1 2017-11-09 Jeff Law <law@redhat.com>
3 * vr-values.h: New file with vr_values class.
4 * tree-vrp.c: Include vr-values.h
5 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6 data objects into the vr_values class.
7 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8 (get_value_range): Make it a member function within vr_values class.
9 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
10 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
11 (op_with_constant_singleton_value_range): Likewise.
12 (extract_range_for_var_from_comparison_expr): Likewise.
13 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
14 (extract_range_from_binary_expr): Likewise.
15 (extract_range_from_unary_expr): Likewise.
16 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
17 (check_for_binary_op_overflow, extract_range_basic): Likewise.
18 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
19 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
20 (compare_name_with_value, compare_names): Likewise.
21 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
22 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
23 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
24 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
25 (extract_range_from_phi_node): Likewise.
26 (simplify_truth_ops_using_ranges): Likewise.
27 (simplify_div_or_mod_using_ranges): Likewise.
28 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
29 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
30 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
31 (simplify_float_conversion_using_ranges): Likewise.
32 (simplify_internal_call_using_ranges): Likewise.
33 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
34 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
35 poitner via x_vr_values for calls into gimple folder.
36 (vrp_initialize_lattice): Make this the vr_values ctor.
37 (vrp_free_lattice): Make this the vr_values dtor.
38 (set_vr_value): New function.
39 (class vrp_prop): Add vr_values data member. Add various member
40 functions as well as member functions that delegate to vr_values.
41 (check_array_ref): Make a member function within vrp_prop class.
42 (search_for_addr_array, vrp_initialize): Likewise.
43 (vrp_finalize): Likewise. Revamp to avoid direct access to
44 vr_value, values_propagated, etc.
45 (check_array_bounds): Extract vrp_prop class instance pointer from
46 walk info structure. Use it to call member functions.
47 (check_all_array_refs): Make a member function within vrp_prop class.
48 Smuggle class instance pointer via walk info structure.
49 (x_vr_values): New local static.
50 (vrp_valueize): Use x_vr_values to get class instance.
51 (vr_valueize_1): Likewise.
52 (class vrp_folder): Add vr_values data member. Add various member
53 functions as well as member functions that delegate to vr_values.
54 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
55 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
56 class instance from vr_values. Use it to call member functions.
57 (vrp_dom_walker): Add vr_values data member.
58 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
59 instance via x_vr_values.
60 (identify_jump_threads): Accept vr_values as argument. Store
61 it into the walker structure.
62 (evrp_dom_walker): Add vr_values class data member. Add various
64 (evrp_dom_walker::try_find_new_range): Use vr_values data
65 member to access the memory allocator.
66 (evrp_dom_walker::before_dom_children): Store vr_values class
67 instance into the vrp_folder class.
68 (evrp_dom_walker::push_value_range): Rework to avoid direct
69 access to num_vr_values and vr_value.
70 (evrp_dom_walker::pop_value_range): Likewise.
71 (execute_early_vrp): Remove call to vrp_initialize_lattice.
72 Use vr_values to get to dump_all_value_ranges member function.
73 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
74 and simplify_cond_using_ranges_2 via vrp_prop class instance.
75 Pass vr_values class instance down to identify_jump_threads.
76 Remove call to vrp_free_lattice.
77 (debug_all_value_ranges): Remove.
79 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
80 (vrp_folder): Likewise.
82 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
83 Get it from the existing bitmap instead.
84 (vrp_intersect_ranges_1): Likewise.
86 2017-11-09 Jakub Jelinek <jakub@redhat.com>
88 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
90 (store_immediate_info::store_immediate_info): Add bitnotp argument,
92 (imm_store_chain_info::coalesce_immediate_stores): Break group
93 if bit_not_p is different.
94 (count_multiple_uses, split_group,
95 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
96 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
97 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
98 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
99 to store_immediate_info ctor.
101 2017-11-09 Jim Wilson <jimw@sifive.com>
103 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
104 (scan_prog_file): Likewise.
106 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
108 * bb-reorder.c (max_entry_frequency): Remove.
109 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
110 connect_traces, push_to_next_round_p): Remove prototypes.
111 (find_traces_1_round): Use counts only.
112 (push_to_next_round_p): Likewise.
113 (find_traces): Likewise.
114 (rotate_loop): Likewise.
115 (find_traces_1_round): Likewise.
116 (connect_traces): Likewise.
117 (edge_order): Likewise.
119 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
121 * config/arm/arm.c (output_return_instruction): Add comments to
122 indicate requirement for cmse_nonsecure_entry return to account
123 for the size of clearing instruction output here.
124 (thumb_exit): Likewise.
125 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
126 return in hardfloat mode.
128 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
130 * config/rs6000/rs6000.c (machine_function): Add a bool,
131 "toc_is_wrapped_separately".
132 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
133 if it wasn't explicitly set or unset, we are optimizing for speed, and
134 doing separate shrink-wrapping.
135 (rs6000_get_separate_components): Enable the TOC component if
136 saving the TOC register in the prologue.
137 (rs6000_components_for_bb): Handle the TOC component.
138 (rs6000_emit_prologue_components): Store the TOC register where needed.
139 (rs6000_set_handled_components): Mark TOC as handled, if handled.
140 (rs6000_emit_prologue): Don't save the TOC if that is already done.
142 2017-11-09 Martin Jambor <mjambor@suse.cz>
144 * ipa-param-manipulation.c: New file.
145 * ipa-param-manipulation.h: Likewise.
146 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
147 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
148 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
149 (ipa_parm_adjustment): Likewise.
150 (ipa_parm_adjustment_vec): Likewise.
151 (ipa_get_vector_of_formal_parms): Moved declaration to
152 ipa-param-manipulation.h.
153 (ipa_get_vector_of_formal_parm_types): Likewise.
154 (ipa_modify_formal_parameters): Likewise.
155 (ipa_modify_call_arguments): Likewise.
156 (ipa_combine_adjustments): Likewise.
157 (ipa_dump_param_adjustments): Likewise.
158 (ipa_modify_expr): Likewise.
159 (ipa_get_adjustment_candidate): Likewise.
160 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
161 ipa-param-manipulation.c.
162 (ipa_get_vector_of_formal_parm_types): Likewise.
163 (ipa_modify_formal_parameters): Likewise.
164 (ipa_modify_call_arguments): Likewise.
165 (ipa_modify_expr): Likewise.
166 (get_ssa_base_param): Likewise.
167 (ipa_get_adjustment_candidate): Likewise.
168 (index_in_adjustments_multiple_times_p): Likewise.
169 (ipa_combine_adjustments): Likewise.
170 (ipa_dump_param_adjustments): Likewise.
171 * tree-sra.c: Also include ipa-param-manipulation.h
172 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
175 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
176 Alan Hayward <alan.hayward@arm.com>
177 David Sherwood <david.sherwood@arm.com>
179 * doc/sourcebuild.texi (vect_masked_store): Document.
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_align_stack_vars): 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_variable_length): 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_unaligned_possible): 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_element_align_preferred): 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_perm_short, vect_perm_byte): Document
210 previously undocumented selectors.
211 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
213 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
215 * doc/rtl.texi (const_vector): Say that elements can be
217 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
218 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
219 (gen_vec_duplicate): Use it instead of CONSTANT_P.
220 * optabs.c (expand_vector_broadcast): Likewise.
222 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
223 Alan Hayward <alan.hayward@arm.com>
224 David Sherwood <david.sherwood@arm.com>
226 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
227 scaled index even if the unscaled address was invalid.
228 Don't increase the complexity of using a scale in that case.
230 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
231 Alan Hayward <alan.hayward@arm.com>
232 David Sherwood <david.sherwood@arm.com>
234 * doc/rtl.texi: Rewrite the subreg rules so that they partition
235 the inner register into REGMODE_NATURAL_SIZE bytes rather than
236 UNITS_PER_WORD bytes.
237 * emit-rtl.c (validate_subreg): Divide subregs into blocks
238 based on REGMODE_NATURAL_SIZE of the inner mode.
239 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
240 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
241 * expmed.c (lowpart_bit_field_p): Divide the value up into
242 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
243 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
244 whether something is likely to occupy more than one register.
246 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
249 * ipa-inline-transform.c (update_noncloned_frequencies): Use
250 profile_count::adjust_for_ipa_scaling.
251 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
252 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
254 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
256 2017-11-09 Jakub Jelinek <jakub@redhat.com>
258 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
259 (split_group): Add total_orig and total_new arguments, estimate the
260 number of statements related to the store group without store merging
261 and with store merging.
262 (imm_store_chain_info::output_merged_store): Adjust split_group
263 callers, punt if estimated number of statements with store merging
264 is not smaller than estimated number of statements without it.
266 (handled_load): Remove has_single_use checks.
267 (pass_store_merging::process_store): Likewise.
269 2017-11-09 Richard Biener <rguenther@suse.de>
271 PR tree-optimization/82902
272 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
274 2017-11-09 Martin Liska <mliska@suse.cz>
277 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
280 2017-11-09 Martin Liska <mliska@suse.cz>
282 PR tree-optimization/82669
283 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
285 2017-11-09 Martin Liska <mliska@suse.cz>
287 PR gcov-profile/48463
288 * coverage.c (coverage_begin_function): Output also end locus
289 of a function and information whether the function is
291 * gcov-dump.c (tag_function): Parse and print the information.
292 * gcov.c (INCLUDE_MAP): Add include.
293 (INCLUDE_SET): Likewise.
294 (struct line_info): Move earlier in the source file because
295 of vector<line_info> in function_info structure.
296 (line_info::line_info): Likewise.
297 (line_info::has_block): Likewise.
298 (struct source_info): Add new member index.
299 (source_info::get_functions_at_location): New function.
300 (function_info::group_line_p): New function.
301 (output_intermediate_line): New function.
302 (output_intermediate_file): Use the mentioned function.
303 (struct function_start): New.
304 (struct function_start_pair_hash): Likewise.
305 (process_file): Add code that identifies group functions.
306 Assign lines either to global or function scope.
307 (generate_results): Skip artificial functions.
308 (find_source): Assign index for each source file.
309 (read_graph_file): Read new flag artificial and end_line.
310 (add_line_counts): Assign it either to global of function scope.
311 (accumulate_line_counts): Isolate core of the function to
312 accumulate_line_info and call it for both function and global
314 (accumulate_line_info): New function.
315 (output_line_beginning): Fix GNU coding style.
316 (print_source_line): New function.
317 (output_line_details): Likewise.
318 (output_function_details): Likewise.
319 (output_lines): Iterate both source (global) scope and function
321 (struct function_line_start_cmp): New class.
322 * doc/gcov.texi: Reflect changes in documentation.
324 2017-11-09 Jakub Jelinek <jakub@redhat.com>
327 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
328 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
329 and similarly for not instead of neg.
331 2017-11-08 Andi Kleen <ak@linux.intel.com>
333 * config/i386/i386.opt: Add -mforce-indirect-call.
334 * config/i386/predicates.md: Check for flag_force_indirect_call.
335 * doc/invoke.texi: Document -mforce-indirect-call
337 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
339 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
341 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
342 config/riscv/riscv.c (predict.h): New include.
343 (riscv_slow_unaligned_access_p): No longer static.
344 (riscv_block_move_straight): Add require.
345 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
347 2017-11-08 Jakub Jelinek <jakub@redhat.com>
350 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
351 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
352 nonimmediate_operand predicate for operand 1 instead of
355 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
357 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
359 * config/aarch64/constraints.md (Uml): New constraint.
360 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
363 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
365 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
366 of two vec_duplicates into a vec_concat.
368 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
370 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
371 Simplify vec_merge of vec_duplicate and vec_concat.
372 * config/aarch64/constraints.md (Utq): New constraint.
373 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
376 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
378 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
379 Simplify vec_merge of vec_duplicate and const_vector.
380 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
382 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
383 mode iterator. Update predicate on operand 1 to
384 handle non-const_vec constants. Delete constraints.
385 (*aarch64_combinez_be<mode>): Likewise for operand 2.
387 2017-11-08 Jakub Jelinek <jakub@redhat.com>
389 PR tree-optimization/78821
390 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
392 (store_operand_info::store_operand_info): Initialize it to false.
393 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
394 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
395 store in the group, and if chain_info is non-NULL, to ignore altogether
397 (compatible_load_p): Fail if bit_not_p does not match.
398 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
399 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
400 (pass_store_merging::process_store): Adjust
401 terminate_all_aliasing_chains calls to pass NULL in all current spots,
402 call terminate_all_aliasing_chains newly when adding a store into
403 a chain with non-NULL chain_info.
405 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
407 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
409 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
411 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
413 (aarch64_layout_frame): Initialise emit_frame_chain.
414 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
415 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
417 2017-11-08 Martin Liska <mliska@suse.cz>
419 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
420 of gimple_call_internal_p.
422 2017-11-07 Tom de Vries <tom@codesourcery.com>
424 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
426 2017-11-07 Tom de Vries <tom@codesourcery.com>
428 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
431 2017-11-08 Martin Liska <mliska@suse.cz>
434 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
436 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
438 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
440 2017-11-07 Jakub Jelinek <jakub@redhat.com>
443 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
444 (*cmp<mode>_ccz_1): New insn with $k alternative.
447 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
448 RTX_COMM_COMPARE as commutative as well.
449 (ix86_binary_operator_ok): Formatting fix.
450 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
451 *<code><mode>3<mask_name><round_saeonly_name>,
452 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
453 *<s>mul<mode>3_highpart<mask_name>,
454 *vec_widen_umult_even_v16si<mask_name>,
455 *vec_widen_umult_even_v8si<mask_name>,
456 *vec_widen_umult_even_v4si<mask_name>,
457 *vec_widen_smult_even_v16si<mask_name>,
458 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
459 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
460 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
461 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
462 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
463 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
464 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
465 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
466 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
467 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
468 ix86_binary_operator_ok. Formatting fixes.
469 (*<plusminus_insn><mode>3<mask_name><round_name>,
470 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
473 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
475 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
476 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
477 eq and ne if TARGET_ISEL.
478 (cmp): New code_iterator.
479 (UNS, UNSU_, UNSIK): New code_attrs.
480 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
481 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
483 ("eq<mode>3"): ... this.
484 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
486 ("ne<mode>3"): ... this.
488 2017-11-07 Julia Koval <julia.koval@intel.com>
491 * common/config/i386/i386-common.c
492 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
493 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
494 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
495 * config/i386/i386.opt: Ditto.
496 * config/i386/i386.c (ix86_target_string): Ditto.
497 (ix86_option_override_internal): Ditto.
498 (ix86_init_mpx_builtins): Move MPX to args2.
499 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
500 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
501 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
502 __builtin_ia32_vgf2p8affineinvqb_v32qi,
503 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
504 __builtin_ia32_vgf2p8affineinvqb_v16qi,
505 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
507 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
510 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
511 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
512 (zero-extendsidi peephole2): Remove peephole.
514 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
517 * stor-layout.c: Include attribs.h.
518 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
519 explicit lookup of "aligned" attribute.
521 2017-11-07 Andrew Waterman <andrew@sifive.com>
523 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
524 (riscv_expand_block_move): Likewise.
525 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
527 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
528 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
529 * config/riscv/riscv.c (riscv_block_move_straight): New function.
530 (riscv_adjust_block_mem): Likewise.
531 (riscv_block_move_loop): Likewise.
532 (riscv_expand_block_move): Likewise.
533 gcc/config/riscv/riscv.md (movmemsi): New pattern.
535 2017-11-07 Michael Clark <michaeljclark@mac.com>
537 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
538 (MUSL_DYNAMIC_LINKER): Likewise.
540 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
542 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
543 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
546 2017-11-07 Richard Biener <rguenther@suse.de>
548 * match.pd: Fix build.
550 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
551 Jackson Woodruff <jackson.woodruff@arm.com>
553 PR tree-optimization/71026
554 * match.pd: Canonicalize negate in division.
556 2017-11-07 Sudakshina Das <sudi.das@arm.com>
559 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
561 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
563 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
564 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
566 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
568 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
569 non-scalar integral types.
570 * match.pd (negate_expr_p): Handle MINUS_EXPR.
571 (-(A-B), -(~A)): New transformations.
573 2017-11-07 Tom de Vries <tom@codesourcery.com>
575 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
576 semicolon after "do {} while (0)".
577 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
578 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
579 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
580 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
581 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
583 2017-11-07 Tom de Vries <tom@codesourcery.com>
585 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
586 after "do {} while (0)".
587 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
588 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
589 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
590 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
591 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
593 2017-11-07 Tom de Vries <tom@codesourcery.com>
596 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
598 (arm_rtx_costs_internal): Add missing semicolon after
599 HANDLE_NARROW_SHIFT_ARITH call.
601 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
603 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
604 disable isel if it was not set explicitly.
606 2017-11-06 James Bowman <james.bowman@ftdichip.com>
608 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
610 (add_type_attribute) likewise.
612 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
614 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
615 of struct ix86_frame.
616 (ix86_initial_elimination_offset): Likewise.
617 (ix86_expand_split_stack_prologue): Likewise.
619 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
621 * tree-vrp.h (enum value_range_type): Update stale comment.
623 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
624 Alan Hayward <alan.hayward@arm.com>
625 David Sherwood <david.sherwood@arm.com>
627 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
628 (aarch64_expand_vec_perm_const): Take the number of units too.
629 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
630 (aarch64_expand_vec_perm_const): Likewise.
631 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
632 (vec_perm<mode>): Update accordingly.
634 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
635 Alan Hayward <alan.hayward@arm.com>
636 David Sherwood <david.sherwood@arm.com>
638 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
639 Take the number of units too.
640 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
641 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
642 but check for a vector mode before rather than after the call.
643 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
644 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
645 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
646 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
647 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
648 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
649 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
650 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
651 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
652 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
653 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
654 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
655 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
656 (aarch64_sqdmull2_n<mode>): Update accordingly.
