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32174 * ipa-cp.c (create_specialized_node): Likewise.
32175 * symtab.c (symtab_node::maybe_create_reference): Handle
32176 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32178 2017-01-20 Martin Liska <mliska@suse.cz>
32180 * read-rtl-function.c (function_reader::create_function): Use
32181 build_decl instread of build_decl_stat.
32183 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32185 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32186 * config/i386/avx512dqintrin.h: Ditto.
32187 * config/i386/avx512fintrin.h: Ditto.
32188 * config/i386/i386-builtin-types.def: Add new types.
32189 * config/i386/i386.c: Handle new types.
32190 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32191 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32192 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32193 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32194 (__builtin_ia32_kshiftridi): New.
32195 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32197 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32201 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32202 define to rs6000_init_stack_protect_guard.
32203 (rs6000_init_stack_protect_guard): New function.
32205 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32206 Yunqiang Su <yunqiang.su@imgtec.com>
32208 * config.gcc (supported_defaults): Add madd4.
32209 (with_madd4): Add validation.
32210 (all_defaults): Add madd4.
32211 * config/mips/mips.opt (mmadd4): New option.
32212 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32214 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32216 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32217 (ISA_HAS_FUSED_MADD4): Likewise.
32218 * doc/invoke.texi (-mmadd4): Document the new option.
32219 * doc/install.texi (--with-madd4): Document the new option.
32221 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32223 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32224 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32225 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32226 (aarch64_init_pauth_hint_builtins): New.
32227 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32228 (aarch64_expand_builtin): Expand new builtins.
32230 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32232 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32233 * combine-stack-adj.c (no_unhandled_cfa): Handle
32234 REG_CFA_TOGGLE_RA_MANGLE.
32235 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32236 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32237 info for return address signing.
32238 (aarch64_expand_epilogue): Likewise.
32240 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32242 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32243 * config/aarch64/aarch64-protos.h
32244 (aarch64_return_address_signing_enabled): New declaration.
32245 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32247 (aarch64_expand_prologue): Sign return address before it's pushed onto
32249 (aarch64_expand_epilogue): Authenticate return address fetched from
32251 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32252 (aarch64_attributes): New function attributes for "sign-return-address".
32253 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32254 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32255 ("*do_return"): Generate combined instructions according to key index.
32256 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32257 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32259 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32260 * config/aarch64/aarch64.opt (msign-return-address=): New.
32261 * doc/extend.texi (AArch64 Function Attributes): Documents
32262 "sign-return-address=".
32263 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32265 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32267 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32268 overall option summary.
32270 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32272 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32273 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32274 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32275 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32277 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32279 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32280 -mpower9-minmax by default for -mcpu=power9.
32281 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32282 128-bit floating point.
32284 2017-01-20 Alan Modra <amodra@gmail.com>
32286 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32287 optimizing for size.
32289 2017-01-20 Alan Modra <amodra@gmail.com>
32292 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32293 for strcmp and strncmp from corresponding builtin decl.
32295 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32297 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32298 instead of i386/rtems-64.h.
32299 * config/i386/rtems-64.h: Remove.
32301 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32305 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32307 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32309 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32311 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32312 Change int to HOST_WIDE_INT.
32313 * config/aarch64/aarch64-protos.h
32314 (aarch64_simd_gen_const_vector_dup): Likewise.
32315 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32317 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32319 * langhooks-def.h (lhd_type_for_size): New decl.
32320 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32321 * langhooks.c (lhd_type_for_size): New function, taken from
32324 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32326 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32327 define_bypass for CR latency.
32328 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32329 (power9-alu2): Add define_bypass for CR latency.
32331 (power9-mul): Update insn latency.
32332 (power9-mul-compare): Update insn latency, bypass latency and remove
32335 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32337 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32339 * config/aarch64/aarch64.md
32340 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32341 aarch64_nopcrelative_literal_loads.
32342 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32344 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32346 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32347 TARGET_LOONGSON_3A.
32348 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32350 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32353 * config.gcc (supported_defaults): Add lxc1-sxc1.
32354 (with_lxc1_sxc1): Add validation.
32355 (all_defaults): Add lxc1-sxc1.
32356 * config/mips/mips.opt (mlxc1-sxc1): New option.
32357 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32359 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32360 __mips_no_lxc1_sxc1.
32361 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32362 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32363 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32365 2017-01-19 Richard Biener <rguenther@suse.de>
32367 PR tree-optimization/72488
32368 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32369 sure to restore SSA info.
32370 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32372 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32374 PR rtl-optimization/79121
32375 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32376 of the inner type when shifting an extended value.
32378 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32381 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32382 interposable aliases.
32384 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32387 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32388 Use the evmergelohi instruction.
32389 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32390 (mov_sitf_e500_subreg8_2_be): Likewise.
32391 (mov_sitf_e500_subreg12_2_le): Likewise.
32392 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32393 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32394 (mov_sitf_e500_subreg8_2_le): Likewise.
32395 (mov_sitf_e500_subreg12_2_be): Likewise.
32397 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32399 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32400 attribute from vecsimple to vecperm.
32401 (altivec_vbpermq2): Likewise.
32403 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32406 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32408 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32409 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32410 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32411 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32412 case where N arg is SIZE_MAX.
32413 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32414 (cmpstrsi): Add pattern.
32416 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32418 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32419 __builtin_vec_revb builtins.
32420 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32421 built-in functions to support generation of the ISA 3.0 XXBR<x>
32422 vector byte reverse instructions.
32423 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32424 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32425 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32426 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32427 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32428 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32429 (P9V_BUILTIN_VEC_REVB): Likewise.
32430 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32431 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32432 (p9_xxbrq_v16qi): Likewise.
32433 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32434 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32435 (p9_xxbrh_v8hi): Likewise.
32436 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32437 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32438 vec_revb built-in functions.
32440 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32442 PR rtl-optimization/78952
32443 * config/i386/i386.md (any_extract): New code iterator.
32444 (*insvqi_2): Use any_extract for source operand.
32445 (*insvqi_3): Use any_shiftrt for source operand.
32447 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32449 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32451 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32453 2017-01-18 Matthias Klose <doko@ubuntu.com>
32455 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32457 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32459 * config/rs6000/altivec.h (vec_bperm): Change #define.
32460 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32461 (altivec_vbpermq2): New define_insn.
32462 (altivec_vbpermd): Likewise.
32463 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32464 function interface.
32465 (VBPERMD): Likewise.
32466 (VBPERM): New polymorphic function interface.
32467 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32468 Add entries for P9V_BUILTIN_VEC_VBPERM.
32469 * doc/extend.texi: Add interfaces for vec_bperm.
32471 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32473 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32474 first letter of error messages.
32475 (s390_resolve_overloaded_builtin): Likewise.
32476 * config/s390/s390.c (s390_expand_builtin): Likewise.
32477 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32478 (s390_valid_target_attribute_inner_p): Likewise.
32479 * config/s390/s390.md ("tabort"): Likewise.
32481 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32483 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32484 (ISA_AVOID_DIV_HILO): New macro.
32485 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32486 (ISA_HAS_DDIV): Likewise.
32488 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32490 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32492 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32494 * doc/invoke.texi (fabi-version): Spelling fix.
32496 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32499 * doc/invoke.texi (fabi-version): Mention mangling fix for
32502 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32505 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32507 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32510 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32511 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32513 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32515 (stack_protect_test): Ditto.