658 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
659 Alan Hayward <alan.hayward@arm.com>
660 David Sherwood <david.sherwood@arm.com>
662 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
663 the number of units too.
664 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
665 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
666 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
667 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
668 (vec_store_lanesxi<mode>): Update accordingly.
670 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
671 Alan Hayward <alan.hayward@arm.com>
672 David Sherwood <david.sherwood@arm.com>
674 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
675 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
676 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
677 of units rather than the mode.
678 * config/aarch64/iterators.md (nunits): New mode attribute.
679 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
680 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
681 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
682 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
683 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
684 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
685 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
686 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
687 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
688 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
689 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
690 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
691 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
692 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
693 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
694 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
695 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
696 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
697 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
698 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
699 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
700 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
701 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
702 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
703 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
704 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
705 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
706 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
707 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
708 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
709 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
710 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
711 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
712 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
713 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
714 (aarch64_simd_vec_setv2di): Likewise.
716 2017-11-06 Carl Love <cel@us.ibm.com>
718 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
720 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
721 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
723 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
725 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
726 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
727 Add power 8 macro expansions.
728 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
729 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
730 power 8 instructions. (VSX_XXBR): Add iterator.
732 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
734 * config/arm/arm.md (predicable_short_it): Change default to "no",
735 improve documentation, remove uses that are identical to the default.
736 (enabled_for_depr_it): Rename to enabled_for_short_it.
737 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
738 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
739 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
740 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
741 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
743 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
746 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
747 float128 helper macros, which are no longer used after deleting
748 the old 'q' built-in functions, and moving the round to odd
749 built-in functions to being special built-in functions.
750 (BU_FLOAT128_2): Likewise.
751 (BU_FLOAT128_1_HW): Likewise.
752 (BU_FLOAT128_2_HW): Likewise.
753 (BU_FLOAT128_3_HW): Likewise.
754 (FABSQ): Delete old 'q' built-in functions.
755 (COPYSIGNQ): Likewise.
756 (SQRTF128_ODD): Move round to odd built-in functions to be
757 special built-in functions, so that we can handle
758 -mabi=ieeelongdouble.
759 (TRUNCF128_ODD): Likewise.
760 (ADDF128_ODD): Likewise.
761 (SUBF128_ODD): Likewise.
762 (MULF128_ODD): Likewise.
763 (DIVF128_ODD): Likewise.
764 (FMAF128_ODD): Likewise.
765 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
766 built-in names to 'f128'.
767 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
768 old 'q' built-in functions, as the machine independent code for
769 'f128' built-in functions handles this.
770 (rs6000_expand_builtin): Add expansion for float128 round to odd
771 functions, keying off on -mabi=ieeelongdouble of whether to use
772 the KFmode or TFmode variant.
773 (rs6000_init_builtins): Initialize the _Float128 round to odd
775 * doc/extend.texi (PowerPC Built-in Functions): Document the old
776 _Float128 'q' built-in functions are now mapped into the new
777 'f128' built-in functions.
779 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
781 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
782 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
784 2017-11-06 David Malcolm <dmalcolm@redhat.com>
787 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
788 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
789 * toplev.c: Include "ipa-fnsummary.h".
790 (toplev::finalize): Call ipa_fnsummary_c_finalize.
792 2017-11-06 Jakub Jelinek <jakub@redhat.com>
794 PR tree-optimization/82838
795 * gimple-ssa-store-merging.c
796 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
797 on a separate gimple_seq which is then appended to seq.
799 2017-11-06 Jeff Law <law@redhat.com>
802 * config/i386/i386.c (PROBE_INTERVAL): Remove.
803 (get_probe_interval): New functions.
804 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
805 (ix86_adjust_stack_and_probe): Likewise.
806 (output_adjust_stack_and_probe): Likewise.
807 (ix86_emit_probe_stack_range): Likewise.
808 (ix86_expand_prologue): Likewise.
810 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
812 PR tree-optimization/82816
813 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
814 if the modes of the two types are the same.
815 (convert_plusminus_to_widen): Likewise.
817 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
819 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
821 (usadv16qi): New define_expand.
822 (usadv8hi): New define_expand.
824 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
827 * ipa-inline-transform.c (update_noncloned_frequencies): Always
829 (inline_transform): Likewise.
830 * predict.c (counts_to_freqs): Remove useless conditional.
831 * profile-count.h (profile_count::apply_scale): Move sanity check.
832 * tree-inline.c (copy_bb): Always scale.
833 (copy_cfg_body): Likewise.
835 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
838 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
840 (*cmp_ite0): Add enabled_for_depr_it attribute.
841 (*cmp_ite1): Likewise.
843 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
845 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
848 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
850 * tree-vrp.c (vrp_int_const_binop): Return true on success and
851 return the value by pointer.
852 (extract_range_from_multiplicative_op_1): Update accordingly.
853 Return as soon as an operation fails.
855 2017-11-05 Tom de Vries <tom@codesourcery.com>
858 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
859 (DEF_SANITIZER_BUILTIN): ... here.
860 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
861 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
863 2017-11-05 Tom de Vries <tom@codesourcery.com>
866 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
868 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
869 ASM_OUTPUT_BEFORE_CASE_LABEL call.
870 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
872 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
873 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
874 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
876 2017-11-05 Tom de Vries <tom@codesourcery.com>
879 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
882 2017-11-04 Michael Clark <michaeljclark@mac.com>
884 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
885 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
887 (*addsi3_extended): Likewise.
888 (*addsi3_extended2): Likewise.
889 (<optab>si3): Likewise.
890 (<optab>di3): Likewise.
891 (<optab><mode>3): Likewise.
892 (<*optabe>si3_internal): Likewise.
893 (zero_extendqi<SUPERQI:mode>2): Likewise.
894 (*add<mode>hi3): Likewise.
895 (*xor<mode>hi3): Likewise.
896 (<optab>di3): Likewise.
897 (*<optab>si3_extend): Likewise.
898 (*sge<u>_<X:mode><GPR:mode>): Likewise.
899 (*slt<u>_<X:mode><GPR:mode>): Likewise.
900 (*sle<u>_<X:mode><GPR:mode>): Likewise.
902 2017-11-04 Andrew Waterman <andrew@sifive.com>
904 * config/riscv/riscv.c (riscv_option_override): Conditionally set
905 TARGET_STRICT_ALIGN based upon -mtune argument.
907 2017-11-04 Andrew Waterman <andrew@sifive.com>
909 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
911 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
913 config/i386/i386.c (choose_basereg): Use optional scratch
914 register and add assertion.
915 (x86_emit_outlined_ms2sysv_save): Use scratch register when
916 needed, and don't allocate stack.
917 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
918 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
919 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
921 2017-11-03 Jeff Law <law@redhat.com>
923 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
924 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
925 to probe at the start of a noreturn function.
927 2017-11-03 Jakub Jelinek <jakub@redhat.com>
929 PR tree-optimization/78821
930 * gimple-ssa-store-merging.c: Update the file comment.
931 (MAX_STORE_ALIAS_CHECKS): Define.
932 (struct store_operand_info): New type.
933 (store_operand_info::store_operand_info): New constructor.
934 (struct store_immediate_info): Add rhs_code and ops data members.
935 (store_immediate_info::store_immediate_info): Add rhscode, op0r
936 and op1r arguments to the ctor, initialize corresponding data members.
937 (struct merged_store_group): Add load_align_base and load_align
939 (merged_store_group::merged_store_group): Initialize them.
940 (merged_store_group::do_merge): Update them.
941 (merged_store_group::apply_stores): Pick the constant for
942 encode_tree_to_bitpos from one of the two operands, or skip
943 encode_tree_to_bitpos if neither operand is a constant.
944 (class pass_store_merging): Add process_store method decl. Remove
945 bool argument from terminate_all_aliasing_chains method decl.
946 (pass_store_merging::terminate_all_aliasing_chains): Remove
947 var_offset_p argument and corresponding handling.
948 (stmts_may_clobber_ref_p): New function.
949 (compatible_load_p): New function.
950 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
951 if there is overlap and rhs_code is not INTEGER_CST. For
952 non-overlapping stores terminate group if rhs is not mergeable.
953 (get_alias_type_for_stmts): Change first argument from
954 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
955 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
956 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
958 (get_location_for_stmts): Change first argument from
959 auto_vec<gimple *> & to vec<gimple *> &.
960 (struct split_store): Remove orig_stmts data member, add orig_stores.
961 (split_store::split_store): Create orig_stores rather than orig_stmts.
962 (find_constituent_stmts): Renamed to ...
963 (find_constituent_stores): ... this. Change second argument from
964 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
965 to info structures rather than the statements.
966 (split_group): Rename ALLOW_UNALIGNED argument to
967 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
968 it. Adjust find_constituent_stores caller.
969 (imm_store_chain_info::output_merged_store): Handle rhs_code other
970 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
971 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
972 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
973 (mem_valid_for_store_merging): New function.
974 (handled_load): New function.
975 (pass_store_merging::process_store): New method.
976 (pass_store_merging::execute): Use process_store method. Adjust
977 terminate_all_aliasing_chains caller.
979 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
981 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
982 Return true for more constants, symbols and label references.
983 (aarch64_valid_floating_const): Remove unused function.
985 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
988 * config/aarch64/aarch64.c (aarch64_layout_frame):
989 Undo forcing of LR at bottom of frame.
991 2017-11-03 Jeff Law <law@redhat.com>
994 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
995 for int_registers_saved.
997 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
998 extracted from tree-ssa-dom.c.
999 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1000 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1001 (record_equivalences_from_incoming_edge): Add additional argument
1002 to single_pred_edge_ignoring_loop_edges call.
1003 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1004 (uncprop_dom_walker::before_dom_children): Add additional argument
1005 to single_pred_edge_ignoring_loop_edges call.
1006 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1007 single_pred_edge_ignoring_loop_edges rather than open coding.
1008 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1010 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1012 * match.pd (-(-A)): Rewrite.
1014 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1016 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1017 (rs6000_emit_int_cmove): New declaration.
1018 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1019 (rs6000_emit_sISEL): Delete.
1020 (rs6000_emit_int_cmove): Make non-static.
1021 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1022 instead of rs6000_emit_sISEL.
1024 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1026 * asan.c (create_cond_insert_point): Maintain profile.
1027 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1029 * basic-block.h (struct basic_block_def): Remove frequency.
1030 (EDGE_FREQUENCY): Use to_frequency
1031 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1033 (find_traces): Update to use to_frequency.
1034 (find_traces_1_round): Likewise; use only IPA counts.
1035 (bb_to_key): Likewise.
1036 (connect_traces): Use IPA counts only.
1037 (copy_bb_p): Update to use to_frequency.
1038 (fix_up_crossing_landing_pad): Likewise.
1039 (sanitize_hot_paths): Likewise.
1040 * bt-load.c (basic_block_freq): Likewise.
1041 * cfg.c (init_flow): Set count_max to uninitialized.
1042 (check_bb_profile): Remove frequencies; check counts.
1043 (dump_bb_info): Do not dump frequencies.
1044 (update_bb_profile_for_threading): Update counts only.
1045 (scale_bbs_frequencies_int): Likewise.
1046 (MAX_SAFE_MULTIPLIER): Remove.
1047 (scale_bbs_frequencies_gcov_type): Update counts only.
1048 (scale_bbs_frequencies_profile_count): Update counts only.
1049 (scale_bbs_frequencies): Update counts only.
1050 * cfg.h (struct control_flow_graph): Add count-max.
1051 (update_bb_profile_for_threading): Update prototype.
1052 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1053 (find_many_sub_basic_blocks): Likewise.
1054 * cfgcleanup.c (try_forward_edges): Likewise.
1055 (try_crossjump_to_edge): Likewise.
1056 * cfgexpand.c (expand_gimple_cond): Likewise.
1057 (expand_gimple_tailcall): Likewise.
1058 (construct_init_block): Likewise.
1059 (construct_exit_block): Likewise.
1060 * cfghooks.c (verify_flow_info): Check consistency of counts.
1061 (dump_bb_for_graph): Do not dump frequencies.
1062 (split_block_1): Do not update frequencies.
1063 (split_edge): Do not update frequencies.
1064 (make_forwarder_block): Do not update frequencies.
1065 (duplicate_block): Do not update frequencies.
1066 (account_profile_record): Do not update frequencies.
1067 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1068 for global heuristics.
1069 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1070 (expected_loop_iterations_unbounded): Use counts only.
1071 * cfgloopmanip.c (scale_loop_profile): Simplify.
1072 (create_empty_loop_on_edge): Simplify
1074 (duplicate_loop_to_header_edge): Simplify
1075 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1076 (update_br_prob_note): Take care of removing note when profile
1078 (relink_block_chain): Do not dump frequency.
1079 (rtl_account_profile_record): Use to_frequency.
1080 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1081 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1082 (cgraph_update_edges_for_call_stmt_node): Likewise.
1083 (cgraph_edge::verify_count_and_frequency): Update.
1084 (cgraph_node::verify_node): Temporarily disable frequency verification.
1085 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1086 to_cgraph_frequency.
1087 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1088 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1090 (cgraph_node::expand_thunk): Update profile.
1091 * except.c (dw2_build_landing_pads): Do not update frequency.
1092 * final.c (compute_alignments): Use to_frequency.
1093 (dump_basic_block_info): Do not dump frequency.
1094 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1095 (dump_gimple_bb_header): Do not dump frequency.
1096 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1098 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1099 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1100 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1101 (init_before_recovery): Do not update frequency.
1102 (sched_create_recovery_edges): Do not update frequency.
1103 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1104 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1105 (ipa_cp_c_finalize): Set max_count to uninitialized.
1106 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1107 (param_change_prob): Use counts.
1108 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1110 * ipa-split.c (consider_split): Use to_frequency.
1111 (split_function): Use to_frequency.
1112 * ira-build.c (loop_compare_func): Likewise.
1113 (mark_loops_for_removal): Likewise.
1114 (mark_all_loops_for_removal): Likewise.
1115 * loop-doloop.c (doloop_modify): Do not update frequency.
1116 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1118 * lto-streamer-in.c (input_function): Update count_max.
1119 * omp-expand.c (expand_omp_taskreg): Update count_max.
1120 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1121 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1122 (maybe_hot_count_p): Use ipa counts only.
1123 (maybe_hot_bb_p): Simplify.
1124 (maybe_hot_edge_p): Simplify.
1125 (probably_never_executed): Do not take frequency argument.
1126 (probably_never_executed_bb_p): Do not pass frequency.
1127 (probably_never_executed_edge_p): Likewise.
1128 (combine_predictions_for_bb): Check that profile is nonzero.
1129 (propagate_freq): Do not set frequency.
1130 (drop_profile): Simplify.
1131 (counts_to_freqs): Simplify.
1132 (expensive_function_p): Use to_frequency.
1133 (propagate_unlikely_bbs_forward): Simplify.
1134 (determine_unlikely_bbs): Simplify.
1135 (estimate_bb_frequencies): Add hack to silence graphite issues.
1136 (compute_function_frequency): Use ipa counts.
1137 (pass_profile::execute): Update.
1138 (rebuild_frequencies): Use counts only.
1139 (force_edge_cold): Use counts only.
1140 * profile-count.c (profile_count::dump): Dump new count types.
1141 (profile_count::differs_from_p): Check compatiblity.
1142 (profile_count::to_frequency): New function.
1143 (profile_count::to_cgraph_frequency): New function.
1144 * profile-count.h (struct function): Declare.
1145 (enum profile_quality): Add profile_guessed_local and
1146 profile_guessed_global0.
1147 (class profile_proability): Decrease number of bits to 29;
1148 update from_reg_br_prob_note and to_reg_br_prob_note.
1149 (class profile_count: Update comment; decrease number of bits
1150 to 61. Check compatibility.
1151 (profile_count::compatible_p): New private member function.
1152 (profile_count::ipa_p): New member function.
1153 (profile_count::operator<): Handle global zero correctly.
1154 (profile_count::operator>): Handle global zero correctly.
1155 (profile_count::operator<=): Handle global zero correctly.
1156 (profile_count::operator>=): Handle global zero correctly.
1157 (profile_count::nonzero_p): New member function.
1158 (profile_count::force_nonzero): New member function.
1159 (profile_count::max): New member function.
1160 (profile_count::apply_scale): Handle IPA scalling.
1161 (profile_count::guessed_local): New member function.
1162 (profile_count::global0): New member function.
1163 (profile_count::ipa): New member function.
1164 (profile_count::to_frequency): Declare.
1165 (profile_count::to_cgraph_frequency): Declare.
1166 * profile.c (OVERLAP_BASE): Delete.
1167 (compute_frequency_overlap): Delete.
1168 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1169 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1170 * sched-ebb.c (rank): Use counts only.
1171 * shrink-wrap.c (handle_simple_exit): Use counts only.
1172 (try_shrink_wrapping): Use counts only.
1173 (place_prologue_for_one_component): Use counts only.
1174 * tracer.c (find_best_predecessor): Use to_frequency.
1175 (find_trace): Use to_frequency.
1176 (tail_duplicate): Use to_frequency.
1177 * trans-mem.c (expand_transaction): Do not update frequency.
1178 * tree-call-cdce.c: Do not update frequency.
1179 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1180 (gimple_merge_blocks): Likewise.
1181 (gimple_split_edge): Likewise.
1182 (gimple_duplicate_sese_region): Likewise.
1183 (gimple_duplicate_sese_tail): Likewise.
1184 (move_sese_region_to_fn): Likewise.
1185 (gimple_account_profile_record): Likewise.
1186 (insert_cond_bb): Likewise.
1187 * tree-complex.c (expand_complex_div_wide): Likewise.
1188 * tree-eh.c (lower_resx): Update profile.
1189 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1191 (initialize_cfun): Do not initialize frequencies
1192 (freqs_to_counts): Delete.
1193 (copy_cfg_body): Ignore count parameter.
1194 (copy_body): Update.
1195 (expand_call_inline): Update count_max.
1196 (optimize_inline_calls): Update count_max.
1197 (tree_function_versioning): Update count_max.
1198 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1199 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1201 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1202 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1203 (try_peel_loop): Likewise.
1204 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1206 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1207 (tree_transform_and_unroll_loop): Do not use frequencies
1208 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1209 Use reliable prediction only.
1210 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1211 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1212 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1213 probability scaling.
1214 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1215 not update frequency
1216 (any_remaining_duplicated_blocks): Likewise.
1217 (update_profile): Likewise.
1218 (estimated_freqs_path): Delete.
1219 (freqs_to_counts_path): Delete.
1220 (clear_counts_path): Delete.
1221 (ssa_fix_duplicate_block_edges): Likewise.
1222 (duplicate_thread_path): Likewise.
1223 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1224 * tree-tailcall.c (decrease_profile): Do not update frequency.
1225 (eliminate_tail_call): Likewise.
1226 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1227 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1228 (optimize_mask_stores): Likewise.
1229 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1230 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1231 (ubsan_expand_ptr_ifn): Update profile.
1232 * value-prof.c (gimple_ic): Simplify.
1233 * value-prof.h (gimple_ic): Update prototype.
1234 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1235 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1237 (want_inline_self_recursive_call_p): Likewise.
1238 (resolve_noninline_speculation): Only cummulate defined counts.
1239 (inline_small_functions): Use nonzero_p.
1240 (ipa_inline): Do not access freed node.
1242 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1244 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1245 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1247 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1249 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1250 non-legitimate address.
1252 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1254 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1255 (*lt0_<mode>di, *lt0_<mode>si): New.
1257 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1259 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1260 TARGET_PAIRED_FLOAT.
1262 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1263 Jim Wilson <jim.wilson@linaro.org>
1265 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1266 * config/aarch64/aarch64-tune.md: Regenerated.
1267 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1268 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1270 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1272 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1273 (arc_expand_prologue): Restore blink for millicode.
1274 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1276 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1279 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1280 gen_vec_duplicate after forcing the scalar into a register.
1282 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1284 * combine (try_combine): Print the insns input to try_combine to the
1287 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1290 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1291 Remove second argument from aarch64_process_target_attr call.
1292 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1294 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1296 (aarch64_handle_attr_arch): Remove second argument.
1297 (aarch64_handle_attr_cpu): Ditto.
1298 (aarch64_handle_attr_tune): Ditto.
1299 (aarch64_handle_attr_isa_flags): Ditto.
1300 (aarch64_process_one_target_attr): Ditto.
1301 (aarch64_process_target_attr): Ditto.
1302 (aarch64_option_valid_attribute_p): Remove second argument.
1303 on aarch64_process_target_attr call.
1305 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1307 * diagnostic.c: Include "selftest-diagnostic.h".
1308 (selftest::assert_location_text): New function.
1309 (selftest::test_diagnostic_get_location_text): New function.
1310 (selftest::diagnostic_c_tests): Call it.
1312 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1314 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1315 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1316 (class selftest::test_diagnostic_context): Move to...
1317 * selftest-diagnostic.c: New file.
1318 * selftest-diagnostic.h: New file.
1320 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1322 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1323 offset range for FT32B.
1324 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1325 * config/ft32/ft32.md: Add TARGET_NOPM.
1326 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1327 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1329 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1331 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1333 2017-11-02 Jeff Law <law@redhat.com>
1335 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1336 virtual keyword on FINAL OVERRIDE members.
1338 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1339 virtuals together. Add virtual destructor.
1340 (substitute_and_fold_engine): Similarly.
1342 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1344 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1346 2017-11-02 Richard Biener <rguenther@suse.de>
1348 PR tree-optimization/82795
1349 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1351 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1353 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1355 (gcc_GAS_CHECK_FEATURE): Remove.
1356 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1357 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1358 * configure: Regenerate.
1360 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1362 * config/arc/arc.c (hwloop_optimize): Account for empty
1365 2017-11-02 Richard Biener <rguenther@suse.de>
1368 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1369 Truncate ARRAY_REF index and element size.
1371 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1373 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1375 2017-11-01 Jeff Law <law@redhat.com>
1377 * tree-ssa-ccp.c (ccp_folder): New class derived from
1378 substitute_and_fold_engine.
1379 (ccp_folder::get_value): New member function.
1380 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1381 (ccp_fold_stmt): Remove prototype.
1382 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1383 * tree-ssa-copy.c (copy_folder): New class derived from
1384 substitute_and_fold_engine.
1385 (copy_folder::get_value): Renamed from get_value.
1386 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1387 * tree-vrp.c (vrp_folder): New class derived from
1388 substitute_and_fold_engine.
1389 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1390 (vrp_folder::get_value): New member function.
1391 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1392 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1393 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1394 provide a class interface to folder/substitute routines.
1395 (ssa_prop_fold_stmt_fn): Remove typedef.
1396 (ssa_prop_get_value_fn): Likewise.
1397 (subsitute_and_fold): Remove prototype.
1398 (replace_uses_in): Likewise.
1399 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1400 Renamed from replace_uses_in. Call the virtual member function
1401 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1402 (substitute_and_fold_dom_walker): Remove initialization of
1403 data member entries for calbacks. Add substitute_and_fold_engine
1404 member and initialize it.
1405 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1406 member functions for get_value, replace_phi_args_in c
1407 replace_uses_in, and fold_stmt calls.
1408 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1409 substitute_and_fold. Remove assert. Update ctor call.
1411 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1412 (ssa_prop_visit_phi_fn): Likewise.
1413 (class ssa_propagation_engine): New class to provide an interface
1415 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1417 (ssa_prop_visit_phi): Likewise.
1418 (ssa_propagation_engine::simulate_stmt): Moved into class.
1419 Call visit_phi/visit_stmt from the class rather than via
1420 file scoped static variables.
1421 (ssa_propagation_engine::simulate_block): Moved into class.
1422 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1423 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1424 set file scoped statics for the visit_stmt/visit_phi callbacks.
1425 * tree-complex.c (complex_propagate): New class derived from
1426 ssa_propagation_engine.
1427 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1428 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1429 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1431 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1432 ssa_propagation_engine.
1433 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1434 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1435 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1436 * tree-ssa-copy.c (copy_prop): New class derived from
1437 ssa_propagation_engine.
1438 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1439 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1440 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1441 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1442 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1443 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1444 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1446 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1448 PR rtl-optimization/82778
1449 PR rtl-optimization/82597
1450 * compare-elim.c (struct comparison): Add in_a_setter field.
1451 (find_comparison_dom_walker::before_dom_children): Remove killed
1452 bitmap and df_simulate_find_defs call, instead walk the defs.
1453 Compute last_setter and initialize in_a_setter. Merge definitions
1454 with first initialization for a few variables.
1455 (try_validate_parallel): Use insn_invalid_p instead of
1456 recog_memoized. Return insn rather than just the pattern.
1457 (try_merge_compare): Fix up comment. Don't uselessly test if
1458 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1460 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1463 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1464 Alan Hayward <alan.hayward@arm.com>
1465 David Sherwood <david.sherwood@arm.com>
1467 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1468 aarch64_hard_regno_nregs to get the number of registers
1471 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1472 Alan Hayward <alan.hayward@arm.com>
1473 David Sherwood <david.sherwood@arm.com>
1475 * config/aarch64/constraints.md (Upl): Rename to...
1477 * config/aarch64/aarch64.md
1478 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1481 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1482 Alan Hayward <alan.hayward@arm.com>
1483 David Sherwood <david.sherwood@arm.com>
1485 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1486 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1489 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1490 Alan Hayward <alan.hayward@arm.com>
1491 David Sherwood <david.sherwood@arm.com>
1493 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1494 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1495 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1497 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1498 of a single element.
1499 * config/aarch64/iterators.md: Add a comment above the permute
1500 unspecs to say that they are generated directly by
1501 aarch64_expand_vec_perm_const.
1502 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1504 2017-11-01 Nathan Sidwell <nathan@acm.org>
1506 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1508 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1510 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1511 and medany code models, and describe what they do.
1513 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1515 Revert accidental duplicate:
1517 * combine.c (can_change_dest_mode): Reject changes in
1518 REGMODE_NATURAL_SIZE.
1520 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1522 PR rtl-optimization/64682
1523 PR rtl-optimization/69567
1524 PR rtl-optimization/69737
1525 PR rtl-optimization/82683
1526 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1527 register mentioned in the note, drop the note, unless it came from I3,
1528 in which case it should go to I3 again.
1530 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1532 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1533 and return false if not.
1534 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1536 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1538 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1539 range is known to be empty.
1541 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1542 Alan Hayward <alan.hayward@arm.com>
1543 David Sherwood <david.sherwood@arm.com>
1545 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1546 and CONST_VECTOR_NUNITS instead of computing the number of units from
1547 the byte sizes of the vector and element.
1548 (simplify_binary_operation_1): Likewise.
1549 (simplify_const_binary_operation): Likewise.
1550 (simplify_ternary_operation): Likewise.
1552 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1553 Alan Hayward <alan.hayward@arm.com>
1554 David Sherwood <david.sherwood@arm.com>
1556 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1557 (int_mem_offset): ...this new function.
1558 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1559 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1560 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1563 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1564 Alan Hayward <alan.hayward@arm.com>
1565 David Sherwood <david.sherwood@arm.com>
1567 * lower-subreg.c (interesting_mode_p): New function.
1568 (compute_costs, find_decomposable_subregs, decompose_register)
1569 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1570 (resolve_clobber, dump_choices): Use it.
1572 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1573 Alan Hayward <alan.hayward@arm.com>
1574 David Sherwood <david.sherwood@arm.com>
1576 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
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 * alias.c (find_base_value, find_base_term): Only process integer
1583 truncations. Check the precision rather than the size.
1585 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1586 Alan Hayward <alan.hayward@arm.com>
1587 David Sherwood <david.sherwood@arm.com>
1589 * machmode.h (is_narrower_int_mode): New function
1590 * optabs.c (expand_float, expand_fix): Use it.
1591 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1593 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1594 Alan Hayward <alan.hayward@arm.com>
1595 David Sherwood <david.sherwood@arm.com>
1597 * rtl.h (narrower_subreg_mode): New function.
1598 * ira-color.c (update_costs_from_allocno): Use it.
1600 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1601 Alan Hayward <alan.hayward@arm.com>
1602 David Sherwood <david.sherwood@arm.com>
1604 * optabs-query.h (convert_optab_p): New function, split out from...
1605 (convert_optab_handler): ...here.
1606 (widening_optab_handler): Delete.
1607 (find_widening_optab_handler): Remove permit_non_widening parameter.
1608 (find_widening_optab_handler_and_mode): Likewise. Provide an
1609 override that operates on mode class wrappers.
1610 * optabs-query.c (widening_optab_handler): Delete.
1611 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1612 parameter. Assert that the two modes are the same class and that
1613 the "from" mode is narrower than the "to" mode. Use
1614 convert_optab_handler instead of widening_optab_handler.
1615 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1616 instead of widening_optab_handler.
1617 * expr.c (expand_expr_real_2): Update calls to
1618 find_widening_optab_handler.
1619 * optabs.c (expand_widen_pattern_expr): Likewise.
1620 (expand_binop_directly): Take the insn_code as a parameter.
1621 (expand_binop): Only call find_widening_optab_handler for
1622 conversion optabs; use optab_handler otherwise. Update calls
1623 to find_widening_optab_handler and expand_binop_directly.
1624 Use convert_optab_handler instead of widening_optab_handler.
1625 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1626 find_widening_optab_handler and use scalar_mode rather than
1628 (convert_plusminus_to_widen): Likewise.
1630 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1631 Alan Hayward <alan.hayward@arm.com>
1632 David Sherwood <david.sherwood@arm.com>
1634 * machmode.h (fixed_size_mode): New class.
1635 * rtl.h (get_pool_mode): Return fixed_size_mode.
1636 * gengtype.c (main): Add fixed_size_mode.
1637 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1638 (get_raw_arg_mode): Likewise.
1639 * doc/tm.texi: Regenerate.
1640 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1641 * targhooks.c (default_get_reg_raw_mode): Likewise.
1642 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1643 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1644 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1645 (msp430_get_raw_result_mode): Likewise.
1646 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1647 * dbxout.c (dbxout_parms): Require fixed-size modes.
1648 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1649 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1650 * omp-low.c (lower_oacc_reductions): Likewise.
1651 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1652 (simplify_subreg): Update accordingly.
1653 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1654 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1655 that aren't fixed-size.
1656 (get_pool_mode): Return a fixed_size_mode.
1657 (output_constant_pool_2): Take a fixed_size_mode.
1659 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1660 Alan Hayward <alan.hayward@arm.com>
1661 David Sherwood <david.sherwood@arm.com>
1663 * doc/rtl.texi (vec_series): Document.
1664 (const): Say that the operand can be a vec_series.
1665 * rtl.def (VEC_SERIES): New rtx code.
1666 * rtl.h (const_vec_series_p_1): Declare.
1667 (const_vec_series_p): New function.
1668 * emit-rtl.h (gen_const_vec_series): Declare.
1669 (gen_vec_series): Likewise.
1670 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1671 (gen_vec_series): Likewise.
1672 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1673 * simplify-rtx.c (simplify_unary_operation): Handle negations
1675 (simplify_binary_operation_series): New function.
1676 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1677 (test_vector_ops_series): New function.
1678 (test_vector_ops): Call it.
1679 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1680 gen_const_vec_series.
1681 (altivec_lvsr): Likewise.
1682 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1684 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1685 Alan Hayward <alan.hayward@arm.com>
1686 David Sherwood <david.sherwood@arm.com>
1688 * doc/rtl.texi (const): Update description of address constants.
1689 Say that vector constants are allowed too.
1690 * common.md (E, F): Use CONSTANT_P instead of checking for
1692 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1693 checking for CONST_VECTOR.
1694 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1696 * expr.c (expand_expr_real_2): Check for vector modes instead
1697 of checking for CONST_VECTOR.
1698 * rtl.h (const_vec_p): New function.
1699 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1700 (unwrap_const_vec_duplicate): Handle them here too.
1702 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1703 David Malcolm <dmalcolm@redhat.com>
1704 Alan Hayward <alan.hayward@arm.com>
1705 David Sherwood <david.sherwood@arm.com>
1707 * rtl.h (vec_duplicate_p): New function.
1708 * selftest-rtl.c (assert_rtx_eq_at): New function.
1709 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1710 (assert_rtx_eq_at): Declare.
1711 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1712 * selftest-run-tests.c (selftest::run_tests): Call it.
1713 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1714 (simplify_unary_operation_1): Recursively handle vector duplicates.
1715 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1717 (simplify_subreg): Handle subregs of vector duplicates.
1718 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1719 (selftest::simplify_rtx_c_tests): New functions.
1721 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1722 Alan Hayward <alan.hayward@arm.com>
1723 David Sherwood <david.sherwood@arm.com>
1725 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1726 (gen_vec_duplicate): Likewise.
1727 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1729 (gen_const_vector): ...here.
1730 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1731 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1732 whose elements are all equal.
1733 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1734 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1735 (simplify_relational_operation): Likewise.
1736 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1738 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1739 (aarch64_expand_vector_init): Likewise.
1740 * config/arm/arm.c (neon_vdup_constant): Likewise.
1741 (neon_expand_vector_init): Likewise.
1742 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1743 (arm_block_set_unaligned_vect): Likewise.
1744 (arm_block_set_aligned_vect): Likewise.
1745 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1746 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1747 (expand_vec_perm_even_odd_pack): Likewise.
1748 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1749 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1750 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1751 gen_const_vec_duplicate.
1752 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1753 * config/mips/mips.c (mips_gen_const_int_vector): Use
1754 gen_const_vec_duplicate.
1755 (mips_expand_vector_init): Use CONST0_RTX.
1756 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1757 (define_split): Use gen_const_vec_duplicate.
1758 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1759 (define_split): Use gen_const_vec_duplicate.
1760 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1761 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1762 * config/spu/spu.c (spu_const): Likewise.
1764 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1765 Alan Hayward <alan.hayward@arm.com>
1766 David Sherwood <david.sherwood@arm.com>
1768 * combine.c (can_change_dest_mode): Reject changes in
1769 REGMODE_NATURAL_SIZE.
1771 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1773 * configure.ac (--enable-libssp): New.
1774 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1775 trying to determine target-specific default. Adjust indentation.
1776 * configure: Regenerated.
1777 * doc/install.texi (Configuration): Expand --disable-libssp
1780 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1782 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1785 2017-10-31 Martin Jambor <mjambor@suse.cz>
1788 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1790 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1792 * auto-profile.c (autofdo_source_profile::read): Use
1793 UNKNOWN_LOCATION rather than 0.
1794 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1795 (warning_at): ...this overload.
1796 (warning_at_rich_loc_n): Rename to...
1797 (warning_n): ...this overload.
1798 (error_at_rich_loc): Rename to...
1799 (error_at): ...this overload.
1800 (pedwarn_at_rich_loc): Rename to...
1801 (pedwarn): ...this overload.
1802 (permerror_at_rich_loc): Rename to...
1803 (permerror): ...this overload.
1804 (inform_at_rich_loc): Rename to...
1805 (inform): ...this overload.
1806 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1807 (diagnostic_n_impl_richloc): Rename to...
1808 (diagnostic_n_impl): ...this rich_location *-based decl.
1809 (inform_at_rich_loc): Rename to...
1810 (inform): ...this, and add an assertion.
1811 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1812 (warning_at_rich_loc): Rename to...
1813 (warning_at): ...this, and add an assertion.
1814 (warning_at_rich_loc_n): Rename to...
1815 (warning_n): ...this, and add an assertion.
1816 (warning_n): Update location_t-based implementation for removal of
1817 location_t-based diagnostic_n_impl.
1818 (pedwarn_at_rich_loc): Rename to...
1819 (pedwarn): ...this, and add an assertion.
1820 (permerror_at_rich_loc): Rename to...