32516 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32517 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32519 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32520 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32521 -mstack-protector-guard-offset=.
32522 (RS/6000 and PowerPC Options): Ditto.
32524 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32526 * config/i386/i386.h (MASK_CLASS_P): New define.
32527 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32528 there are no registers from different register sets also when
32529 mask registers are used. Update function comment.
32530 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32531 to (*k/*r) and (*k/*km) alternatives.
32533 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32535 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32536 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32537 (EH_RETURN_HANDLER_RTX): New define.
32538 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32539 Force frame pointer in EH return functions.
32540 (aarch64_expand_epilogue): Add barrier for eh_return.
32541 (aarch64_final_eh_return_addr): Remove.
32542 (aarch64_eh_return_handler_rtx): New function.
32543 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32545 (aarch64_eh_return_handler_rtx): New prototype.
32547 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32549 * config/rs6000/altivec.h (vec_rlmi): New #define.
32550 (vec_vrlnm): Likewise.
32551 (vec_rlnm): Likewise.
32552 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32553 (UNSPEC_VRLNM): Likewise.
32554 (VIlong): New mode iterator.
32555 (altivec_vrl<VI_char>mi): New define_insn.
32556 (altivec_vrl<VI_char>nm): Likewise.
32557 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32559 (VRLDNM): Likewise.
32560 (RLNM): New polymorphic function entry.
32561 (VRLWMI): New monomorphic function entry.
32562 (VRLDMI): Likewise.
32563 (RLMI): New polymorphic function entry.
32564 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32565 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32566 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32569 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32572 * dwarf2out.c (field_byte_offset): Restore the
32573 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32574 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32575 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32578 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32581 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32583 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32586 * dwarf2out.c (add_data_member_location_attribute): For constant
32587 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32588 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32589 DW_AT_byte_size attributes.
32591 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32593 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32594 after forcing to constant memory when the code model is medium.
32596 2017-01-17 Julia Koval <julia.koval@intel.com>
32599 * config/i386/avx512fintrin.h
32600 (_mm512_i32gather_ps): Change __addr type to void const*.
32601 (_mm512_mask_i32gather_ps): Ditto.
32602 (_mm512_i32gather_pd): Ditto.
32603 (_mm512_mask_i32gather_pd): Ditto.
32604 (_mm512_i64gather_ps): Ditto.
32605 (_mm512_mask_i64gather_ps): Ditto.
32606 (_mm512_i64gather_pd): Ditto.
32607 (_mm512_mask_i64gather_pd): Ditto.
32608 (_mm512_i32gather_epi32): Ditto.
32609 (_mm512_mask_i32gather_epi32): Ditto.
32610 (_mm512_i32gather_epi64): Ditto.
32611 (_mm512_mask_i32gather_epi64): Ditto.
32612 (_mm512_i64gather_epi32): Ditto.
32613 (_mm512_mask_i64gather_epi32): Ditto.
32614 (_mm512_i64gather_epi64): Ditto.
32615 (_mm512_mask_i64gather_epi64): Ditto.
32616 (_mm512_i32scatter_ps): Change __addr type to void*.
32617 (_mm512_mask_i32scatter_ps): Ditto.
32618 (_mm512_i32scatter_pd): Ditto.
32619 (_mm512_mask_i32scatter_pd): Ditto.
32620 (_mm512_i64scatter_ps): Ditto.
32621 (_mm512_mask_i64scatter_ps): Ditto.
32622 (_mm512_i64scatter_pd): Ditto.
32623 (_mm512_mask_i64scatter_pd): Ditto.
32624 (_mm512_i32scatter_epi32): Ditto.
32625 (_mm512_mask_i32scatter_epi32): Ditto.
32626 (_mm512_i32scatter_epi64): Ditto.
32627 (_mm512_mask_i32scatter_epi64): Ditto.
32628 (_mm512_i64scatter_epi32): Ditto.
32629 (_mm512_mask_i64scatter_epi32): Ditto.
32630 (_mm512_i64scatter_epi64): Ditto.
32631 (_mm512_mask_i64scatter_epi64): Ditto.
32632 * config/i386/avx512pfintrin.h
32633 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32634 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32635 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32636 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32637 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32638 (_mm512_prefetch_i32scatter_ps): Ditto.
32639 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32640 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32641 (_mm512_prefetch_i64scatter_pd): Ditto.
32642 (_mm512_prefetch_i64scatter_ps): Ditto.
32643 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32644 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32645 * config/i386/avx512vlintrin.h
32646 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32647 (_mm_mmask_i32gather_ps): Ditto.
32648 (_mm256_mmask_i32gather_pd): Ditto.
32649 (_mm_mmask_i32gather_pd): Ditto.
32650 (_mm256_mmask_i64gather_ps): Ditto.
32651 (_mm_mmask_i64gather_ps): Ditto.
32652 (_mm256_mmask_i64gather_pd): Ditto.
32653 (_mm_mmask_i64gather_pd): Ditto.
32654 (_mm256_mmask_i32gather_epi32): Ditto.
32655 (_mm_mmask_i32gather_epi32): Ditto.
32656 (_mm256_mmask_i32gather_epi64): Ditto.
32657 (_mm_mmask_i32gather_epi64): Ditto.
32658 (_mm256_mmask_i64gather_epi32): Ditto.
32659 (_mm_mmask_i64gather_epi32): Ditto.
32660 (_mm256_mmask_i64gather_epi64): Ditto.
32661 (_mm_mmask_i64gather_epi64): Ditto.
32662 (_mm256_i32scatter_ps): Change __addr type to void*.
32663 (_mm256_mask_i32scatter_ps): Ditto.
32664 (_mm_i32scatter_ps): Ditto.
32665 (_mm_mask_i32scatter_ps): Ditto.
32666 (_mm256_i32scatter_pd): Ditto.
32667 (_mm256_mask_i32scatter_pd): Ditto.
32668 (_mm_i32scatter_pd): Ditto.
32669 (_mm_mask_i32scatter_pd): Ditto.
32670 (_mm256_i64scatter_ps): Ditto.
32671 (_mm256_mask_i64scatter_ps): Ditto.
32672 (_mm_i64scatter_ps): Ditto.
32673 (_mm_mask_i64scatter_ps): Ditto.
32674 (_mm256_i64scatter_pd): Ditto.
32675 (_mm256_mask_i64scatter_pd): Ditto.
32676 (_mm_i64scatter_pd): Ditto.
32677 (_mm_mask_i64scatter_pd): Ditto.
32678 (_mm256_i32scatter_epi32): Ditto.
32679 (_mm256_mask_i32scatter_epi32): Ditto.
32680 (_mm_i32scatter_epi32): Ditto.
32681 (_mm_mask_i32scatter_epi32): Ditto.
32682 (_mm256_i32scatter_epi64): Ditto.
32683 (_mm256_mask_i32scatter_epi64): Ditto.
32684 (_mm_i32scatter_epi64): Ditto.
32685 (_mm_mask_i32scatter_epi64): Ditto.
32686 (_mm256_i64scatter_epi32): Ditto.
32687 (_mm256_mask_i64scatter_epi32): Ditto.
32688 (_mm_i64scatter_epi32): Ditto.
32689 (_mm_mask_i64scatter_epi32): Ditto.
32690 (_mm256_i64scatter_epi64): Ditto.
32691 (_mm256_mask_i64scatter_epi64): Ditto.
32692 (_mm_i64scatter_epi64): Ditto.
32693 (_mm_mask_i64scatter_epi64): Ditto.
32694 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32695 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32696 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32697 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32698 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32699 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32700 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32701 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32702 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32703 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32704 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32705 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32706 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32707 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32708 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32709 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32710 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32711 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32712 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32713 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32714 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32715 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32716 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32717 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32718 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32719 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32720 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32721 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32722 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32723 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32724 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32725 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32726 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32727 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32728 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32729 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32730 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32731 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32732 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32733 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32734 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32735 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32736 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32737 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32738 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32739 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32740 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32741 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32742 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32743 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32744 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32745 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32746 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32747 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32748 definitions accordingly.
32750 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32751 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32754 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32757 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32760 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32761 hard regs for inner regno.
32763 2017-01-17 Martin Liska <mliska@suse.cz>
32766 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32767 assumption and add comment.
32769 2017-01-17 Nathan Sidwell <nathan@acm.org>
32771 * ipa-visibility.c (localize_node): New function, broken out of ...
32772 (function_and_variable_visibility): ... here. Call it.
32774 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32776 PR middle-end/77445
32777 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32778 correctly set frequency of oudgoing edge.
32779 (duplicate_thread_path): Fix profile updating.
32781 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32784 * configure.ac: Add GCC_BASE_VER.
32785 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32786 version from BASE-VER file.
32787 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32788 (gcc.o): Depend on $(BASEVER).
32789 * common.opt (dumpfullversion): New option.
32790 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32791 * doc/invoke.texi: Document -dumpfullversion.
32792 * doc/install.texi: Document --with-gcc-major-version-only.
32793 * configure: Regenerated.
32795 2017-01-17 Richard Biener <rguenther@suse.de>
32797 PR tree-optimization/71433
32798 * tree-vrp.c (register_new_assert_for): Merge same asserts
32799 on all incoming edges.
32800 (process_assert_insertions_for): Handle insertions at the
32803 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32805 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32806 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32808 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32811 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32814 2017-01-17 Alan Modra <amodra@gmail.com>
32817 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32818 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32819 symbolic stack limit when pic.
32821 2017-01-16 Martin Sebor <msebor@redhat.com>
32823 PR tree-optimization/78608
32824 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32826 2017-01-16 Jeff Law <law@redhat.com>
32829 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32830 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32831 for several include directories that may be relative to sysroot.
32832 * config/i386/x-mingw32 (gplus_includedir): Define.
32833 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32834 (native_system_includedir): Likewise.
32835 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32836 override if TARGET_SYSTEM_ROOT is defined.
32837 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32839 PR tree-optimization/79090
32840 PR tree-optimization/33562
32841 PR tree-optimization/61912
32842 PR tree-optimization/77485
32843 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32844 and computed trims into the dump file.
32846 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32848 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32850 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32853 * gimplify.c (gimplify_init_constructor): If want_value and
32854 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32858 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32859 sequence. Formatting fixes.
32860 (doloop_optimize): Formatting fixes.
32863 * gcc.c (debug_level_greater_than_spec_func): New function.
32864 (static_spec_functions): Add debug-level-gt spec function.
32865 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32867 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32868 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32869 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32870 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32871 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32872 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32874 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32876 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32877 QImode fixups to general and mask registers only.
32879 2017-01-16 Carl Love <cel@us.ibm.com>
32881 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32882 for built-in functions
32883 vector signed char vec_nabs (vector signed char)
32884 vector signed short vec_nabs (vector signed short)
32885 vector signed int vec_nabs (vector signed int)
32886 vector signed long long vec_nabs (vector signed long long)
32887 vector float vec_nabs (vector float)
32888 vector double vec_nabs (vector double)
32889 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32891 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32892 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32893 * doc/extend.texi: Update the documentation file for the new built-in
32896 2017-01-16 Martin Sebor <msebor@redhat.com>
32898 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32901 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32903 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32904 UNSPEC_VSX__XXSPLTD to require special splat handling.
32906 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32909 * system.h: Poison strndup.
32911 2017-01-16 Alan Modra <amodra@gmail.com>
32914 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32917 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32919 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32921 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32923 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32924 call recog here. Assert that INSN_CODE (insn) is non-negative.
32926 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32929 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32931 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32932 in the currently scheduled RTL fragment.
32934 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32936 PR rtl-optimization/78751
32937 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32940 2017-01-14 Jeff Law <law@redhat.com>
32942 PR tree-optimization/79090
32943 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32944 variable length stores.
32945 (compute_trims): Delete dead assignment to *trim_tail.
32946 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32949 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32951 PR rtl-optimization/78626
32952 PR rtl-optimization/78727
32953 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32954 of a block, and split such blocks after everything else is finished.
32956 2017-01-14 Alan Modra <amodra@gmail.com>
32959 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32960 target legitimate_combined_insn.
32961 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32962 (rs6000_legitimate_combined_insn): New function.
32963 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32965 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32966 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32967 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32969 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32971 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32973 2017-01-13 Jeff Law <law@redhat.com>
32975 PR tree-optimization/33562
32976 PR tree-optimization/61912
32977 PR tree-optimization/77485
32978 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32980 (delete_dead_assignment): Likewise.
32981 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32982 statement to delete_dead_call and delete_dead_assignment.
32984 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32987 * substring-locations.c (format_warning_va): Strengthen case 1 so
32988 that both endpoints of the substring must be within the format
32989 range for just the substring to be printed.
32991 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32993 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32994 * config/i386/i386.c (ix86_target_string): Add missing options
32995 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32996 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32997 flags_other and ix86_target_other to flags2_other. Display unknown
32999 (ix86_valid_target_attribute_inner_p): Add missing options and
33000 reorder options by implied ISAs, as in ix86_target_string.
33002 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33004 * hash-table.h (hash_table::too_empty_p): New function.
33005 (hash_table::expand): Use it.
33006 (hash_table::traverse): Likewise.
33007 (hash_table::empty_slot): Use sizeof (value_type) instead of
33008 sizeof (PTR) to convert bytes to elements. Shrink the table
33009 if the current size is excessive for the current number of
33012 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33014 * ira-costs.c (record_reg_classes): Break from the inner loop
33015 early once alt_fail is known to be true. Update outer loop
33016 handling accordingly.
33018 2017-01-13 Jeff Law <law@redhat.com>
33020 * tree-ssa-dse.c (decrement_count): New function.
33021 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33022 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33023 when we know the partially dead statement is a mem* function.
33025 PR tree-optimization/61912
33026 PR tree-optimization/77485
33027 * tree-ssa-dse.c: Include expr.h.
33028 (maybe_trim_constructor_store): New function.
33029 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33031 PR tree-optimization/33562
33032 PR tree-optimization/61912
33033 PR tree-optimization/77485
33034 * doc/invoke.texi: Document new dse-max-object-size param.
33035 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33036 * tree-ssa-dse.c: Include params.h.
33037 (dse_store_status): New enum.
33038 (initialize_ao_ref_for_dse): New, partially extracted from
33040 (valid_ao_ref_for_dse, normalize_ref): New.
33041 (setup_live_bytes_from_ref, compute_trims): Likewise.
33042 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33043 (maybe_trim_partially_dead_store): Likewise.