1821 (permerror): ...this, and add an assertion.
1822 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1823 (error_at_rich_loc): Rename to...
1824 (error_at): ...this, and add an assertion.
1825 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1826 (driver::do_spec_on_infiles): Likewise.
1827 * substring-locations.c (format_warning_va): Update for renaming
1828 of inform_at_rich_loc.
1830 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1832 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1833 _Float<N>X built-in functions so that the variant without the
1834 "__builtin_" prefix is only enabled for the GNU C and Objective C
1835 languages when they are in non-strict ANSI/ISO mode.
1836 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1837 * target.def (floatn_builtin_p): Add a target hook to control
1838 whether _Float<N> and _Float<N>X built-in functions without the
1839 "__builtin_" prefix are enabled, and return true for C and
1840 Objective C in the default hook. Include langhooks.h in
1842 * targhooks.h (default_floatn_builtin_p): Likewise.
1843 * targhooks.c (default_floatn_builtin_p): Likewise.
1844 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1845 floatn_builtin_p target hook.
1846 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1848 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1849 Eric Botcazou <ebotcazou@adacore.com>
1851 PR rtl-optimization/81803
1852 * lra-constraints.c (curr_insn_transform): Also reload the whole
1853 register for a strict subreg no wider than a word if this is for
1854 a WORD_REGISTER_OPERATIONS target.
1856 2017-10-31 Jason Merrill <jason@redhat.com>
1858 * gdbinit.in: Skip over inlines from timevar.h.
1860 2017-10-31 Martin Liska <mliska@suse.cz>
1862 * doc/gcov.texi: Document new option.
1863 * gcov.c (print_usage): Likewise print it.
1864 (process_args): Support the argument.
1865 (format_count): New function.
1866 (format_gcov): Use the function.
1868 2017-10-31 Martin Liska <mliska@suse.cz>
1870 * gcov.c (struct name_map): do not use typedef.
1871 Define operator== and operator<.
1872 (name_search): Remove.
1873 (name_sort): Remove.
1874 (main): Do not allocate names.
1875 (process_file): Add vertical space.
1876 (generate_results): Use std::find.
1877 (release_structures): Do not release memory.
1878 (find_source): Use std::find.
1880 2017-10-31 Martin Liska <mliska@suse.cz>
1882 * gcov.c (struct line_info): Remove it's typedef.
1883 (line_info::line_info): Add proper ctor.
1884 (line_info::has_block): Do not use a typedef.
1885 (struct source_info): Do not use typedef.
1886 (circuit): Likewise.
1887 (get_cycles_count): Likewise.
1888 (output_intermediate_file): Iterate via vector iterator.
1889 (add_line_counts): Use std::vector methods.
1890 (accumulate_line_counts): Likewise.
1891 (output_lines): Likewise.
1893 2017-10-31 Martin Liska <mliska@suse.cz>
1895 * gcov.c (struct source_info): Remove typedef.
1896 (source_info::source_info): Add proper ctor.
1897 (accumulate_line_counts): Use struct, not it's typedef.
1898 (output_gcov_file): Likewise.
1899 (output_lines): Likewise.
1900 (main): Do not allocate an array.
1901 (output_intermediate_file): Use size of vector container.
1902 (process_file): Resize the vector.
1903 (generate_results): Do not preallocate, use newly added vector
1905 (release_structures): Do not release sources.
1906 (find_source): Use vector methods.
1907 (add_line_counts): Do not use typedef.
1909 2017-10-31 Martin Liska <mliska@suse.cz>
1911 * doc/gcov.texi: Document that.
1912 * gcov.c (add_line_counts): Mark lines with a non-executed
1914 (output_line_beginning): Handle such lines.
1915 (output_lines): Pass new argument.
1916 (output_intermediate_file): Print it in intermediate format.
1918 2017-10-31 Martin Liska <mliska@suse.cz>
1920 * color-macros.h: New file.
1921 * diagnostic-color.c: Factor out color related to macros to
1923 * doc/gcov.texi: Document -k option.
1924 * gcov.c (INCLUDE_STRING): Include string.h.
1925 (print_usage): Add -k option.
1926 (process_args): Parse it.
1927 (pad_count_string): New function.
1928 (output_line_beginning): Likewise.
1929 (DEFAULT_LINE_START): New macro.
1930 (output_lines): Support color output.
1932 2017-10-31 Martin Liska <mliska@suse.cz>
1934 PR gcov-profile/82633
1935 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1936 their interaction with GCOV infrastructure.
1937 * configure.ac: Add -fkeep-{inline,static}-functions to
1939 * configure: Regenerate.
1941 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1944 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1946 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1949 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1950 into a register if it does not fit into an immediate offset field.
1952 2017-10-31 Olivier Hainque <hainque@adacore.com>
1954 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1956 2017-10-31 Julia Koval <julia.koval@intel.com>
1958 * config.gcc: Add gfniintrin.h.
1959 * config/i386/gfniintrin.h: New.
1960 * config/i386/i386-builtin-types.def
1961 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1962 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1963 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1964 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1965 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1966 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1967 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1968 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1969 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1970 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1971 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1972 * config/i386/immintrin.h: Include gfniintrin.h.
1973 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1975 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1977 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1979 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1981 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1982 (arm_ashldi3_1bit): Remove pattern.
1983 (ashrdi3): Remove shift by 1 expansion.
1984 (arm_ashrdi3_1bit): Remove pattern.
1985 (lshrdi3): Remove shift by 1 expansion.
1986 (arm_lshrdi3_1bit): Remove pattern.
1987 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1988 cost of ashldi3 by 1.
1989 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1990 (<shift>di3_neon): Likewise.
1992 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1994 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1995 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1996 and (*aarch64_simd_mov<VQ:mode>).
1997 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1998 pattern alternative.
1999 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2000 attributes to match pattern alternative.
2002 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2004 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2005 * config/rs6000/emmintrin.h: New file.
2006 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2008 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2010 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2011 * (movdi_vfp_cortexa8): Remove pattern.
2013 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2015 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2017 (Specific, alpha*-dec-osf5.1): Remove.
2018 (Specific, mips-sgi-irix5): Remove.
2019 (Specific, mips-sgi-irix6): Remove.
2021 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2024 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2025 arguments to clear_bit_region_be.
2027 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2029 * gimplify.c: Include memmodel.h.
2031 2017-10-30 Martin Jambor <mjambor@suse.cz>
2033 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2034 condition whether loop should be executed at all.
2036 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2038 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2039 gimple folding of vec_madd() intrinsics.
2040 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2041 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2042 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2044 2017-10-30 Richard Biener <rguenther@suse.de>
2046 PR tree-optimization/82762
2048 2017-10-23 Richard Biener <rguenther@suse.de>
2050 PR tree-optimization/82129
2052 2017-08-01 Richard Biener <rguenther@suse.de>
2054 PR tree-optimization/81181
2055 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2056 (compute_antic): ... end of iteration here.
2058 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2060 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2061 -std=iso9899:2017 and -std=gnu17.
2062 * doc/standards.texi (C Language): Document C17 support.
2063 * doc/cpp.texi (Overview): Mention -std=c17.
2064 (Standard Predefined Macros): Document C11 and C17 values of
2065 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2066 * doc/extend.texi (Inline): Do not list individual options for
2067 standards newer than C99.
2068 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2070 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2073 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2075 * asan.c (asan_finish_file): Align asan globals array by shadow
2078 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2081 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2082 (struct store_immediate_info): Add bitregion_start and bitregion_end
2084 (store_immediate_info::store_immediate_info): Add brs and bre
2085 arguments and initialize bitregion_{start,end} from those.
2086 (struct merged_store_group): Add bitregion_start, bitregion_end,
2087 align_base and mask fields. Drop unnecessary struct keyword from
2088 struct store_immediate_info. Add do_merge method.
2089 (clear_bit_region_be): Use memset instead of loop storing zeros.
2090 (merged_store_group::do_merge): New method.
2091 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2092 stores as long as the surrounding bitregions have no gaps.
2093 (merged_store_group::merge_overlapping): Use do_merge.
2094 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2095 is byte aligned, rather than requiring that start and width are
2096 byte aligned. Drop unnecessary struct keyword from
2097 struct store_immediate_info. Allocate and populate also mask array.
2098 Make start of the arrays relative to bitregion_start rather than
2099 start and size them according to bitregion_{end,start} difference.
2100 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2101 struct store_immediate_info.
2102 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2103 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2104 struct keyword from struct store_immediate_info.
2105 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2106 between stores as long as the surrounding bitregions have no gaps.
2108 (struct split_store): Add orig non-static data member.
2109 (split_store::split_store): Initialize orig to false.
2110 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2111 if there is exactly a single original stmt. Change stmts argument
2112 to pointer from reference, if NULL, don't push anything to it. Add
2113 first argument, use it to optimize skipping over orig stmts that
2114 are known to be before bitpos already. Simplify.
2115 (split_group): Return unsigned int count how many stores are or
2116 would be needed rather than a bool. Add allow_unaligned argument.
2117 Change split_stores argument from reference to pointer, if NULL,
2118 only do a dry run computing how many stores would be produced.
2119 Rewritten algorithm to use both alignment and misalign if
2120 !allow_unaligned and handle bitfield stores with gaps.
2121 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2122 from bitregion_start instead of start. Compute allow_unaligned
2123 here, if true, do 2 split_group dry runs to compute which one
2124 produces fewer stores and prefer aligned if equal. Punt if
2125 new count is bigger or equal than original before emitting any
2126 statements, rather than during that. Remove no longer needed
2127 new_ssa_names tracking. Replace num_stmts with
2128 split_stores.length (). Use 32-bit stack allocated entries
2129 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2130 if possible. Handle bitfields with gaps.
2131 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2132 Compute bitregion_{start,end} for the stores and construct
2133 store_immediate_info with that. Formatting fixes.
2135 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2138 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2139 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2141 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2143 * gimplify.c: Include tm_p.h.
2145 * common.opt (gcoff): Re-add as ignored option.
2146 (gcoff1, gcoff2, gcoff3): Likewise.
2148 * Makefile.in (OBJS): Delete sdbout.o.
2149 (GTFILES): Delete $(srcdir)/sdbout.c.
2150 * debug.h: Delete sdb_debug_hooks.
2151 * final.c: Delete sdbout.h include.
2152 (final_scan_insn): Delete SDB_DEBUG check.
2153 (rest_of_clean_state): Likewise.
2154 * output.h: Delete sdb_begin_function_line.
2157 * toplev.c: Delete sdbout.h include.
2158 (process_options): Delete SDB_DEBUG check.
2159 * tree-core.h (tree_type_common): Delete pointer field of
2161 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2162 TYPE_SYMTAB_POINTER.
2163 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2164 (TYPE_SYMTAB_IS_POINTER): Delete.
2165 (TYPE_SYMTAB_IS_DIE): Renumber.
2166 * xcoffout.c: Refer to former sdbout.c file.
2167 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2169 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2170 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2171 (Debugging Options): Delete -gcoff.
2172 (-gstabs): Delete SDB reference.
2174 (-gcoff@var{level}): Delete.
2175 * doc/passes.texi (Debugging information output): Delete SDB and
2176 sdbout.c references.
2177 * doc/tm.texi: Regenerate.
2178 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2179 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2181 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2182 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2183 (SDB_DEBUGGING_INFO): Delete.
2184 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2185 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2186 * target.def (output_source_filename): Delete COFF reference.
2188 * common.opt (gcoff): Delete.
2189 (gxcoff+): Update Negative chain.
2190 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2192 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2193 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2194 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2196 (expand_function_start): Change sdb reference to past tense.
2197 (expand_function_end): Change sdb reference to past tense.
2198 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2199 * opts.c (debug_type_names): Delete coff entry.
2200 (common_handle_option): Delete OPT_gcoff case.
2201 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2203 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2204 * config/cris/cris.h: Delete SDB reference in comment.
2205 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2206 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2207 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2208 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2210 (ix86_expand_prologue): Likewise.
2211 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2212 * config/ia64/ia64.h: Likewise.
2213 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2214 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2216 * config/mmix/mmix.h: Likewise.
2217 * config/nds32/nds32.c: Likewise.
2218 * config/stormy/storym16.h: Likewise.
2219 * config/visium/visium.h: Likewise.
2220 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2222 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2224 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2225 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2226 FRAME_POINTER_REGNUM point at high end of local var area.
2228 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2230 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2231 Move comment around. Do not reset best_edge for a copiable
2232 destination if the copy would cause a partition change.
2233 (better_edge_p): Remove redundant check.
2235 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2237 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2239 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2241 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2242 for math functions that have _Float<N> and _Float<N>X variants.
2243 (mathfn_built_in_2): Add support for math functions that have
2244 _Float<N> and _Float<N>X variants.
2245 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2246 (expand_builtin_mathfn_ternary): Add support for fma with
2247 _Float<N> and _Float<N>X variants.
2248 (expand_builtin): Likewise.
2249 (fold_builtin_3): Likewise.
2250 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2251 create math function _Float<N> and _Float<N>X variants as external
2253 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2254 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2255 the __builtin_ prefix and if not strict ansi, without the prefix.
2256 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2257 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2258 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2259 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2260 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2261 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2262 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2263 function signatures for fma _Float<N> and _Float<N>X variants.
2264 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2265 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2266 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2267 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2268 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2269 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2270 * gencfn-macros.c (print_case_cfn): Add support for math functions
2271 that have _Float<N> and _Float<N>X variants.
2272 (print_define_operator_list): Likewise.
2273 (fltfn_suffixes): Likewise.
2275 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2276 for math functions that have _Float<N> and _Float<N>X variants.
2277 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2278 and _Float<N>X variants.
2279 (COPYSIGN): Likewise.
2282 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2283 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2285 (integer_valued_read_call_p): Likewise.
2286 * fold-const-call.c (fold_const_call_ss): Likewise.
2287 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2288 _Float<N> and _Float<N>X variants.
2289 (fold_const_call_ssss): Add support for fma _Float<N> and
2290 _Float<N>X variants.
2291 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2292 support for copysign and fma _Float<N> and _Float<N>X variants.
2293 (backprop::process_builtin_call_use): Likewise.
2294 * tree-call-cdce.c (can_test_argument_range); Add support for
2295 sqrt _Float<N> and _Float<N>X variants.
2296 (edom_only_function): Likewise.
2297 (get_no_error_domain): Likewise.
2298 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2299 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2300 copysign _Float<N> and _Float<N>X variants.
2301 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2302 handled by machine independent code.
2303 (FMAF128): Likewise.
2304 * doc/cpp.texi (Common Predefined Macros): Document defining
2305 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2306 fma _Float<N> and _Float<N>X variants.
2308 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2311 * config/i386/i386-modes.def (CCFPU): Remove definition.
2312 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2313 (ix86_cc_modes_compatible): Ditto.
2314 (ix86_expand_carry_flag_compare): Ditto.
2315 (ix86_expand_int_movcc): Ditto.
2316 (ix86_expand_int_addcc): Ditto.
2317 (ix86_reverse_condition): Ditto.
2318 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2319 Return true/false for unordered/ordered fp comparisons.
2320 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2321 (ix86_prepare_fp_compare_args): Update for rename.
2322 (ix86_expand_fp_compare): Update for rename. Generate unordered
2323 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2324 (ix86_expand_sse_compare_and_jump): Ditto.
2325 * config/i386/predicates.md (fcmov_comparison_operator):
2326 Remove CCFPU mode handling.
2327 (ix86_comparison_operator): Ditto.
2328 (ix86_carry_flag_operator): Ditto.
2329 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2330 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2331 (*cmpu<mode>_cc_i387): Ditto.
2332 (FPCMP): Remove mode iterator.
2333 (unord): Remove mode attribute.
2334 (unord_subst): New define_subst transformation
2335 (unord): New define_subst attribute.
2337 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2338 (*cmpi<unord>xf_i387): Ditto.
2339 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2340 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2341 using unord_subst transformation.
2342 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2343 (round_saeonly): Also handle CCFP mode.
2344 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2345 Remove UNSPEC_SAHF unspec handling.
2347 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2349 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2351 2017-10-27 Jeff Law <law@redhat.com>
2353 * gimple-ssa-sprintf.c: Include domwalk.h.
2354 (class sprintf_dom_walker): New class, derived from dom_walker.
2355 (sprintf_dom_walker::before_dom_children): New function.
2356 (struct call_info): Moved into sprintf_dom_walker class
2357 (compute_formath_length, handle_gimple_call): Likewise.
2358 (sprintf_length::execute): Call the dominator walker rather
2359 than walking the statements.
2361 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2362 gimple statement locations.
2363 (check_array_bounds): Corresponding changes. Get the statement's
2364 location directly from wi->stmt.
2366 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2369 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2371 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2373 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2374 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2376 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2379 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2380 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2381 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2382 maybe_get_pool_constant.
2383 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2386 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2388 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2389 2.26 caveat. Update gas and gld versions.
2390 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2393 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2395 * cgraph.h (set_malloc_flag): Declare.
2396 * cgraph.c (set_malloc_flag_1): New function.
2397 (set_malloc_flag): Likewise.
2398 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2399 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2401 (read_ipa_call_summary): Add support for reading is_return_callee.
2402 (write_ipa_call_summary): Stream is_return_callee.
2403 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2404 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2405 ipa-prop.h, ipa-fnsummary.h.
2406 (pure_const_names): Change to static.
2407 (malloc_state_e): Define.
2408 (malloc_state_names): Define.
2409 (funct_state_d): Add field malloc_state.
2410 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2411 (check_retval_uses): New function.
2412 (malloc_candidate_p): Likewise.
2413 (analyze_function): Add support for malloc attribute.
2414 (pure_const_write_summary): Stream malloc_state.
2415 (pure_const_read_summary): Add support for reading malloc_state.
2416 (dump_malloc_lattice): New function.
2417 (propagate_malloc): New function.
2418 (warn_function_malloc): New function.