33044 (maybe_trim_complex_store): Likewise.
33045 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33046 Track what bytes live from the original store. Return tri-state
33047 for dead, partially dead or live.
33048 (dse_dom_walker): Add constructor, destructor and new private members.
33049 (delete_dead_call, delete_dead_assignment): New extracted from
33051 (dse_optimize_stmt): Make a member of dse_dom_walker.
33052 Use initialize_ao_ref_for_dse.
33054 PR tree-optimization/33562
33055 PR tree-optimization/61912
33056 PR tree-optimization/77485
33057 * sbitmap.h (bitmap_count_bits): Prototype.
33058 (bitmap_clear_range, bitmap_set_range): Likewise.
33059 * sbitmap.c (bitmap_clear_range): New function.
33060 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33062 2017-01-13 Martin Liska <mliska@suse.cz>
33065 * function.c (set_cfun): Add new argument force.
33066 * function.h (set_cfun): Likewise.
33067 * ipa-inline-transform.c (inline_call): Use the function when
33068 strict alising from is dropped for function we inline to.
33070 2017-01-13 Richard Biener <rguenther@suse.de>
33072 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33073 for dumping GIMPLE INTEGER_CSTs.
33075 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33077 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33078 to 201112L since C++17.
33080 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33083 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33084 if -fsanitize=kernel-address is present.
33086 2017-01-13 Richard Biener <rguenther@suse.de>
33088 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33089 as _Literal ( type ) number in case usual suffixes do not
33090 preserve all information.
33092 2017-01-13 Richard Biener <rguenther@suse.de>
33094 PR tree-optimization/77283
33095 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33096 and ssa-iterators.h.
33097 (is_feasible_trace): Implement a cost model based on joiner
33100 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33103 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33104 char or short to __float128/_Float128 directly.
33106 2017-01-12 Martin Sebor <msebor@redhat.com>
33108 to -Wformat-overflow.
33109 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33110 (min_bytes_remaining): Same.
33111 (get_string_length): Same.
33112 (format_string): Same.
33113 (format_directive): Same.
33115 (pass_sprintf_length::handle_gimple_call): Same.
33117 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33119 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33120 info.nowrite calls with no lhs that can't throw. Return bool
33121 whether gsi_remove has been called or not.
33122 (pass_sprintf_length::handle_gimple_call): Return bool whether
33123 try_substitute_return_value called gsi_remove. Formatting fix.
33124 (pass_sprintf_length::execute): Don't use gsi_remove if
33125 handle_gimple_call returned true.
33128 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33129 be removed due to side-effects, don't remove following barrier nor
33130 turn the successor edge into fallthru edge.
33132 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33135 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33136 element-reversing loads and stores as not swappable.
33138 2017-01-12 Nathan Sidwell <nathan@acm.org>
33139 Nicolai Stange <nicstange@gmail.com>
33141 * combine.c (try_combine): Don't ignore result of overlap checking
33142 loop. Combine overlap & asm check into single loop.
33144 2017-01-12 Richard Biener <rguenther@suse.de>
33146 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33147 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33149 2017-01-12 Richard Biener <rguenther@suse.de>
33151 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33152 and TS_TARGET_OPTION directly derive from TS_BASE.
33153 * tree-core.h (tree_optimization_option): Derive from tree_base.
33154 (tree_target_option): Likewise.
33156 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33158 * config/i386/i386.c (memory_address_length): Increase len
33159 only when rip_relative_addr_p returns false.
33161 2017-01-11 Julia Koval <julia.koval@intel.com>
33163 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33164 (OPTION_MASK_ISA_SGX_SET): New.
33165 (ix86_handle_option): Handle OPT_msgx.
33166 * config.gcc: Added sgxintrin.h.
33167 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33168 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33169 * config/i386/i386.c (ix86_target_string): Add -msgx.
33171 (ix86_option_override_internal): Handle new options.
33172 (ix86_valid_target_attribute_inner_p): Add sgx.
33173 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33174 * config/i386/i386.opt: Add msgx.
33175 * config/i386/sgxintrin.h: New file.
33176 * config/i386/x86intrin.h: Add sgxintrin.h.
33178 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33181 * fold-const.c (maybe_nonzero_address): Return 1 for function
33183 (tree_single_nonzero_warnv_p): Don't handle function local objects
33187 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33190 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33193 * opts.c: Include "spellcheck.h"
33194 (struct string_fragment): New struct.
33195 (struct edit_distance_traits<const string_fragment &>): New
33197 (get_closest_sanitizer_option): New function.
33198 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33200 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33202 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33204 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33205 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33206 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33207 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33208 for initial die_offset if dwarf_split_debug_info.
33209 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33210 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33212 (output_skeleton_debug_sections): Formatting fix. Use
33213 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33214 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33216 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33218 * config/arm/cortex-a53.md: Add bypasses for
33219 cortex_a53_r2f_cvt.
33220 (cortex_a53_r2f): Only use for transfers.
33221 (cortex_a53_f2r): Likewise.
33222 (cortex_a53_r2f_cvt): Add reservation for conversions.
33223 (cortex_a53_f2r_cvt): Likewise.
33225 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33227 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33228 to all inlined functions, change static to extern.
33230 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33233 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33235 (arm_assemble_integer): Likewise.
33237 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33239 * config.gcc: Use new awk script to check CPU, FPU and architecture
33240 parameters for --with-... options.
33241 * config/arm/parsecpu.awk: New file
33242 * config/arm/arm-cpus.in: New file.
33243 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33245 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33247 * config/arm/t-arm: Update dependency rules.
33248 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33249 of processing .def files.
33250 * config/arm/genopt.sh: Deleted.
33251 * config/arm/gentune.sh: Deleted.
33252 * config/arm/arm-cores.def: Deleted.
33253 * config/arm/arm-arches.def: Deleted.
33254 * config/arm/arm-fpus.def: Deleted.
33255 * config/arm/arm-tune.md: Regenerated.
33256 * config/arm/arm-tables.opt: Regenerated.
33257 * config/arm/arm-cpu.h: New generated file.
33258 * config/arm/arm-cpu-data.h: New generated file.
33259 * config/arm/arm-cpu-cdata.h: New generated file.
33261 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33264 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33266 (input_varpool_node): Unpack dynamically_initialized bit.
33268 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33270 PR rtl-optimization/79032
33271 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33272 the alignment of the adjusted memory reference against that of MODE,
33273 instead of the alignment of the original memory reference.
33275 2017-01-11 Martin Jambor <mjambor@suse.cz>
33277 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33279 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33280 decorated functions.
33282 2017-01-11 Richard Biener <rguenther@suse.de>
33284 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33285 set range/nonnull info for PHI results. Do not set it on
33286 stmts marked for removal.
33288 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33290 * expr.c (store_field): In the bitfield case, fetch the return value
33291 from the registers before applying a single big-endian adjustment.
33292 Always do a final load for a BLKmode value not larger than a word.
33294 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33297 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33298 that we correctly handle column numbers greater than
33299 LINE_MAP_MAX_COLUMN_NUMBER.
33301 2017-01-10 Martin Sebor <msebor@redhat.com>
33303 PR middle-end/78245
33304 * gimple-ssa-sprintf.c (get_destination_size): Call
33305 {init,fini}object_sizes.
33306 * tree-object-size.c (addr_object_size): Adjust.
33307 (pass_through_call): Adjust.