2419 (ipa_pure_const::execute): Call propagate_malloc and
2420 ipa_free_fn_summary.
2421 (pass_local_pure_const::execute): Add support for malloc attribute.
2422 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2423 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2425 2017-10-27 Martin Liska <mliska@suse.cz>
2427 PR gcov-profile/82457
2428 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2429 for fork-like functions to be properly instrumented.
2431 2017-10-27 Richard Biener <rguenther@suse.de>
2434 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2435 info when we redirected EH.
2437 2017-10-26 Michael Collison <michael.collison@arm.com>
2439 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2441 (<optab>_trunchf<GPI:mode>2: New pattern.
2442 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2443 * config/aarch64/iterators.md (wv): New mode attribute.
2444 (vf, VF): New mode attributes.
2445 (vgp, VGP): New mode attributes.
2446 (s): Update attribute with SImode and DImode prefixes.
2448 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2450 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2451 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2452 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2453 (nios2_option_overide): Initialize it. Don't allow R0-relative
2454 addressing with PIC.
2455 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2456 (nios2_symbolic_constant_p): Likewise.
2457 (nios2_legitimate_address_p): Likewise.
2458 (nios2_r0rel_section_name_p): New.
2459 (nios2_symbol_ref_in_r0rel_data_p): New.
2460 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2461 (r0rel_constant_p): New.
2462 (nios2_print_operand_address): Handle r0rel_constant_p.
2463 (nios2_cdx_narrow_form_p): Likewise.
2464 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2465 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2466 (Nios II Options): Document -mr0rel-sec.
2468 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2470 * config/nios2/nios2.c: Include xregex.h.
2471 (nios2_gprel_sec_regex): New.
2472 (nios2_option_overide): Initialize it. Don't allow GP-relative
2473 addressing with PIC.
2474 (nios2_small_section_name_p): Check for regex match.
2475 * config/nios2/nios2.opt (mgprel-sec=): New option.
2476 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2477 (Nios II Options): Document -mgprel-sec.
2479 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2481 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2483 2017-10-26 Tom de Vries <tom@codesourcery.com>
2485 PR tree-optimization/82707
2486 * gimple.c (gimple_copy): Fix unsharing of
2487 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2489 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2491 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2492 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2493 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2494 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2495 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2496 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2497 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2498 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2499 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2500 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2501 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2502 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2503 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2504 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2505 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2506 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2507 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2509 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2511 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2513 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2514 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2515 warning to rs6000.c. Remove the Undocumented flag, since it has
2517 (-mabi=ibmlongdouble): Likewise.
2518 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2519 already set, set the default format for long double.
2520 (rs6000_debug_reg_global): Print whether long double is IBM or
2522 (rs6000_option_override_internal): Rework setting long double
2523 format. Only warn if the user is changing the long double default
2524 and they did not use -Wno-psabi.
2525 * doc/invoke.texi (PowerPC options): Update the documentation for
2526 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2528 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2529 Alan Hayward <alan.hayward@arm.com>
2530 David Sherwood <david.sherwood@arm.com>
2532 * rtl.h (wider_subreg_mode): New function.
2533 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2534 rather than an unsigned int *.
2535 * ira-color.c (regno_max_ref_width): Replace with...
2536 (regno_max_ref_mode): ...this new variable.
2537 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2538 Use wider_subreg_mode.
2539 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2540 rather than an unsigned int *.
2541 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2542 (process_alt_operands): Likewise.
2543 (invariant_p): Likewise.
2544 * lra-spills.c (assign_mem_slot): Likewise.
2545 (add_pseudo_to_slot): Likewise.
2546 * lra.c (collect_non_operand_hard_regs): Likewise.
2547 (add_regs_to_insn_regno_info): Likewise.
2548 * reload1.c (regno_max_ref_width): Replace with...
2549 (regno_max_ref_mode): ...this new variable.
2550 (reload): Update accordingly. Update call to
2551 ira_sort_regnos_for_alter_reg.
2552 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2553 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2554 (scan_paradoxical_subregs): Likewise.
2556 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2558 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2559 (aarch64_frame): Add emit_frame_chain boolean.
2560 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2561 Move eh_return case to aarch64_layout_frame.
2562 (aarch64_layout_frame): Initialize emit_frame_chain.
2563 (aarch64_expand_prologue): Use emit_frame_chain.
2565 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2567 * config/aarch64/aarch64.c (aarch64_layout_frame):
2568 Ensure LR is always stored at the bottom of the callee-saves.
2569 Remove rarely used frame layout which saves callee-saves at top of
2570 frame, so the store of LR can be used as a valid probe in all cases.
2572 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2574 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2575 Improve unaligned TImode/TFmode base/offset split.
2577 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2578 Alan Hayward <alan.hayward@arm.com>
2579 David Sherwood <david.sherwood@arm.com>
2581 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2582 * combine.c (find_single_use_1): Likewise.
2583 (expand_field_assignment): Likewise.
2584 (move_deaths): Likewise.
2585 * lra-constraints.c (simplify_operand_subreg): Likewise.
2586 (curr_insn_transform): Likewise.
2587 * lra.c (collect_non_operand_hard_regs): Likewise.
2588 (add_regs_to_insn_regno_info): Likewise.
2589 * rtlanal.c (reg_referenced_p): Likewise.
2590 (covers_regno_no_parallel_p): Likewise.
2592 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2594 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2595 Don't print any bits outside the precision of the value.
2596 * wide-int.cc (test_printing): Add some new tests.
2598 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2600 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2601 supports -xbrace_comment option.
2602 * configure: Regenerate.
2603 * config.in: Regenerate.
2604 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2605 (ASM_CPU_SPEC): Use it.
2607 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2609 * target.def (static_rtx_alignment): New hook.
2610 * targhooks.h (default_static_rtx_alignment): Declare.
2611 * targhooks.c (default_static_rtx_alignment): New function.
2612 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2613 * doc/tm.texi: Regenerate.
2614 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2615 instead of targetm.constant_alignment. Remove call to
2617 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2618 (cris_preferred_mininum_alignment): New function, split out from...
2619 (cris_constant_alignment): ...here.
2620 (cris_static_rtx_alignment): New function.
2621 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2623 (ix86_constant_alignment): ...here.
2624 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2625 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2626 (mmix_static_rtx_alignment): New function.
2627 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2628 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2630 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2633 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2634 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2636 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2638 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2639 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2640 mark as a sign-extending load.
2641 (local_pic_load_u): Define.
2643 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2646 * fold-const.c (operand_equal_for_comparison_p): Also return true
2647 if ARG0 is a simple variant of ARG1 with narrower precision.
2648 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2650 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2652 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2654 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2655 scatter_static, scatter_per_elt.
2656 * x86-tune-costs.h: Add new cost entries.
2658 2017-10-25 Richard Biener <rguenther@suse.de>
2660 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2661 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2662 class pass_fre): Move to ...
2663 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2664 class pass_fre): ... here and adjust for statistics.
2666 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2669 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2670 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2671 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2672 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2675 2017-10-25 Richard Biener <rguenther@suse.de>
2677 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2678 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2679 eliminate_push_avail, eliminate_insert): Move inside...
2680 (class eliminate_dom_walker): ... this class in preparation
2682 (fini_eliminate): Remove by merging with ...
2683 (eliminate): ... this function. Adjust for class changes.
2684 (pass_pre::execute): Remove fini_eliminate call.
2685 (pass_fre::execute): Likewise.
2687 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2690 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2691 (VPERMI2, VPERMI2I): New mode iterators.
2692 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2693 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2695 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2696 mode iterator. Remove 3 old define_insn patterns.
2697 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2698 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2699 VPERMI2 mode iterator, remove the other two expanders.
2700 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2701 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2702 instructions, remove the other two patterns.
2703 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2704 mode iterator, remove the other two patterns.
2705 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2706 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2707 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2708 and adjust argument order accordingly.
2709 (ix86_expand_vec_perm): Adjust caller.
2710 (expand_vec_perm_1): Likewise.
2711 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2712 (expand_vec_perm_vpermt2_vpshub2): ... this.
2713 (ix86_expand_vec_perm_const_1): Adjust caller.
2714 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2717 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2718 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2719 (vec_shl_<mode>): Remove unused expander.
2720 (avx512bw_<shift_insn><mode>3): New define_insn.
2721 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2722 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2724 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2727 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2730 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2732 PR rtl-optimization/82396
2733 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2734 (autopref_multipass_init): Simplify initialization.
2735 (autopref_rank_data): Simplify sort order.
2736 * gcc/sched-int.h (autopref_multipass_data_): Remove
2737 multi_mem_insn_p, min_offset and max_offset.
2739 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2742 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2743 Check special value of flag_omit_frame_pointer.
2744 (aarch64_can_eliminate): Likewise.
2745 (aarch64_override_options_after_change_1): Simplify handling of
2746 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2748 2017-10-24 Richard Biener <rguenther@suse.de>
2750 PR tree-optimization/82697
2751 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2752 zero for conditional load and unconditional store.
2754 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2756 * doc/install.texi: Document bootstrap-cet.
2758 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2761 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2762 ENDBR instruction at function entrance if function is only
2765 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2768 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2769 patterns to better describe from which operation the CF is computed.
2770 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2771 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2772 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2773 is 0, use _0 suffixed expanders instead of emitting a comparison
2776 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2778 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2779 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2781 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2784 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2785 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2786 * loop-iv.c (iv_get_reaching_def): Likewise.
2787 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2788 variable is promoted and the partition contains undefined values.
2790 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2792 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2794 (nios2_address_cost): Define.
2795 (nios2_legitimize_address): Recognize (exp + constant) directly.
2796 (TARGET_ADDRESS_COST): Define.
2798 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2800 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2801 (nios2_symbolic_memory_operand_p): Declare.
2802 (nios2_split_large_constant): Declare.
2803 (nios2_split_symbolic_memory_operand): Declare.
2804 * config/nios2/nios2.c: Adjust includes.
2805 (nios2_symbolic_constant_allowed): New.
2806 (nios2_symbolic_constant_p): New.
2807 (nios2_plus_symbolic_constant_p): New.
2808 (nios2_valid_addr_expr_p): Recognize addresses involving
2810 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2811 (nios2_symbolic_memory_operand_p): New.
2812 (nios2_large_constant_p): New.
2813 (nios2_split_large_constant): New.
2814 (nios2_split_plus_large_constant): New.
2815 (nios2_split_symbolic_memory_operand): New.
2816 (nios2_legitimize_address): Code refactoring. Handle addresses
2817 involving symbolic constants.
2818 (nios2_emit_move_sequence): Likewise.
2819 (nios2_print_operand): Improve error output.
2820 (nios2_print_operand_address): Handle LO_SUM.
2821 (nios2_cdx_narrow_form_p): Likewise.
2822 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2823 operands involving symbolic constants.
2824 (movhi_internal, movsi_internal): Likewise.
2825 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2826 (extendhisi2, extendqi<mode>2): Likewise.
2828 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2830 * tree-pass.h (PROP_rtl_split_insns): Define.
2831 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2833 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2835 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2837 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2840 * target.def (const_not_ok_for_debug_p): Default to
2841 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2842 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2843 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2844 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2845 which targetm.const_not_ok_for_debug_p returned true.
2846 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2848 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2850 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2851 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2852 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2854 (ix86_const_not_ok_for_debug_p): New function.
2855 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2856 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2858 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2861 * system.h: Conditionally include "unique-ptr.h" if
2862 INCLUDE_UNIQUE_PTR is defined.
2863 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2864 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2866 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2868 * config/rl78/rl78.md: New define_expand "subdi3".
2870 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2873 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2874 DF_REF_INSN if DF_REF_INSN_INFO is false.
2876 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2878 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2879 xmm_move instead of sse_move.
2880 (sse_store_index): New function.
2881 (ix86_register_move_cost): Be more sensible about mismatch stall;
2882 model AVX moves correctly; make difference between sse->integer and
2884 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2885 moves; make difference between SSE and AVX.
2886 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2887 and zmm_move. Increase size of sse load and store tables;
2888 add unaligned load and store tables; add ssemmx_to_integer.
2889 * x86-tune-costs.h: Update all entries according to real
2890 move latencies from Agner Fog's manual and chip documentation.
2892 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2895 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2896 * config/i386/constraints.md (Wf): New constraint.
2897 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2898 (cmp<dwi>_doubleword): Removed.
2899 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2900 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2901 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2902 expand with cmp<dwi>_doubleword. For LTU and GEU use
2903 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2905 * common.opt (gcolumn-info): Enable by default.
2906 * doc/invoke.texi (gcolumn-info): Document new default.
2908 2017-10-23 Richard Biener <rguenther@suse.de>
2910 PR tree-optimization/82672
2911 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2912 Fold the stmt if we propagated into it.
2914 2017-10-23 Richard Biener <rguenther@suse.de>
2916 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2917 (bitmap_remove_expr_from_set): ... this. All callers call this
2918 for non-constant values.
2919 (bitmap_set_subtract): Rename to...
2920 (bitmap_set_subtract_expressions): ... this. Adjust and
2922 (bitmap_set_contains_value): Remove superfluous check.
2923 (bitmap_set_replace_value): Inline into single caller ...
2924 (bitmap_value_replace_in_set): ... here and simplify.
2925 (dependent_clean): Merge into ...
2926 (clean): ... this using an overload. Adjust.
2927 (prune_clobbered_mems): Adjust.
2928 (compute_antic_aux): Likewise.
2929 (compute_partial_antic_aux): Likewise.
2931 2017-10-23 Richard Biener <rguenther@suse.de>
2933 PR tree-optimization/82129
2935 2017-08-01 Richard Biener <rguenther@suse.de>
2937 PR tree-optimization/81181
2938 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2939 (compute_antic): ... end of iteration here.
2941 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2943 * target.def (starting_frame_offset): New hook.
2944 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2945 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2946 * doc/tm.texi.in: Regenerate.
2947 * hooks.h (hook_hwi_void_0): Declare.
2948 * hooks.c (hook_hwi_void_0): New function.
2949 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2950 STARTING_FRAME_OFFSET.
2951 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2952 * reload1.c (reload): Likewise.
2953 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2954 instead of STARTING_FRAME_OFFSET.
2955 * function.c (try_fit_stack_local): Likewise.
2956 (assign_stack_local_1): Likewise
2957 (instantiate_virtual_regs): Likewise.
2958 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2959 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2960 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2961 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2962 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2963 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2964 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2965 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2966 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2967 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2968 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2969 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2970 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2971 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2972 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2973 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2974 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2975 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2976 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2977 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2978 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2979 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2980 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2981 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2982 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2983 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2984 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2985 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2986 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2987 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2988 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2989 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2990 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2991 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2992 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2993 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2994 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2995 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2996 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2997 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2998 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2999 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3000 return a HOST_WIDE_INT.
3001 (avr_builtin_setjmp_frame_value): Use it instead of
3002 STARTING_FRAME_OFFSET.
3003 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3004 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3005 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3007 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3008 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3009 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3010 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3011 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3012 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3013 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3014 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3015 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3016 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3017 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3018 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3020 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3021 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3022 * config/mips/mips.c (mips_compute_frame_info): Refer to
3023 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3024 (mips_starting_frame_offset): New function.
3025 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3026 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3027 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3028 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3029 and return a HOST_WIDE_INT.
3030 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3031 (mmix_initial_elimination_offset): Refer to
3032 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3033 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3034 * config/pa/pa.c (pa_starting_frame_offset): New function.
3035 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3036 (pa_expand_prologue): Likewise.
3037 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3038 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3039 !FRAME_GROWS_DOWNWARD handling to...
3040 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3041 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3042 !FRAME_GROWS_DOWNWARD handling to...
3043 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3044 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3045 !FRAME_GROWS_DOWNWARD handling to...
3046 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3047 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3049 (rs6000_starting_frame_offset): New function.
3050 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3051 !FRAME_GROWS_DOWNWARD handling to...
3052 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3053 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3054 !FRAME_GROWS_DOWNWARD handling to...
3055 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3056 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3057 !FRAME_GROWS_DOWNWARD handling to...
3058 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3059 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3060 (rs6000_starting_frame_offset): New function.
3061 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3062 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3063 * config/vax/vax.c (vax_starting_frame_offset): New function.
3064 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3065 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3066 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3067 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3068 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3069 * system.h (STARTING_FRAME_OFFSET): Poison.
3071 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3073 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3074 SCALAR_TYPE_MODE instead of TYPE_MODE.
3076 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3077 Alan Hayward <alan.hayward@arm.com>
3078 David Sherwood <david.sherwood@arm.com>
3080 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_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 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3088 2017-10-23 Richard Biener <rguenther@suse.de>
3090 PR tree-optimization/82129
3091 * tree-ssa-pre.c (bitmap_set_and): Remove.
3092 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3093 canonicalizing expressions in the set to those with lowest
3094 ID rather than taking that from the first edge.
3096 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3098 * combine.c (rtx_equal_for_field_assignment_p): Use
3099 byte_lowpart_offset.
3101 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3102 Alan Hayward <alan.hayward@arm.com>
3103 David Sherwood <david.sherwood@arm.com>
3105 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3106 to a SUBREG_PROMOTED_VAR.
3108 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3109 Alan Hayward <alan.hayward@arm.com>
3110 David Sherwood <david.sherwood@arm.com>
3112 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3113 (expand_debug_source_expr): Likewise.
3114 * combine.c (combine_simplify_rtx): Likewise.
3115 * cse.c (fold_rtx): Likewise.
3116 * optabs.c (expand_float): Likewise.
3117 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3118 (simplify_binary_operation_1): Likewise.
3120 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3121 Alan Hayward <alan.hayward@arm.com>
3122 David Sherwood <david.sherwood@arm.com>
3124 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3125 (record_promoted_value): Likewise.
3126 * expr.c (expand_expr_real_2): Likewise.
3127 * ree.c (update_reg_equal_equiv_notes): Likewise.
3128 (combine_set_extension): Likewise.
3129 * rtlanal.c (low_bitmask_len): Likewise.
3130 * simplify-rtx.c (neg_const_int): Likewise.
3131 (simplify_binary_operation_1): Likewise.
3133 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3134 Alan Hayward <alan.hayward@arm.com>
3135 David Sherwood <david.sherwood@arm.com>
3137 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3138 * regcprop.c (maybe_mode_change): Likewise.
3139 * reload1.c (alter_reg): Likewise.
3141 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3143 * inchash.h (inchash::hash::add_wide_int): New function.
3144 * lto-streamer-out.c (hash_tree): Use it.
3146 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3148 * inchash.h (inchash::hash::add_wide_int): Rename to...
3149 (inchash::hash::add_hwi): ...this.
3150 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3151 (polymorphic_call_target_hasher::hash): Likewise.
3152 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3153 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3154 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3155 * lto-streamer-out.c (hash_tree): Likewise.
3156 * optc-save-gen.awk: Likewise.
3157 * tree.c (add_expr): Likewise.
3159 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3162 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3163 for ordered inequality comparisons even with TARGET_IEEE_FP.
3165 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3168 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3169 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3170 Expand with cmp<dwi>_doubleword.
3172 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3174 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3175 List CET intrinsics.
3176 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3177 specific to -fcf-protection option.
3179 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3181 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3182 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3183 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3184 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3185 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3186 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3187 (extra_objs): Add cet.o for Linux/x86 targets.
3188 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3189 * config/i386/cet.c: New file.
3190 * config/i386/cetintrin.h: Likewise.
3191 * config/i386/t-cet: Likewise.
3192 * config/i386/cpuid.h (bit_SHSTK): New.
3193 (bit_IBT): Likewise.
3194 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3195 pass IBT and SHSTK bits.
3196 * config/i386/i386-builtin-types.def
3197 (VOID_FTYPE_UNSIGNED_PVOID): New.
3198 (VOID_FTYPE_UINT64_PVOID): Likewise.
3199 * config/i386/i386-builtin.def: Add CET intrinsics.
3200 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3201 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3202 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3203 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3205 * config/i386/i386.c (rest_of_insert_endbranch): New.
3206 (pass_data_insert_endbranch): Likewise.
3207 (pass_insert_endbranch): Likewise.
3208 (make_pass_insert_endbranch): Likewise.
3209 (ix86_notrack_prefixed_insn_p): Likewise.
3210 (ix86_target_string): Add -mibt, -mshstk flags.
3211 (ix86_option_override_internal): Add flag_cf_protection
3213 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3214 (ix86_print_operand): Add 'notrack' prefix output.
3215 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3216 (ix86_expand_builtin): Expand CET intrinsics.
3217 (x86_output_mi_thunk): Add 'endbranch' instruction.
3218 * config/i386/i386.h (TARGET_IBT): New.
3219 (TARGET_IBT_P): Likewise.
3220 (TARGET_SHSTK): Likewise.
3221 (TARGET_SHSTK_P): Likewise.
3222 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3223 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3224 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3225 (builtin_setjmp_setup): New pattern.
3226 (builtin_longjmp): Likewise.
3227 (rdssp<mode>): Likewise.
3228 (incssp<mode>): Likewise.
3229 (saveprevssp): Likewise.
3230 (rstorssp): Likewise.
3231 (wrss<mode>): Likewise.
3232 (wruss<mode>): Likewise.
3233 (setssbsy): Likewise.
3234 (clrssbsy): Likewise.
3235 (nop_endbr): Likewise.
3236 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3238 * config/i386/immintrin.h: Include <cetintrin.h>.
3239 * config/i386/linux-common.h
3240 (file_end_indicate_exec_stack_and_cet): New prototype.
3241 (TARGET_ASM_FILE_END): New.
3243 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3245 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3246 latencies instead of having separate table; make difference between
3247 integer and float costs.
3248 * i386.h (processor_costs): Remove scalar_stmt_cost,
3249 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3250 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3252 * x86-tune-costs.h: Remove entries which has been removed in
3253 procesor_costs from all tables; make cond_taken_branch_cost
3254 and cond_not_taken_branch_cost COST_N_INSNS based.
3256 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3258 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3260 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3262 * config/i386/i386.md (isa): Remove fma_avx512f.
3263 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3264 <avx512>_fmadd_<mode>_mask3<round_name>,
3265 <avx512>_fmsub_<mode>_mask<round_name>,
3266 <avx512>_fmsub_<mode>_mask3<round_name>,
3267 <avx512>_fnmadd_<mode>_mask<round_name>,
3268 <avx512>_fnmadd_<mode>_mask3<round_name>,
3269 <avx512>_fnmsub_<mode>_mask<round_name>,
3270 <avx512>_fnmsub_<mode>_mask3<round_name>,
3271 <avx512>_fmaddsub_<mode>_mask<round_name>,
3272 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3273 <avx512>_fmsubadd_<mode>_mask<round_name>,
3274 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3275 (*vec_widen_umult_even_v16si<mask_name>,
3276 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3277 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3279 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3281 * extend.texi: Add 'nocf_check' documentation.
3282 * gimple.texi: Add second parameter to
3283 gimple_build_call_from_tree.
3284 * invoke.texi: Add -fcf-protection documentation.
3285 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3287 2017-10-20 Richard Biener <rguenther@suse.de>
3289 PR tree-optimization/82473
3290 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3291 the largest input type.
3293 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3295 * c-attribs.c (handle_nocf_check_attribute): New function.
3296 (c_common_attribute_table): Add 'nocf_check' handling.
3297 * gimple-parser.c: Add second argument NULL to
3298 gimple_build_call_from_tree.
3299 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3300 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3302 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3303 * common.opt: Add fcf-protection flag.
3304 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3305 * flag-types.h: Add enum cf_protection_level.
3306 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3307 Add 'nocf_check' attribute propagation to gimple call.
3308 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3309 (gimple_build_call_from_tree): Update prototype.
3310 (gimple_call_nocf_check_p): New function.
3311 (gimple_call_set_nocf_check): Likewise.
3312 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3313 * ipa-icf.c: Add nocf_check attribute in statement hash.
3314 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3315 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3316 * toplev.c (process_options): Add flag_cf_protection handling.
3318 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3320 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3322 2017-10-20 Richard Biener <rguenther@suse.de>
3324 PR tree-optimization/82603
3325 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3326 remove false predicated stores.
3328 2017-10-20 Richard Biener <rguenther@suse.de>
3330 * graphite-isl-ast-to-gimple.c
3331 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3332 Remove return value and simplify, dump copied stmt after lhs
3334 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3335 Reduce dump verbosity.
3336 (gsi_insert_earliest): Likewise.
3337 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3338 * graphite.c (print_global_statistics): Adjust dumping.
3339 (print_graphite_scop_statistics): Likewise.
3340 (print_graphite_statistics): Do not dump loops here.
3341 (graphite_transform_loops): But here.
3343 2017-10-20 Nicolas Roche <roche@adacore.com>
3345 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3346 * configure: Regenerate.
3348 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3351 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3352 functions when optimizing replace GIMPLE_RETURN stmts with
3353 calls to __builtin_unreachable ().
3356 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3357 for -fsanitize=thread link of executables.
3358 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3359 link of executables.
3362 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3364 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3365 define_insns for logical vector shifts to use VI248_AVX512BW
3366 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3367 condition, useless isa and prefix attributes. Change the first
3368 2 of these define_insns to ...
3369 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3370 define_insn for avx512vl.
3371 (<shift_insn><mode>3): ... and this, new define_insn without
3372 masking for non-avx512vl.
3375 * config/i386/sse.md (*andnot<mode>3,
3376 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3377 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3378 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3379 not applied use empty suffix even for TARGET_AVX512VL.
3380 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3381 is applied, supply evex,evex or evex,evex,evex instead of just
3384 2017-10-20 Julia Koval <julia.koval@intel.com>
3386 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3387 (OPTION_MASK_ISA_GFNI_UNSET): New.
3388 (ix86_handle_option): Handle OPT_mgfni.
3389 * config/i386/cpuid.h (bit_GFNI): New.
3390 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3391 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3392 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3393 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3394 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3395 * config/i386/i386.opt: Add mgfni.
3397 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3399 * config/msp430/msp430.c (msp430_option_override): Disable
3400 -fdelete-null-pointer-checks.
3401 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3403 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3405 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3406 of x87 and SSE instructions.
3408 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3410 * asan.c (create_cond_insert_point): Do not update edge count.
3411 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3412 (afdo_propagate_circuit): Likewise.
3413 (afdo_calculate_branch_prob): Likewise.
3414 (afdo_annotate_cfg): Likewise.
3415 * basic-block.h (struct edge_def): Remove count.
3416 (edge_def::count): New accessor.
3417 * bb-reorder.c (rotate_loop): Update.
3418 (find_traces_1_round): Update.
3419 (connect_traces): Update.
3420 (sanitize_hot_paths): Update.
3421 * cfg.c (unchecked_make_edge): Update.
3422 (make_single_succ_edge): Update.
3423 (check_bb_profile): Update.
3424 (dump_edge_info): Update.
3425 (update_bb_profile_for_threading): Update.
3426 (scale_bbs_frequencies_int): Update.
3427 (scale_bbs_frequencies_gcov_type): Update.
3428 (scale_bbs_frequencies_profile_count): Update.
3429 (scale_bbs_frequencies): Update.
3430 * cfganal.c (connect_infinite_loops_to_exit): Update.
3431 * cfgbuild.c (compute_outgoing_frequencies): Update.
3432 (find_many_sub_basic_blocks): Update.
3433 * cfgcleanup.c (try_forward_edges): Update.
3434 (try_crossjump_to_edge): Update
3435 * cfgexpand.c (expand_gimple_cond): Update
3436 (expand_gimple_tailcall): Update
3437 (construct_exit_block): Update
3438 * cfghooks.c (verify_flow_info): Update
3439 (redirect_edge_succ_nodup): Update
3440 (split_edge): Update
3441 (make_forwarder_block): Update
3442 (duplicate_block): Update
3443 (account_profile_record): Update
3444 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3445 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3446 * cfgloopmanip.c (scale_loop_profile): Update.
3448 (lv_adjust_loop_entry_edge): Update.
3449 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3450 (force_nonfallthru_and_redirect): Update.
3451 (purge_dead_edges): Update.
3452 (rtl_flow_call_edges_add): Update.
3453 * cgraphunit.c (init_lowered_empty_function): Update.
3454 (cgraph_node::expand_thunk): Update.
3455 * gimple-pretty-print.c (dump_probability): Update.
3456 (dump_edge_probability): Update.
3457 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3458 * haifa-sched.c (sched_create_recovery_edges): Update.
3459 * hsa-gen.c (convert_switch_statements): Update.
3460 * ifcvt.c (dead_or_predicable): Update.
3461 * ipa-inline-transform.c (inline_transform): Update.
3462 * ipa-split.c (split_function): Update.
3463 * ipa-utils.c (ipa_merge_profiles): Update.
3464 * loop-doloop.c (add_test): Update.
3465 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3466 * lto-streamer-in.c (input_cfg): Update.
3467 (input_function): Update.
3468 * lto-streamer-out.c (output_cfg): Update.
3469 * modulo-sched.c (sms_schedule): Update.
3470 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3471 * predict.c (maybe_hot_edge_p): Update.
3472 (unlikely_executed_edge_p): Update.
3473 (probably_never_executed_edge_p): Update.
3474 (dump_prediction): Update.
3475 (drop_profile): Update.
3476 (propagate_unlikely_bbs_forward): Update.
3477 (determine_unlikely_bbs): Update.
3478 (force_edge_cold): Update.
3479 * profile.c (compute_branch_probabilities): Update.
3480 * reg-stack.c (better_edge): Update.
3481 * shrink-wrap.c (handle_simple_exit): Update.
3482 * tracer.c (better_p): Update.
3483 * trans-mem.c (expand_transaction): Update.
3484 (split_bb_make_tm_edge): Update.
3485 * tree-call-cdce.c: Update.
3486 * tree-cfg.c (gimple_find_sub_bbs): Update.
3487 (gimple_split_edge): Update.
3488 (gimple_duplicate_sese_region): Update.
3489 (gimple_duplicate_sese_tail): Update.
3490 (gimple_flow_call_edges_add): Update.
3491 (insert_cond_bb): Update.
3492 (execute_fixup_cfg): Update.
3493 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3494 * tree-complex.c (expand_complex_div_wide): Update.
3495 * tree-eh.c (lower_resx): Update.
3496 (unsplit_eh): Update.
3497 (cleanup_empty_eh_move_lp): Update.
3498 * tree-inline.c (copy_edges_for_bb): Update.
3499 (freqs_to_counts): Update.
3500 (copy_cfg_body): Update.
3501 * tree-ssa-dce.c (remove_dead_stmt): Update.
3502 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3503 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3504 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3505 (unloop_loops): Update.
3506 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3507 * tree-ssa-loop-split.c (connect_loops): Update.
3508 (split_loop): Update.
3509 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3510 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3511 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3512 * tree-ssa-reassoc.c (branch_fixup): Update.
3513 * tree-ssa-tail-merge.c (replace_block_by): Update.
3514 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3515 (compute_path_counts): Update.
3516 (update_profile): Update.
3517 (recompute_probabilities): Update.
3518 (update_joiner_offpath_counts): Update.
3519 (estimated_freqs_path): Update.
3520 (freqs_to_counts_path): Update.
3521 (clear_counts_path): Update.
3522 (ssa_fix_duplicate_block_edges): Update.
3523 (duplicate_thread_path): Update.
3524 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3525 (case_bit_test_cmp): Update.
3526 (collect_switch_conv_info): Update.
3527 (gen_inbound_check): Update.
3528 (do_jump_if_equal): Update.
3529 (emit_cmp_and_jump_insns): Update.
3530 * tree-tailcall.c (decrease_profile): Update.
3531 (eliminate_tail_call): Update.
3532 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3533 (vect_do_peeling): Update.
3534 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3535 * ubsan.c (ubsan_expand_null_ifn): Update.
3536 (ubsan_expand_ptr_ifn): Update.
3537 * value-prof.c (gimple_divmod_fixed_value): Update.
3538 (gimple_mod_pow2): Update.
3539 (gimple_mod_subtract): Update.
3540 (gimple_ic): Update.
3541 (gimple_stringop_fixed_value): Update.
3543 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3546 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3548 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3550 PR rtl-optimization/82395
3551 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3552 based on non_spilled_static_chain_regno_p.
3554 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3556 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3557 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3558 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3560 2017-10-19 Martin Sebor <msebor@redhat.com>
3562 PR tree-optimization/82596
3563 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3565 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3567 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3568 (get_mem_refs_of_builtin_call): Likewise.
3569 * builtins.c (expand_builtin_apply): Adjust call to
3570 allocate_dynamic_stack_space.
3571 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3572 the third argument to allocate_dynamic_stack_space, otherwise -1.
3573 (expand_builtin): Deal with all alloca variants.
3574 (is_inexpensive_builtin): Likewise.
3575 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3576 * calls.c (special_function_p): Deal with all alloca variants.
3577 (initialize_argument_information): Adjust call to
3578 allocate_dynamic_stack_space.
3579 (expand_call): Likewise.
3580 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3581 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3582 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3583 use it for the stack usage computation.
3584 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3585 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3586 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3587 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3588 (in_loop_p): Remove first argument and useless check.
3589 (pass_walloca::execute): Remove useless test and adjust call to above.
3590 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3591 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3592 (gimplify_call_expr): Deal with all alloca variants.
3593 * hsa-gen.c (gen_hsa_alloca): Likewise.
3594 (gen_hsa_insns_for_call): Likewise.
3595 * ipa-pure-const.c (special_builtin_state): Likewise.
3596 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3597 * tree-object-size.c (alloc_object_size): Likewise.
3598 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3599 (call_may_clobber_ref_p_1): Likewise.
3600 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3601 (ccp_fold_stmt): Likewise.
3602 (optimize_stack_restore): Likewise.
3603 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3604 (mark_all_reaching_defs_necessary_1): Likewise.
3605 (propagate_necessity): Likewise.
3606 (eliminate_unnecessary_stmts): Likewise.
3607 * tree.c (build_common_builtin_nodes): Build
3608 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3609 (build_alloca_call_expr): New function.
3610 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3611 (CASE_BUILT_IN_ALLOCA): Likewise.
3612 (build_alloca_call_expr): Declare.
3613 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3615 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3618 * dwarf2out.c (new_die_raw): New static inline function.
3619 (new_die): Use it to create the DIE.
3620 (add_AT_external_die_ref): Likewise.
3621 (clone_die): Likewise.
3622 (clone_as_declaration): Likewise.
3623 (dwarf2out_vms_debug_main_pointer): Likewise.
3624 (base_type_die): Likewise. Remove early return for corner cases.
3625 Do not call add_pubtype on the DIE here.
3626 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3627 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3628 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3629 native order exists for base types, attach the DIE manually and call
3630 add_pubtype on it. Do not equate a reverse order DIE to the type.
3632 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3634 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3635 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3637 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3639 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3641 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3642 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3643 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3644 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3645 * opts.c (sanitizer_opts): Add builtin.
3646 * ubsan.c (instrument_builtin): New function.
3647 (pass_ubsan::execute): Call it.
3648 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3649 * doc/invoke.texi: Document -fsanitize=builtin.
3651 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3652 builtins, store max (log2 (align), 0) into uchar field instead of
3653 align into uptr field.
3654 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3655 store uchar 0 field instead of uptr 0 field.
3656 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3657 instead of passing one address of struct with 2 locations pass
3658 two addresses of structs with 1 location each.
3659 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3660 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3661 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3662 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3663 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3664 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3665 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3666 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3668 2017-10-19 Martin Liska <mliska@suse.cz>
3671 * file-find.c (remove_prefix): Remove.
3672 * file-find.h (remove_prefix): Likewise.
3673 * gcc-ar.c: Remove smartness of lookup.