33308 (pass_object_sizes::execute): Adjust.
33309 * tree-object-size.h (fini_object_sizes): Declare.
33311 2017-01-10 Martin Sebor <msebor@redhat.com>
33313 PR tree-optimization/78775
33314 * builtins.c (get_size_range): Move...
33315 * calls.c: ...to here.
33316 (alloc_max_size): Accept zero argument.
33317 (operand_signed_p): Remove.
33318 (maybe_warn_alloc_args_overflow): Call get_size_range.
33319 * calls.h (get_size_range): Declare.
33321 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33323 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33324 from TI's devices.csv file as of September 2016.
33325 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33327 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33329 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33330 * doc/invoke.texi: Likewise.
33331 * doc/md.texi: Likewise.
33332 * doc/objc.texi: Likewise.
33334 2017-01-10 Joshua Conner <joshconner@google.com>
33336 * config/arm/fuchsia-elf.h: New file.
33337 * config/fuchsia.h: New file.
33338 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33339 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33341 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33343 2016-01-10 Richard Biener <rguenther@suse.de>
33345 PR tree-optimization/79034
33346 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33347 Propagate out degenerate PHIs in the joiner.
33349 2017-01-10 Martin Liska <mliska@suse.cz>
33351 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33352 (sort_congruence_classes_by_decl_uid): Likewise.
33353 (sort_congruence_class_groups_by_decl_uid): Likewise.
33354 (sem_item_optimizer::merge_classes): Sort class, groups in these
33355 classes and members in the groups by DECL_UID of declarations.
33356 This would make merge operations stable.
33358 2017-01-10 Martin Liska <mliska@suse.cz>
33360 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33361 usage of m_classes_vec.
33362 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33363 (sem_item_optimizer::get_group_by_hash): Likewise.
33364 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33365 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33366 (sem_item_optimizer::verify_classes): Likewise.
33367 (sem_item_optimizer::process_cong_reduction): Likewise.
33368 (sem_item_optimizer::dump_cong_classes): Likewise.
33369 (sem_item_optimizer::merge_classes): Likewise.
33370 * ipa-icf.h (congruence_class_hash): Rename from
33371 congruence_class_group_hash. Remove declaration of m_classes_vec.
33373 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33375 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33376 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33377 * config.gcc: Add avx512vpopcntdqintrin.h.
33378 * config/i386/avx512vpopcntdqintrin.h: New.
33379 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33380 * config/i386/i386-builtin-types.def: Add new types.
33381 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33382 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33383 __builtin_ia32_vpopcountq_v8di_mask): New.
33384 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33385 __AVX512VPOPCNTDQ__.
33386 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33387 (PTA_AVX512VPOPCNTDQ): Define.
33388 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33389 TARGET_AVX512VPOPCNTDQ_P): Define.
33390 * config/i386/i386.opt: Add mavx512vpopcntdq.
33391 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33392 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33394 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33396 PR middle-end/77484
33397 * predict.def (PRED_CALL): Set to 67.
33399 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33401 * expr.c (store_field): In the bitfield case, if the value comes from
33402 a function call and is of an aggregate type returned in registers, do
33403 not modify the field mode; extract the value in all cases if the mode
33404 is BLKmode and the size is not larger than a word.
33406 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33409 * config/i386/cpuid.h: Fix undefined behavior.
33411 2017-01-04 Jeff Law <law@redhat.com>
33413 PR tree-optimization/79007
33414 PR tree-optimization/67955
33415 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33416 conservative for pt.null when flag_non_call_exceptions is on.
33418 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33420 PR translation/79019
33421 PR translation/79020
33422 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33423 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33424 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33426 * config/avr/avr.opt (maccumulate-args): Likewise.
33427 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33428 * common.opt (freport-bug): Likewise.
33429 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33430 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33431 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33432 translatable string.
33433 * config/i386/i386.c (function_value_32): Likewise.
33434 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33435 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33437 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33438 * common/config/msp430/msp430-common.c (msp430_handle_option):
33440 * symtab.c (symtab_node::verify_base): Likewise.
33441 * opts.c (set_debug_level): Likewise.
33442 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33443 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33444 missing whitespace to translatable strings.
33445 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33446 * config/sh/superh.h: Likewise.
33447 * config/i386/xopintrin.h: Likewise.
33448 * config/i386/znver1.md: Likewise.
33449 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33450 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33451 * double-int.h (struct double_int): Likewise.
33452 * double-int.c (div_and_round_double): Likewise.
33453 * wide-int.cc: Likewise.
33454 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33455 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33456 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33457 (crossjumps_occurred): ... this.
33458 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33461 PR tree-optimization/78899
33462 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33463 returning bool return struct loop *, NULL for failure and the new
33465 (versionable_outer_loop_p): Don't version outer loop if it has
33466 dont_vectorized bit set.
33467 (tree_if_conversion): When versioning outer loop, ensure
33468 tree_if_conversion is performed also on the inner loop of the
33469 non-vectorizable outer loop copy.
33470 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33471 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33472 prevent vectorization of it.
33473 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33474 the outer loop vectorization of the non-scalar version is attempted
33475 before vectorization of the inner loop in scalar version. If
33476 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33477 vectorization of its inner loop.
33478 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33479 has 2 inner loops, rename also on edges from bb whose single pred
33480 is outer_loop->header. Fix typo in function comment.
33482 2017-01-09 Martin Sebor <msebor@redhat.com>
33485 * asan.c (asan_emit_stack_protection): Increase local buffer size
33486 to avoid snprintf truncation warning.
33488 2017-01-09 Andrew Pinski <apinski@cavium.com>
33490 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33491 to reference thunderx2t99 for the tuning structure
33492 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33494 (thunderx2t99_extra_costs): This.
33495 * config/aarch64/aarch64-tune.md: Regenerate.
33496 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33497 (vulcan_addrcost_table): This.
33498 (vulcan_regmove_cost): Rename to ...
33499 (thunderx2t99_regmove_cost): This.
33500 (vulcan_vector_cost): Rename to ...
33501 (thunderx2t99_vector_cost): this.
33502 (vulcan_branch_cost): Rename to ...
33503 (thunderx2t99_branch_cost): This.
33504 (vulcan_tunings): Rename to ...
33505 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33506 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33508 2017-01-09 Martin Jambor <mjambor@suse.cz>
33512 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33513 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33514 (propagate_vr_accross_jump_function): Use the above function for all
33515 value range computations for pass-through jump functions and type
33516 converasion from explicit value range values.
33517 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33518 parameters from TYPE_ARG_TYPES.
33519 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33520 (ipa_write_node_info): Stream type of the actual argument.
33521 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33523 2017-01-09 Martin Liska <mliska@suse.cz>
33526 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33527 (lookup_compiler): Do not show error message with have_E.
33529 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33531 PR tree-optimization/78938
33532 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33533 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33534 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33535 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33538 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33540 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33543 2017-01-09 Richard Biener <rguenther@suse.de>
33545 PR tree-optimization/78997
33546 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33547 name condition properly.
33549 2017-01-09 Richard Biener <rguenther@suse.de>
33552 * dwarf2out.c (is_cxx): New overload with context.
33553 (is_naming_typedef_decl): Use it.
33555 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33557 * invoke.texi (Option Summary): Correct spacing in option lists
33558 and add line breaks to fix over-long lines.
33560 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33562 PR middle-end/17660
33564 * extend.texi (Common Variable Attributes): Add xref to GCC
33565 Internals manual to explain mode attribute keywords.
33567 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33570 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33571 and Preprocessor Options.
33572 (Options for Linking): Document -pthread here....
33573 (RS/6000 and PowerPC Options): ...not here.
33574 (Solaris 2 Options): ...or here.
33575 * doc/cppopts.texi: Document -pthread.
33577 2017-01-08 Martin Sebor <msebor@redhat.com>
33579 PR middle-end/77708
33580 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33581 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33582 New member functions.
33583 (format_directive): Used them.
33585 (pass_sprintf_length::handle_gimple_call): Same.
33586 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33587 to avoid truncation for any argument.
33588 (extract_affine_mul): Same.
33589 * tree.c (get_file_function_name): Same.
33591 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33593 PR middle-end/77484
33594 * predict.def (PRED_INDIR_CALL): Set to 86.
33596 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33598 PR preprocessor/54124
33599 * doc/cppopts.texi: Reformat -d subtable to list the full name
33600 of the options. Add cross-reference to the docs for the general
33601 compiler -d options.
33602 * doc/invoke.texi (Developer Options): Add cross-reference to the
33603 preprocessor-specific -d option documentation.
33605 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33607 PR preprocessor/13498
33608 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33609 redudant material, and reflect new command-line options.
33610 (System Headers): Likewise.
33612 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33614 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33615 -isystem, and -idirafter. Copy-edit.
33616 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33617 default for -ftrack-macro-expansion. Delete obsolete and
33618 badly-formatted implementation details about -fdebug-cpp output.
33619 * doc/cppwarnopts.texi: Copy-edit.
33621 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33624 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33625 that the transition from a max line width >= 1<<10 to narrower
33626 lines works correctly.
33628 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33630 * doc/options.texi (PerFunction): New.
33631 * opt-functions.awk (switch_flags): Map both Optimization and
33632 PerFunction to CL_OPTIMIZATION.
33633 * opth-gen.awk: Test for PerFunction flag along with
33635 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33636 it only when the latter is present. Skip those that don't in
33637 the hash function generator.
33638 * common.opt (fvar-tracking): Mark as PerFunction instead of
33640 (fvar-tracking-assignments): Likewise.
33641 (fvar-tracking-assignments-toggle): Likewise.
33642 (fvar-tracking-uninit): Likewise.
33644 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33646 PR translation/79018
33647 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33650 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33653 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33654 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33655 TARGET_LONG_JUMP_TABLE_OFFSETS.
33656 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33657 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33658 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33659 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33660 * config/m68k/m68k.md (tablejump expander): Likewise.
33661 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33662 TARGET_LONG_JUMP_TABLE_OFFSETS.
33663 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33664 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33666 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33667 David Holsgrove <david.holsgrove@xilinx.com>
33669 * common/config/microblaze/microblaze-common.c
33670 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33671 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33673 * config/microblaze/microblaze.c (microblaze_must_save_register)
33674 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33676 (microblaze_eh_return): New function.
33677 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33678 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33679 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33680 * config/microblaze/microblaze.md (eh_return): New pattern.
33682 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33684 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33685 GCC_DIAGNOSTIC_STRINGIFY): Define.
33687 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33689 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33691 * config/arm/arm.md (<mcrr>): New.
33693 * config/arm/arm.c (arm_arch5te): New.
33694 (arm_option_override): Set arm_arch5te.
33695 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33697 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33698 (arm_mcrr_qualifiers): ... this. New.
33699 (MRRC_QUALIFIERS): Define to...
33700 (arm_mrrc_qualifiers): ... this. New.
33701 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33703 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33704 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33705 (MRRCI, mrrc, MRRC): New.
33706 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33707 VUNSPEC_MRRC2): New.
33709 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33711 * config/arm/arm.md (<mcr>): New.
33713 * config/arm/arm.c (arm_coproc_builtin_available): Add
33714 support for mcr, mrc, mcr2 and mrc2.
33715 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33716 (arm_mcr_qualifiers): ... this. New.
33717 (MRC_QUALIFIERS): Define to ...
33718 (arm_mrc_qualifiers): ... this. New.
33719 (MCR_QUALIFIERS): Define to ...
33720 (arm_mcr_qualifiers): ... this. New.
33721 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33723 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33724 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33725 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33726 VUNSPEC_MRC2): New.
33728 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33730 * config/arm/arm.md (*ldc): New.
33734 * config/arm/arm.c (arm_coproc_builtin_available): Add
33735 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33736 (arm_coproc_ldc_stc_legitimate_address): New.
33737 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33738 'qualifier_const_pointer'.
33739 (LDC_QUALIFIERS): Define to...
33740 (arm_ldc_qualifiers): ... this. New.
33741 (STC_QUALIFIERS): Define to...
33742 (arm_stc_qualifiers): ... this. New.
33743 * config/arm/arm-protos.h
33744 (arm_coproc_ldc_stc_legitimate_address): New.
33745 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33746 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33747 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33748 stc2, stcl, stc2l): New.
33749 * config/arm/constraints.md (Uz): New.
33750 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33751 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33752 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33753 VUNSPEC_STC2L): New.
33755 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33757 * config/arm/arm.md (<cdp>): New.
33758 * config/arm/arm.c (neon_const_bounds): Rename this ...
33759 (arm_const_bounds): ... this.
33760 (arm_coproc_builtin_available): New.
33761 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33762 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33763 (CDP_QUALIFIERS): Define to...
33764 (arm_cdp_qualifiers): ... this. New.
33766 (arm_expand_builtin_args): Add case for 6 arguments.
33767 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33768 (arm_const_bounds): ... this.
33769 (arm_coproc_builtin_available): New.
33770 * config/arm/arm_acle.h (__arm_cdp): New.
33772 * config/arm/arm_acle_builtins.def (cdp): New.
33774 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33775 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33776 'arm_const_bounds'.
33777 * config/arm/types.md (coproc): New.
33778 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33779 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33780 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33781 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33783 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33785 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33786 (UBINOP_QUALIFIERS): New.
33788 (acle_builtin_data): New. Change comment.
33789 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33790 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33791 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33792 arm_acle_builtins.def.
33793 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33794 (arm_init_acle_builtins): New.
33795 (CRC32_BUILTIN): Remove.
33796 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33797 crc32cb, crc32ch and crc32cw.
33798 (arm_init_crc32_builtins): Remove.
33799 (arm_init_builtins): Use arm_init_acle_builtins rather
33800 than arm_init_crc32_builtins.
33801 (arm_expand_acle_builtin): New.
33802 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33803 * config/arm/arm_acle_builtins.def: New.
33805 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33807 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33808 (arm_builtin_datum): ... this.
33809 (arm_init_neon_builtin): Rename to ...
33810 (arm_init_builtin): ... this. Add a new parameters PREFIX
33811 and USE_SIG_IN_NAME.
33812 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33813 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33814 'arm_builtin_datum'.
33815 (arm_init_vfp_builtins): Likewise.
33816 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33817 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33818 (arm_expand_neon_args): Rename to ...
33819 (arm_expand_builtin_args): ... this. Rename builtin_arg
33820 enum values and differentiate between ARG_BUILTIN_MEMORY
33821 and ARG_BUILTIN_NEON_MEMORY.
33822 (arm_expand_neon_builtin_1): Rename to ...