3675 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3677 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3678 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3679 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3681 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3684 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3685 (setcc + and to xor + setcc): New peephole2.
3687 2017-10-19 Tom de Vries <tom@codesourcery.com>
3689 * doc/sourcebuild.texi (Test Directives, Variants of
3690 dg-require-support): Add dg-require-stack-size.
3692 2017-10-19 Martin Liska <mliska@suse.cz>
3695 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3696 that have a large alignment.
3697 (gimplify_target_expr): Likewise.
3699 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3701 PR rtl-optimization/82602
3702 * ira.c (rtx_moveable_p): Return false for volatile asm.
3704 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3707 * config/i386/i386-modes.def (CCGZ): New CC mode.
3708 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3709 * config/i386/predicates.md (ix86_comparison_operator):
3711 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3712 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3713 with double-word subtraction.
3714 (put_condition_code): Handle CCGZmode.
3716 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3718 * wide-int.cc (debug (const wide_int &)): New.
3719 (debug (const wide_int *)): New.
3720 (debug (const widest_int &)): New.
3721 (debug (const widest_int *)): New.
3723 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3726 * lra-constraints.c (curr_insn_transform): Use non-input operand
3727 instead of output one for matched reload.
3729 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3731 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3732 (tree-ssa-loop-ivopts.h): New header file.
3733 (struct builtin_info): New fields.
3734 (classify_builtin_1): Compute and record base and offset parts for
3735 memset builtin partition by calling strip_offset.
3736 (offset_cmp, fuse_memset_builtins): New functions.
3737 (finalize_partitions): Fuse adjacent memset partitions by calling
3739 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3740 Expose the interface.
3741 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3743 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3745 PR tree-optimization/82574
3746 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3747 that data reference must be executed exactly once per iteration
3748 against the outermost loop in nest.
3749 (classify_partition): Update call to above function.
3751 2017-10-18 Richard Biener <rguenther@suse.de>
3753 PR tree-optimization/82591
3754 * graphite.c (graphite_transform_loops): Move code gen message
3756 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3757 Here. Handle scop_to_isl_ast failing.
3758 (scop_to_isl_ast): Limit the number of ISL operations.
3760 2017-10-18 Richard Biener <rguenther@suse.de>
3762 * graphite-isl-ast-to-gimple.c
3763 (translate_isl_ast_to_gimple::set_rename): Simplify.
3764 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3765 (graphite_copy_stmts_from_block): ... here.
3766 (copy_bb_and_scalar_dependences): Simplify.
3767 (add_parameters_to_ivs_params): Canonicalize.
3768 (generate_entry_out_of_ssa_copies): Simplify.
3769 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3770 by passing in ISL dimension.
3771 (parameter_index_in_region_1): Rename to ...
3772 (parameter_index_in_region): ... this.
3773 (extract_affine): Adjust assert, pass down parameter index.
3774 (add_param_constraints): Use range-info when available.
3775 (build_scop_context): Adjust.
3776 * sese.c (new_sese_info): Adjust.
3777 (free_sese_info): Likewise.
3778 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3779 Remove unused typedefs.
3780 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3782 2017-10-18 Martin Liska <mliska@suse.cz>
3784 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3786 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3788 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3789 (s390_sched_init): Do not reset s390_sched_state if we entered the
3790 current basic block via a fallthru edge and all others are unlikely.
3792 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3794 * config/s390/s390.c (NUM_SIDES): New variable.
3795 (LONGRUNNING_THRESHOLD): New variable.
3796 (LATENCY_FACTOR): New variable.
3797 (s390_sched_score): Decrease score for long-running instructions on
3799 (s390_sched_variable_issue): Perform bookkeeping for long-running
3802 2017-10-18 Richard Biener <rguenther@suse.de>
3804 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3805 Simplify with removal of the parameter rename map.
3806 (set_rename): Likewise.
3807 (should_copy_to_new_region): Likewise.
3808 (graphite_copy_stmts_from_block): Likewise.
3809 (copy_bb_and_scalar_dependences): Remove initialization of
3810 unused copied_bb_map.
3812 (copy_internal_parameters): Likewise.
3813 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3814 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3815 Use INTEGRAL_TYPE_P.
3816 (parameter_index_in_region_1): Rename to ...
3817 (assign_parameter_index_in_region): ... this. Assert we have
3818 a parameter we handle.
3819 (scan_tree_for_params): Adjust.
3820 * sese.h (parameter_rename_map_t): Remove.
3821 (struct sese_info_t): Remove unused parameter_rename_map and
3822 copied_bb_map members.
3823 * sese.c (new_sese_info): Adjust.
3824 (free_sese_info): Likewise.
3826 2017-10-18 Martin Liska <mliska@suse.cz>
3829 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3830 on an abnormal edge.
3832 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3834 * doc/invoke.texi (ffunction-sections and fdata-sections):
3837 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3839 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3840 the use statement can throw internally.
3842 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3844 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3845 any RTX present on the RHS of a SET.
3846 * compare-elim.c (try_eliminate_compare): Restore comment.
3848 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3850 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3851 may be also called on tcc_type nodes.
3852 * langhooks.c (lhd_tree_size): Likewise.
3854 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3856 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3857 format_warning_at_substring.
3858 (maybe_warn): Convert source_range * param to a location_t. Pass
3859 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3860 (format_directive): Remove code to extract source_ranges and
3861 source_range * in favor of just a location_t.
3862 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3864 * substring-locations.c (format_warning_va): Convert
3865 source_range * param to a location_t.
3866 (format_warning_at_substring): Likewise.
3867 * substring-locations.h (format_warning_va): Likewise.
3868 (format_warning_at_substring): Likewise.
3870 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3872 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3874 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3875 and scatter/gather ops.
3877 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3878 vec_gather_load and vec_scatter_store.
3879 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3880 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3881 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3882 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3883 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3884 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3886 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3888 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3889 Detect FTST instruction and handle its register pops. Only pop
3890 second operand if can_pop_second_op is true.
3891 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3892 set can_pop_second_op to false in the compare_for_stack_reg call.
3894 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3895 output_fp_compare for stack register operands.
3896 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3897 instructions here. Do not emit stack register pops here. Assert
3898 that FCOMPP pops next to top stack register. Rewrite function.
3900 2017-10-17 Nathan Sidwell <nathan@acm.org>
3903 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3904 use DECL_ASSEMBLER_NAME_RAW.
3907 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3910 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3911 Wilco Dijkstra <wilco.dijkstra@arm.com>
3913 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3914 converstion to Pmode from the buf_addr.
3916 2017-10-17 Richard Biener <rguenther@suse.de>
3918 * graphite-dependences.c (scop_get_reads_and_writes): Change
3919 output parameters to references.
3921 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3923 PR 71026/tree-optimization
3924 * fold-const.c (distribute_real_division): Removed.
3925 (fold_binary_loc): Remove calls to distribute_real_divison.
3927 2017-10-17 Richard Biener <rguenther@suse.de>
3929 * graphite-scop-detection.c
3930 (scop_detection::stmt_has_simple_data_refs_p): Always use
3931 the full nest as region.
3932 (try_generate_gimple_bb): Likewise.
3933 * sese.c (scalar_evolution_in_region): Simplify now that
3934 SCEV can handle instantiation in regions.
3935 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3936 in the non-loop part of a function if requested.
3938 2017-10-17 Richard Biener <rguenther@suse.de>
3940 PR tree-optimization/82563
3941 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3943 (graphite_regenerate_ast_isl): Call it.
3944 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3946 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3948 PR tree-optimization/82549
3949 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3950 Formatting fixes. Instead of calling make_bit_field_ref with negative
3953 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3955 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3956 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3957 _mm_maskz_reduce_ss): New.
3958 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3959 __builtin_ia32_reducess_mask): Ditto..
3960 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3961 * config/i386/sse.md (reduces<mode>): Renamed to ...
3962 (reduces<mode><mask_scalar_name>): ... this.
3963 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3965 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3966 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3968 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3970 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3971 * selftest-run-tests.c (selftest::run_tests): Call
3972 selftest::unique_ptr_tests_cc_tests.
3973 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3974 * unique-ptr-tests.cc: New file.
3976 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3979 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3981 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3982 (make_hard_regno_born, make_hard_regno_dead): Update
3983 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3985 2017-10-16 Jeff Law <law@redhat.com>
3987 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3989 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3991 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3993 2017-10-16 Olivier Hainque <hainque@adacore.com>
3995 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3996 with_cpu if we were configured for an e500v2 target cpu name.
3998 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4000 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4001 * doc/invoke.texi: Document +nodsp as a valid extension for
4004 2017-10-16 Martin Liska <mliska@suse.cz>
4006 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4007 (test_set_range): Likewise.
4008 (test_range_functions): Rename to ...
4009 (test_bit_in_range): ... this.
4010 (sbitmap_c_tests): Add new test.
4012 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4014 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4016 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4017 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4019 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4021 * config/aarch64/aarch64-builtins.c
4022 (aarch64_types_quadopu_lane_qualifiers): New.
4023 (TYPES_QUADOPU_LANE): New.
4024 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4025 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4026 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4027 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4028 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4029 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4030 (Vdottype, DOTPROD): New.
4031 (sur): Add SDOT and UDOT.
4033 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4035 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4036 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4037 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4039 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4040 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4041 Enable TARGET_DOTPROD.
4042 (cortex-a75.cortex-a55): Likewise.
4043 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4045 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4047 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4048 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4050 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4052 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4053 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4054 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4055 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4056 * config/arm/types.md (neon_dot, neon_dot_q): New.
4057 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4059 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4061 * config/arm/arm.h (TARGET_DOTPROD): New.
4062 * config/arm/arm.c (arm_arch_dotprod): New.
4063 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4064 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4065 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4066 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4067 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4068 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4069 * doc/invoke.texi (armv8.2-a): Document dotprod
4071 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4073 * i386.c (ix86_vec_cost): New function.
4074 (ix86_rtx_costs): Handle vector operations better.
4075 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4076 * x86-tune-costs.h: Add new costs to all tables.
4078 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4080 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4082 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4083 divsd, sqrtss and sqrtsd
4084 * x86-tune-costs.h: Add new entries to all costs.
4085 (znver1_cost): Fix to match real instruction latencies.
4087 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4088 Michael Collison <michael.collison@arm.com>
4090 * compare-elim.c: Include emit-rtl.h.
4091 (can_merge_compare_into_arith): New function.
4092 (try_validate_parallel): Likewise.
4093 (try_merge_compare): Likewise.
4094 (try_eliminate_compare): Call the above when no previous clobber
4096 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4099 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4103 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4104 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4108 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4109 to be any operand_equal_p operands. For & (B - 1) require
4110 B to be power of 2. Recognize
4111 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4113 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4116 * optabs.c (expand_memory_blockage): Fix call of
4117 targetm.have_memory_blockage.
4119 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4122 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4123 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4124 objects to extra_objs instead of overwriting it.
4126 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4128 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4129 Use any_fp_register_operand as operand[3] predicate. Simplify
4130 equality test for operands[2] and operands[4] memory location.
4131 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4132 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4133 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4134 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4135 any_fp_register_operand as operand[1] predicate. Simplify
4136 equality test for operands[0] and operands[3] memory location.
4137 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4138 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4139 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4141 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4143 * target-insns.def: Add memory_blockage.
4144 * optabs.c (expand_memory_blockage): New function.
4145 (expand_asm_memory_barrier): Rename ...
4146 (expand_asm_memory_blockage): ... to this.
4147 (expand_mem_thread_fence): Call expand_memory_blockage
4148 instead of expand_asm_memory_barrier.
4149 (expand_mem_singnal_fence): Ditto.
4150 (expand_atomic_load): Ditto.
4151 (expand_atomic_store): Ditto.
4152 * doc/md.texi (Standard Pattern Names For Generation):
4153 Document memory_blockage instruction pattern.
4155 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4157 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4158 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4159 * config/rl78/rl78.md: New define_expand "adddi3".
4161 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4163 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4166 2017-10-13 Jeff Law <law@redhat.com>
4168 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4170 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4173 * internal-fn.c (expand_mul_overflow): If both operands have
4174 the same highpart of -1 or 0 and the topmost bit of lowpart
4175 is different, overflow is if res <= 0 rather than res < 0.
4177 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4179 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4180 TARGET_P9_VECTOR code for unaligned_load case.
4182 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4184 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4186 2017-10-13 Nathan Sidwell <nathan@acm.org>
4188 * tree-core.h (tree_contains_struct): Make bool.
4189 * tree.c (tree_contains_struct): Likewise.
4190 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4191 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4192 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4193 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4194 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4196 2017-10-13 Richard Biener <rguenther@suse.de>
4198 * graphite-isl-ast-to-gimple.c
4199 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4200 parameters and dominance check.
4201 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4202 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4203 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4204 Do not update SSA form here or do intermediate IL verification.
4205 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4206 (graphite_initialize): Remove check on the number of loops in
4207 the function and inline into graphite_transform_loops.
4208 (graphite_finalize): Inline into graphite_transform_loops.
4209 (graphite_transform_loops): Perform SSA update and IL verification
4211 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4213 2017-10-13 Richard Biener <rguenther@suse.de>
4215 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4216 graphite_expression_type_precision): Avoid global constructor
4218 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4219 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4220 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4221 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4222 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4223 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4225 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4228 * config/i386/i386.h (ix86_red_zone_size): New.
4229 * config/i386/i386.md (push peephole2s): Replace
4230 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4232 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4233 Alan Hayward <alan.hayward@arm.com>
4234 David Sherwood <david.sherwood@arm.com>
4236 * combine.c (can_change_dest_mode): Reject changes in
4237 REGMODE_NATURAL_SIZE.
4239 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4240 Alan Hayward <alan.hayward@arm.com>
4241 David Sherwood <david.sherwood@arm.com>
4243 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4244 (expand_debug_source_expr): Likewise.
4245 * combine.c (combine_simplify_rtx): Likewise.
4246 * cse.c (fold_rtx): Likewise.
4247 * fwprop.c (canonicalize_address): Likewise.
4248 * targhooks.c (default_shift_truncation_mask): Likewise.
4250 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4251 Alan Hayward <alan.hayward@arm.com>
4252 David Sherwood <david.sherwood@arm.com>
4254 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4255 (widened_mode): Likewise.
4256 (expand_unop): Likewise.
4257 * ree.c (transform_ifelse): Likewise.
4258 (merge_def_and_ext): Likewise.
4259 (combine_reaching_defs): Likewise.
4260 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4262 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4263 Alan Hayward <alan.hayward@arm.com>
4264 David Sherwood <david.sherwood@arm.com>
4266 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4267 * combine.c (gen_lowpart_for_combine): Likewise.
4268 * dwarf2out.c (rtl_for_decl_location): Likewise.
4269 * final.c (alter_subreg): Likewise.
4270 * rtlhooks.c (gen_lowpart_general): Likewise.
4271 (gen_lowpart_if_possible): Likewise.
4273 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4274 Alan Hayward <alan.hayward@arm.com>
4275 David Sherwood <david.sherwood@arm.com>
4277 * calls.c (expand_call): Use subreg_lowpart_offset.
4278 * cse.c (cse_insn): Likewise.
4279 * regcprop.c (copy_value): Likewise.
4280 (copyprop_hardreg_forward_1): Likewise.
4282 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4285 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4286 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4287 =Q constraints to +Q and into insn condition add check
4288 that operands[0] and operands[1] are equal.
4289 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4290 =Q constraints to +Q and into insn condition add check
4291 that operands[0] is equal to either operands[1] or operands[2].
4294 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4295 instead of handling MINUS_EXPR twice (once for each argument),
4296 canonicalize operand order and handle just once, use rtype where
4297 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4300 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4301 any values of __C while still being pattern recognizable as a simple
4304 2017-10-13 Richard Biener <rguenther@suse.de>
4306 PR tree-optimization/82451
4308 2017-10-02 Richard Biener <rguenther@suse.de>
4310 PR tree-optimization/82355
4311 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4312 a mapping for the enclosing loop but avoid generating one for
4314 (copy_bb_and_scalar_dependences): Remove premature codegen
4315 error on PHIs in blocks duplicated into multiple places.
4316 * graphite-scop-detection.c
4317 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4318 in the region use it as loop and nest to analyze the DR in.
4319 (try_generate_gimple_bb): Likewise.
4320 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4321 (add_loop_constraints): For blocks in a loop not in the region
4322 create a dimension with a single iteration.
4323 * sese.h (gbb_loop_at_index): Remove assert.
4325 * cfgloop.c (loop_preheader_edge): For the loop tree root
4326 return the single successor of the entry block.
4327 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4328 Reset the SCEV hashtable and niters.
4329 * graphite-scop-detection.c
4330 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4331 assert that we only have POLYNOMIAL_CHREC that vary in loops
4332 contained in the region.
4333 (scop_detection::graphite_can_represent_expr): Adjust.
4334 (scop_detection::stmt_has_simple_data_refs_p): For loops
4335 not in the region set loop to NULL. The nest is now the
4336 entry edge to the region.
4337 (try_generate_gimple_bb): Likewise.
4338 * sese.c (scalar_evolution_in_region): Adjust for
4339 instantiate_scev change.
4340 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4341 Make nest parameter the edge into the region.
4342 (create_data_ref): Likewise.
4343 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4344 entry edge into a region and adjust instantiate_scev calls.
4345 (create_data_ref): Likewise.
4346 (graphite_find_data_references_in_stmt): Likewise.
4347 (find_data_references_in_stmt): Pass the loop preheader edge
4348 from the nest argument.
4349 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4350 parameter the edge into the region.
4351 (instantiate_parameters): Use the loop preheader edge as entry.
4352 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4354 (get_instantiated_value_entry): Make instantiate_below parameter
4355 the edge into the region.
4356 (instantiate_scev_name): Likewise. Adjust dominance checks,
4357 when we cannot use loop-based instantiation instantiate by
4358 walking use-def chains.