33823 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33824 values, arm_expand_builtin_args and add bool parameter NEON.
33825 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33826 (arm_expand_vfp_builtin): Likewise.
33827 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33829 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33831 PR middle-end/77484
33832 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33833 * predict.c (tree_estimate_probability_bb): Reverse direction of
33834 polymorphic call predictor.
33836 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33838 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33839 (determine_pass_name_match): ...this new function and...
33840 (should_skip_pass_p): ...this new function.
33842 2017-01-06 Nathan Sidwell <nathan@acm.org>
33844 * ipa-visibility.c (function_and_variable_visibility): Reformat
33845 comments and long lines. Remove extrneous if.
33846 * symtab.c (symtab_node::make_decl_local): Fix code format.
33847 (symtab_node::set_section_for_node): Fix comment typo.
33849 2017-01-06 Martin Liska <mliska@suse.cz>
33852 * lra-constraints.c: Rename invariant to lra_invariant.
33853 * predict.c (set_even_probabilities): Initialize e to NULL.
33855 2017-01-05 Martin Sebor <msebor@redhat.com>
33857 PR tree-optimization/78910
33858 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33859 (format_integer): Correct off-by-one error in the handling
33860 of precision with negative numbers in signed conversions..
33862 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33864 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33866 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33868 PR tree-optimization/71016
33869 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33870 factor_out_conditional_conversion. Formatting fix.
33871 (factor_out_conditional_conversion): Add cond_stmt argument.
33872 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33873 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33876 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33878 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33879 read-rtl-function.o, and selftest-rtl.o.
33880 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33881 (selftest::aarch64_test_loading_full_dump): New function.
33882 (selftest::aarch64_run_selftests): New function.
33883 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33884 selftest::aarch64_run_selftests.
33885 * config/i386/i386.c
33886 (selftest::ix86_test_loading_dump_fragment_1): New function.
33887 (selftest::ix86_test_loading_call_insn): New function.
33888 (selftest::ix86_test_loading_full_dump): New function.
33889 (selftest::ix86_test_loading_unspec): New function.
33890 (selftest::ix86_run_selftests): Call the new functions.
33891 * emit-rtl.c (maybe_set_max_label_num): New function.
33892 * emit-rtl.h (maybe_set_max_label_num): New decl.
33893 * function.c (instantiate_decls): Guard call to
33894 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33895 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33897 * gensupport.c (gen_reader::gen_reader): Pass "false"
33898 for new "compact" param of rtx_reader.
33899 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33900 rather than an empty string for NULL strings.
33901 * read-md.c: Potentially include config.h rather than bconfig.h.
33902 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33903 (have_error): New global, copied from errors.c.
33904 (md_reader::read_name): Rename to...
33905 (md_reader::read_name_1): ...this, adding "out_loc" param,
33906 and converting "missing name or number" to returning false, rather
33908 (md_reader::read_name): Reimplement in terms of read_name_1.
33909 (md_reader::read_name_or_nil): New function.
33910 (md_reader::read_string): Handle "(nil)" by returning NULL.
33911 (md_reader::md_reader): Add new param "compact".
33912 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33913 (md_reader::read_file): New method.
33914 * read-md.h (md_reader::md_reader): Add new param "compact".
33915 (md_reader::read_file): New method.
33916 (md_reader::is_compact): New accessor.
33917 (md_reader::read_name): Convert return type from void to file_location.
33918 (md_reader::read_name_or_nil): New decl.
33919 (md_reader::read_name_1): New decl.
33920 (md_reader::m_compact): New field.
33921 (noop_reader::noop_reader): Pass "false" for new "compact" param
33923 (rtx_reader::rtx_reader): Add new "compact" param.
33924 (rtx_reader::read_rtx_operand): Make virtual and convert return
33925 type from void to rtx.
33926 (rtx_reader::read_until): New decl.
33927 (rtx_reader::handle_any_trailing_information): New virtual function.
33928 (rtx_reader::postprocess): New virtual function.
33929 (rtx_reader::finalize_string): New virtual function.
33930 (rtx_reader::m_in_call_function_usage): New field.
33931 (rtx_reader::m_reuse_rtx_by_id): New field.
33932 * read-rtl-function.c: New file.
33933 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33934 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33935 (selftest::verify_three_block_rtl_cfg): New decl.
33936 * read-rtl-function.h: New file.
33937 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33938 For host, include function.h, memmodel.h, and emit-rtl.h.
33939 (one_time_initialization): New function.
33940 (struct compact_insn_name): New struct.
33941 (compact_insn_names): New array.
33942 (find_code): Handle insn codes in compact dumps.
33943 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33944 (bind_subst_iter_and_attr): Likewise.
33945 (add_condition_to_string): Likewise.
33946 (add_condition_to_rtx): Likewise.
33947 (apply_attribute_uses): Likewise.
33948 (add_current_iterators): Likewise.
33949 (apply_iterators): Likewise.
33950 (initialize_iterators): Guard usage of apply_subst_iterator with
33951 #ifdef GENERATOR_FILE.
33952 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33953 (md_reader::read_mapping): Likewise.
33954 (add_define_attr_for_define_subst): Likewise.
33955 (add_define_subst_attr): Likewise.
33956 (read_subst_mapping): Likewise.
33957 (check_code_iterator): Likewise.
33958 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33960 (one_time_initialization): New function.
33961 (rtx_reader::read_until): New method.
33962 (read_flags): New function.
33963 (parse_reg_note_name): New function.
33964 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33965 Handle reuse_rtx ids.
33966 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33967 Add parsing support for RTL dumps, mirroring the special-cases in
33968 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33969 values, and calling handle_any_trailing_information.
33970 (rtx_reader::read_rtx_operand): Convert return type from void
33971 to rtx, returning return_rtx. Handle case 'e'. Call
33972 finalize_string on XSTR and XTMPL fields.
33973 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33974 "(nil)" values were omitted. Call the postprocess vfunc on the
33976 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33977 class ctor. Initialize m_in_call_function_usage. Call
33978 one_time_initialization.
33979 * rtl-tests.c (selftest::test_uncond_jump): Call
33980 set_new_first_and_last_insn.
33981 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33982 * selftest-rtl.c: New file.
33983 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33984 (selftest::get_insn_by_uid): New decl.
33985 * selftest-run-tests.c (selftest::run_tests): Call
33986 read_rtl_function_c_tests.
33987 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33988 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33991 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33993 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33994 operands in a special way. Assert that pos+len <= mode precision.
33996 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33998 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33999 3 argument Alias with unlimited for the negative form.
34000 (fno-vect-cost-model): Removed.
34002 2017-01-05 Martin Liska <mliska@suse.cz>
34004 * hsa-gen.c (gen_hsa_divmod): New function.
34005 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34007 2017-01-05 Martin Liska <mliska@suse.cz>
34010 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34013 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34015 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34016 small constant length operands.
34018 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34020 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34021 between loop iterations.
34023 2017-01-05 Martin Liska <mliska@suse.cz>
34026 * gimplify.c (gimplify_decl_expr): Compare to
34027 asan_poisoned_variables instread of checking flags.
34028 (gimplify_target_expr): Likewise.
34029 (gimplify_expr): Likewise.
34030 (gimplify_function_tree): Conditionally initialize
34031 asan_poisoned_variables.
34033 2017-01-04 Jeff Law <law@redhat.com>
34035 PR tree-optimizatin/78812
34036 * rtl.h (contains_mem_rtx_p): Prototype.
34037 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34038 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34039 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34040 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34041 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34043 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34045 * input.c (assert_char_at_range): Default-initialize actual_range.
34047 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34049 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34050 to match the caller.
34052 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34054 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34055 insns after final jump in test to emit dummy move.
34057 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34059 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34060 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34062 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34064 * multiple_target.c (create_dispatcher_calls): Init e_next.
34065 * tree-ssa-loop-split.c (split_loop): Init border.
34066 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34069 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34074 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34075 (altivec_register_operand): Do not return true if the operand
34076 contains a SUBREG mixing SImode and SFmode.
34077 (vsx_register_operand): Likewise.
34078 (vsx_reg_sfsubreg_ok): New predicate.
34079 (vfloat_operand): Do not return true if the operand contains a
34080 SUBREG mixing SImode and SFmode.
34081 (vint_operand): Likewise.
34082 (vlogical_operand): Likewise.
34083 (gpc_reg_operand): Likewise.
34084 (int_reg_operand): Likewise.
34085 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34086 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34087 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34089 (rs6000_emit_move_si_sf_subreg): New helper function.
34090 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34091 fixup SUBREGs involving SImode and SFmode.
34092 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34093 numbers for the new peephole2 optimization.
34094 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34095 the GLIBC math library that do AND/IOR/XOR operations on single
34096 precision floating point.
34097 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34098 target macros to say whether we need to avoid SUBREGs mixing
34100 (TARGET_ALLOW_SF_SUBREG): Likewise.
34101 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34102 (UNSPEC_SI_FROM_SF): Likewise.
34103 (iorxor): Change spacing.
34104 (and_ior_xor): New iterator for AND, IOR, and XOR.
34105 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34106 (movdi_from_sf_zero_ext): Likewise.
34107 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34108 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34109 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34110 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34111 (fms<mode>4): Likewise.
34112 (fnma<mode>4): Likewise.
34113 (fnms<mode>4): Likewise.
34114 (nfma<mode>4): Likewise.
34115 (nfms<mode>4): Likewise.
34117 2017-01-04 Marek Polacek <polacek@redhat.com>
34120 * doc/invoke.texi: Document -Wpointer-compare.
34122 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34124 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34127 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34128 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34131 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34134 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34135 documentation of the powerpc_popcntb_ok attribute.
34136 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34137 code to issue warning messages if a requested CPU configuration is
34138 not supported by the binary (assembler and loader) toolchain.
34139 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34140 made to define a built-in function that has been disabled.
34141 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34142 made to define a built-in function that has been disabled.
34143 (altivec_init_builtins): Add comment explaining why definition
34144 of the DST built-in functions is not preceded by an assertion
34145 check. Add assertions to prevent ICE if attempts are made to
34146 define an altivec predicate or an abs* built-in function that has
34148 (htm_init_builtins): Add comment explaining why definition of the
34149 htm built-in functions is not preceded by an assertion check.
34151 2017-01-04 Jeff Law <law@redhat.com>
34153 PR tree-optimizatin/67955
34154 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34155 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34156 the points-to solution does not include pt_null. Use DECL_PT_UID
34159 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34161 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34162 Use gen_int_mode instead of gen_lopwart for const_int operands.
34164 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34166 PR tree-optimization/71563
34167 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34168 out of range value - has low bits known to be zero.
34170 2017-01-04 Alan Modra <amodra@gmail.com>
34172 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34173 * configure: Regenerate.
34174 * config.in: Regenerate.
34176 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34179 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34180 a substring of the message, but strcmp with the whole message. Ifdef
34181 ENABLE_NLS, translate the message first using dgettext.
34183 2017-01-03 Jeff Law <law@redhat.com>
34185 PR tree-optimizatin/78856
34186 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34187 (mark_threaded_blocks): Remove code to truncate thread paths that
34188 cross multiple loop headers. Instead invalidate the cached loop
34189 iteration information and handle case of a thread path walking
34190 into an irreducible region.
34192 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34195 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34196 assertions. Add support for doing the signbit if the IEEE 128-bit
34197 floating point value is in a GPR.
34198 * config/rs6000/rs6000.md (Fsignbit): Delete.
34199 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34200 Update the length attribute if the value is in a GPR.
34201 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34202 the sign or zero extension instruction, since the value is always 0/1.
34203 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34206 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34207 extracting SImode to a GPR register so that we can generate a
34208 store, limit the vector to be in a traditional Altivec register
34209 for the vextuwrx instruction.
34211 2017-01-03 Ian Lance Taylor <iant@google.com>
34213 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34215 2017-01-03 Martin Sebor <msebor@redhat.com>
34217 PR tree-optimization/78696
34218 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34219 precision. Use MPFR for %f for greater fidelity. Correct handling
34221 (pass_sprintf_length::compute_format_length): Set width and precision
34222 specified by asrerisk to void_node for vararg functions.
34223 (try_substitute_return_value): Adjust dump output.
34225 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34227 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34229 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34231 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34232 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34233 -mlra/-mno-lra was passed to the compiler.
34235 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34237 PR rtl-optimization/65618
34238 * emit-rtl.c (try_split): Move initialization of "before" and
34239 "after" to just before the call to emit_insn_after_setloc.
34241 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34243 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34245 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34247 * dwarf2out.c (gen_enumeration_type_die): When
34248 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34250 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34252 PR tree-optimization/78965
34253 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34254 Change first argument from const call_info & to call_info &. For %n
34255 set info.nowrite to false.
34257 PR middle-end/78901
34258 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34259 possibly throwing calls.
34261 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34262 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34263 and fns handling, rather than in a separate case SSA_NAME.
34265 2017-01-02 Jeff Law <law@redhat.com>
34267 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34268 fixes for first_period and second_period variables.
34270 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34273 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34274 (*insvqi_1): New insn pattern.
34275 (*insvqi_1_mem_rex64): Ditto.
34276 (*insvqi_2): Ditto.
34277 (*insvqi_3): Rename from *insvqi.
34279 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34281 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34283 * doc/cfg.texi (Edges): Remove reference to Java.
34284 (Maintaining the CFG): Ditto.
34286 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34288 PR middle-end/77674
34289 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34290 transparent aliases.
34292 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34294 PR middle-end/77484
34295 * predict.def (PRED_CALL): Update hitrate.
34296 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34297 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34298 into direct/indirect/polymorphic variants.
34300 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34302 Update copyright years.
34304 * gcc.c (process_command): Update copyright notice dates.
34305 * gcov-dump.c (print_version): Ditto.
34306 * gcov.c (print_version): Ditto.
34307 * gcov-tool.c (print_version): Ditto.
34308 * gengtype.c (create_file): Ditto.
34309 * doc/cpp.texi: Bump @copying's copyright year.
34310 * doc/cppinternals.texi: Ditto.
34311 * doc/gcc.texi: Ditto.
34312 * doc/gccint.texi: Ditto.
34313 * doc/gcov.texi: Ditto.
34314 * doc/install.texi: Ditto.
34315 * doc/invoke.texi: Ditto.
34317 Copyright (C) 2017 Free Software Foundation, Inc.
34319 Copying and distribution of this file, with or without modification,
34320 are permitted in any medium without royalty provided the copyright
34321 notice and this notice are preserved.