4359 (instantiate_scev_poly): Adjust.
4360 (instantiate_scev_binary): Likewise.
4361 (instantiate_scev_convert): Likewise.
4362 (instantiate_scev_not): Likewise.
4363 (instantiate_array_ref): Remove.
4364 (instantiate_scev_3): Likewise.
4365 (instantiate_scev_2): Likewise.
4366 (instantiate_scev_1): Likewise.
4367 (instantiate_scev_r): Do not blindly handle N-operand trees.
4368 Do not instantiate array-refs. Handle all constants and invariants.
4369 (instantiate_scev): Make instantiate_below parameter
4370 the edge into the region.
4371 (resolve_mixers): Use the loop preheader edge for the region
4372 parameter to instantiate_scev_r.
4373 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4375 2017-10-13 Richard Biener <rguenther@suse.de>
4377 PR tree-optimization/82525
4378 * graphite-isl-ast-to-gimple.c
4379 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4381 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4382 Fail code generation when we cannot represent the isl integer.
4383 (binary_op_to_tree): Elide modulo operations that are no-ops
4384 in the type we code generate. Remove now superfluous code
4387 2017-10-13 Richard Biener <rguenther@suse.de>
4389 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4390 (scop_detection::harmful_loop_in_region): Remove premature
4391 IV type restriction.
4392 (scop_detection::graphite_can_represent_scev): We can handle
4393 pointer IVs just fine.
4395 2017-10-13 Alan Modra <amodra@gmail.com>
4397 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4398 "Clobbers and Scratch Registers". Add paragraph on
4399 alternative to clobbers for scratch registers and OpenBLAS
4402 2017-10-13 Alan Modra <amodra@gmail.com>
4404 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4405 example of a memory input for a string of known length. Move
4406 commentary out of table. Add a number of new examples
4407 covering array memory inputs.
4409 2017-10-12 Martin Liska <mliska@suse.cz>
4411 PR tree-optimization/82493
4412 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4413 (test_range_functions): New function.
4414 (sbitmap_c_tests): Likewise.
4415 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4416 * selftest.h (sbitmap_c_tests): New function.
4418 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4420 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4422 * config/rs6000/amo.h: Fix spacing issue.
4424 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4427 * config/i386/i386.md (*ashl<mode>3_mask_1,
4428 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4429 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4432 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4434 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4435 (profile_probability): Set max_probability
4436 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4438 * profile-count.c (profile_probability::differs_from_p): Do not
4439 rely on max_probaiblity == 10000
4441 2017-10-12 Jeff Law <law@redhat.com>
4443 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4446 2017-10-12 Martin Sebor <msebor@redhat.com>
4450 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4451 (handle_alias_pairs): Call it.
4452 * common.opt (-Wattribute-alias): New option.
4453 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4454 * doc/invoke.texi (-Wattribute-alias): Document.
4456 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4459 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4461 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4463 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4464 (make_hard_regno_born, make_hard_regno_dead): Update
4465 bb_killed_pseudos and bb_gen_pseudos.
4467 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4469 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4471 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4473 * config/alpha/alpha.c (alpha_split_conditional_move):
4474 Use std::swap instead of manually swapping.
4475 (alpha_stdarg_optimize_hook): Ditto.
4476 (alpha_canonicalize_comparison): Ditto.
4478 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4480 * tree-loop-distribution.c (struct builtin_info): New struct.
4481 (struct partition): Refactor fields into struct builtin_info.
4482 (partition_free): Free struct builtin_info.
4483 (build_size_arg_loc, build_addr_arg_loc): Delete.
4484 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4485 information from struct builtin_info.
4486 (find_single_drs): New function refactored from classify_partition.
4487 Also moved builtin validity checks to this function.
4488 (compute_access_range, alloc_builtin): New functions.
4489 (classify_builtin_st, classify_builtin_ldst): New functions.
4490 (classify_partition): Refactor code into functions find_single_drs,
4491 classify_builtin_st and classify_builtin_ldst.
4492 (distribute_loop): Don't do runtime alias check when distributing
4494 (find_seed_stmts_for_distribution): New function.
4495 (pass_loop_distribution::execute): Refactor code finding seed
4496 stmts into above function. Support distribution for the innermost
4497 two-level loop nest. Adjust dump information.
4499 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4501 * tree-loop-distribution.c: Adjust the general comment.
4502 (NUM_PARTITION_THRESHOLD): New macro.
4503 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4504 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4505 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4506 in call to build_partition_graph.
4507 (finalize_partitions): New parameter. Make loop distribution more
4508 conservative by fusing more partitions.
4509 (distribute_loop): Don't do runtime alias check in case of loop nest
4511 (find_seed_stmts_for_distribution): New function.
4512 (prepare_perfect_loop_nest): New function.
4513 (pass_loop_distribution::execute): Refactor code finding seed stmts
4514 and loop nest into above functions. Support loop nest distribution.
4515 Adjust dump information accordingly.
4517 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4519 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4520 and set PTYPE_SEQUENTIAL for merged partition.
4522 2017-10-12 Richard Biener <rguenther@suse.de>
4524 PR tree-optimization/69728
4526 2017-09-19 Richard Biener <rguenther@suse.de>
4528 PR tree-optimization/69728
4529 * graphite-sese-to-poly.c (schedule_error): New global.
4530 (add_loop_schedule): Handle empty domain by failing the
4532 (build_original_schedule): Handle schedule_error.
4534 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4535 domain by returning an unchanged schedule.
4537 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4539 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4540 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4542 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4544 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4547 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4550 * expr.c (store_field): Don't optimize away bitsize == 0 store
4551 from CALL_EXPR with addressable return type.
4553 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4555 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4556 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4557 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4558 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4559 TARGET_ISEL instead of TARGET_ISEL<sel>.
4561 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4563 * config/rs6000/rs6000.c
4564 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4566 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4568 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4570 (reg_or_cint_operand): Fix comment.
4571 (reg_or_zero_operand): New predicate.
4572 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4573 * config/rs6000/rs6000.c (output_isel): Delete.
4574 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4575 instead of reg_or_cint_operand. Output instruction directly (not via
4577 (isel_unsigned_<mode>): Ditto.
4578 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4579 gpc_reg_operand. Add an instruction alternative for this. Output
4580 instruction directly.
4581 (*isel_reversed_unsigned_<mode>): Ditto.
4583 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4585 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4586 (TARGET_CANONICALIZE_COMPARISON): Define.
4588 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4591 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4592 Check whether the dest is REG before adding REG_EQUIV note.
4594 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4597 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4599 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4600 (make_hard_regno_born, make_hard_regno_dead): Update
4601 bb_killed_pseudos and bb_gen_pseudos.
4603 2017-10-11 Nathan Sidwell <nathan@acm.org>
4605 * incpath.h (enum incpath_kind): Name enum, prefix values.
4606 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4607 * incpath.c (heads, tails): Use INC_MAX.
4608 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4609 (merge_include_chains, split_quote_chain,
4610 register_include_chains): Update incpath_kind names.
4611 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4612 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4614 (add_framework_path, darwin_register_objc_includes): Likewise.
4615 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4617 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4619 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4620 Do not use float_operator operator predicate.
4621 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4622 * config/i386/predicates.md (float_operator): Remove predicate.
4624 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4626 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4627 (*jccxf_i387): Ditto.
4628 (*jcc<mode>_i387): Ditto.
4629 (*jccu<mode>_i387): Ditto.
4630 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4631 (*jcc_*_i387 splitters): Remove.
4632 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4633 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4634 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4637 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4639 * profile-count.h (slow_safe_scale_64bit): New function.
4640 (safe_scale_64bit): New inline.
4641 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4642 * profile-count.c: Include wide-int.h
4643 (slow_safe_scale_64bit): New.
4645 2017-10-11 Nathan Sidwell <nathan@acm.org>
4647 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4648 HAS_DECL_ASSEMBLER_NAME_P.
4649 * gimple-expr.c (gimple_decl_printable_name: Check
4650 HAS_DECL_ASSEMBLER_NAME_P too.
4651 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4652 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4653 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4654 * passes.c (rest_of_decl_compilation): Check
4655 HAS_DECL_ASSEMBLER_NAME_P too.
4656 * recog.c (verify_changes): Likewise.
4657 * tree-pretty-print.c (dump_decl_name): Likewise.
4658 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4660 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4661 (SET_DECL_ASSEMBLER_NAME): Use it.
4662 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4663 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4664 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4666 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4668 * config.gcc (i386, x86_64): Add extra objects.
4669 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4670 (ix86_min_insn_size): Declare.
4671 (ix86_issue_rate): Declare.
4672 (ix86_adjust_cost): Declare.
4673 (ia32_multipass_dfa_lookahead): Declare.
4674 (ix86_macro_fusion_p): Declare.
4675 (ix86_macro_fusion_pair_p): Declare.
4676 (ix86_bd_has_dispatch): Declare.
4677 (ix86_bd_do_dispatch): Declare.
4678 (ix86_core2i7_init_hooks): Declare.
4679 (ix86_atom_sched_reorder): Declare.
4680 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4681 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4682 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4683 (rip_relative_addr_p): Rename to ...
4684 (ix86_rip_relative_addr_p): ... this one; export.
4685 (memory_address_length): Update.
4686 (ix86_issue_rate): Move to x86-tune-sched.c.
4687 (ix86_flags_dependent): Move to x86-tune-sched.c.
4688 (ix86_agi_dependent): Move to x86-tune-sched.c.
4689 (exact_dependency_1): Move to x86-tune-sched.c.
4690 (exact_store_load_dependency): Move to x86-tune-sched.c.
4691 (ix86_adjust_cost): Move to x86-tune-sched.c.
4692 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4693 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4694 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4695 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4696 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4697 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4698 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4699 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4700 (min_insn_size): Rename to ...
4701 (ix86_min_insn_size): ... this one; export.
4702 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4703 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4704 (core2i7_first_cycle_multipass_backtrack): Move to
4705 x86-tune-sched-core.c.
4706 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4707 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4708 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4709 (ix86_avoid_jump_mispredicts): Update.
4710 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4711 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4712 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4713 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4714 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4715 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4716 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4717 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4718 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4719 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4720 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4721 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4722 (BIG): Move to ix86-tune-sched-bd.c.
4723 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4724 (enum insn_path): Move to ix86-tune-sched-bd.c.
4725 (get_mem_group): Move to ix86-tune-sched-bd.c.
4726 (is_cmp): Move to ix86-tune-sched-bd.c.
4727 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4728 (is_branch): Move to ix86-tune-sched-bd.c.
4729 (is_prefetch): Move to ix86-tune-sched-bd.c.
4730 (init_window): Move to ix86-tune-sched-bd.c.
4731 (allocate_window): Move to ix86-tune-sched-bd.c.
4732 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4733 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4734 (process_end_window): Move to ix86-tune-sched-bd.c.
4735 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4736 (find_constant): Move to ix86-tune-sched-bd.c.
4737 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4738 (has_immediate): Move to ix86-tune-sched-bd.c.
4739 (get_insn_path): Move to ix86-tune-sched-bd.c.
4740 (get_insn_group): Move to ix86-tune-sched-bd.c.
4741 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4742 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4743 (add_insn_window): Move to ix86-tune-sched-bd.c.
4744 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4745 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4746 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4747 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4748 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4749 (do_dispatch): Move to ix86-tune-sched-bd.c.
4750 (has_dispatch): Move to ix86-tune-sched-bd.c.
4751 * config/i386/t-i386: Add new object files.
4752 * config/i386/x86-tune-costs.h: New file.
4753 * config/i386/x86-tune-sched-atom.c: New file.
4754 * config/i386/x86-tune-sched-bd.c: New file.
4755 * config/i386/x86-tune-sched-core.c: New file.
4756 * config/i386/x86-tune-sched.c: New file.
4758 2017-10-11 Liu Hao <lh_mouse@126.com>
4760 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4761 the generic version below instead.
4762 (should_colorize): Recognize Windows consoles as terminals
4764 * pretty-print.c [__MINGW32__] (write_all): New function.
4765 [__MINGW32__] (find_esc_head): Likewise.
4766 [__MINGW32__] (find_esc_terminator): Likewise.
4767 [__MINGW32__] (eat_esc_sequence): Likewise.
4768 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4770 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4773 2017-10-11 Richard Biener <rguenther@suse.de>
4775 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4776 Properly call analyze_scalar_evolution with the loop of the stmt.
4778 2017-10-11 Richard Biener <rguenther@suse.de>
4780 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4781 * tree-core.h (tree_base): Add chrec_var union member.
4782 * tree.h (CHREC_VAR): Remove.
4783 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4784 * tree-chrec.h (build_polynomial_chrec): Adjust.
4785 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4786 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4788 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4790 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4791 * match.pd: ... here.
4792 ((T) X == (T) Y): Relax condition.
4794 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4796 PR tree-optimization/82472
4797 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4799 (break_alias_scc_partitions): Update postorder number.
4801 2017-10-11 Martin Liska <mliska@suse.cz>
4804 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4806 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4808 2017-10-11 Martin Liska <mliska@suse.cz>
4810 * print-rtl.c (print_insn): Move declaration of idbuf
4811 to same scope as name.
4813 2017-10-11 Martin Liska <mliska@suse.cz>
4818 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4821 2017-10-11 Martin Liska <mliska@suse.cz>
4824 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4827 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4829 * config.gcc (default_gnu_indirect_function): Default to yes for
4830 arm*-*-linux* with glibc.
4832 2017-10-11 Richard Biener <rguenther@suse.de>
4834 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4835 default-defs and types we do not want to analyze.
4836 (interpret_loop_phi): Replace unreachable code with an assert.
4837 (compute_scalar_evolution_in_loop): Remove and inline ...
4838 (analyze_scalar_evolution_1): ... here, replacing condition with
4839 what makes the intent clearer. Remove handling of cases
4840 get_scalar_evolution now handles.
4842 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4844 PR rtl-optimization/81434
4845 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4846 comment for main loop. In sched_group_found if, also add checks for
4847 pass and min_cost_group.
4849 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4851 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4852 (rs6000_insn_cost): New function.
4853 * config/rs6000/rs6000.md (cost): New attribute.
4855 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4856 H.J. Lu <hongjiu.lu@intel.com>
4860 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4861 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4862 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4863 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4864 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4865 OPTION_MASK_ISA_MMX and some other bit set require both
4866 mmx and the ISAs without the mmx bit.
4867 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4868 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4869 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4870 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4871 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4872 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4873 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4874 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4875 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4876 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4877 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4878 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4879 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4880 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4881 Add OPTION_MASK_ISA_MMX.
4883 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4885 * config.gcc (armv7*-*-freebsd*): New target.
4886 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4888 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4890 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4891 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4894 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4896 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4897 a template parameter.
4898 (WIDE_INT_REF_FOR): Update accordingly.
4899 * tree.h (wi::int_traits <const_tree>): Delete.
4900 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4901 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4902 (wi::tree_to_wide_ref): New typedef.
4903 (wi::to_wide): New function.
4904 * calls.c (get_size_range): Use wi::to_wide when operating on
4906 * cgraph.c (cgraph_node::create_thunk): Likewise.
4907 * config/i386/i386.c (ix86_data_alignment): Likewise.
4908 (ix86_local_alignment): Likewise.
4909 * dbxout.c (stabstr_O): Likewise.
4910 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4911 * expr.c (const_vector_from_tree): Likewise.
4912 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4913 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4914 (fold_negate_expr_1, int_const_binop_1, const_binop)
4915 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4916 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4917 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4918 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4919 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4920 (fold_not_const, round_up_loc): Likewise.
4921 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4922 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4923 (alloca_call_type): Likewise.
4924 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4925 * godump.c (go_output_typedef): Likewise.
4926 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4927 * internal-fn.c (get_min_precision): Likewise.
4928 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4929 * ipa-polymorphic-call.c
4930 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4931 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4932 (ipa_modify_call_arguments): Likewise.
4933 * match.pd: Likewise.
4934 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4935 * print-tree.c (print_node_brief, print_node): Likewise.
4936 * stmt.c (expand_case): Likewise.
4937 * stor-layout.c (layout_type): Likewise.
4938 * tree-affine.c (tree_to_aff_combination): Likewise.
4939 * tree-cfg.c (group_case_labels_stmt): Likewise.
4940 * tree-data-ref.c (dr_analyze_indices): Likewise.
4941 (prune_runtime_alias_test_list): Likewise.
4942 * tree-dump.c (dequeue_and_dump): Likewise.
4943 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4944 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4945 * tree-pretty-print.c (dump_generic_node): Likewise.
4946 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4947 (simple_iv_with_niters): Likewise.
4948 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4949 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4950 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4951 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4952 (refine_value_range_using_guard, number_of_iterations_ne_max)
4953 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4954 (get_cst_init_from_scev, record_nonwrapping_iv)
4955 (scev_var_range_cant_overflow): Likewise.
4956 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4957 * tree-ssa-pre.c (compute_avail): Likewise.
4958 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4959 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4960 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4961 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4962 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4963 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4964 (dump_case_nodes, try_switch_expansion): Likewise.
4965 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4966 (vect_do_peeling): Likewise.
4967 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4968 * tree-vect-stmts.c (vectorizable_load): Likewise.
4969 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4970 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4971 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4972 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4973 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4974 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4975 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4976 (evrp_dom_walker::before_dom_children): Likewise.
4977 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4978 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4979 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4980 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4981 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4982 (get_range_pos_neg): Likewise.
4983 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4984 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4985 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4986 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4987 * config/avr/avr.c (avr_fold_builtin): Likewise.
4988 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4989 * config/msp430/msp430.c (msp430_attr): Likewise.
4990 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4991 * config/powerpcspe/powerpcspe-c.c
4992 (altivec_resolve_overloaded_builtin): Likewise.
4993 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4994 (rs6000_expand_ternop_builtin): Likewise.
4995 * config/rs6000/rs6000-c.c
4996 (altivec_resolve_overloaded_builtin): Likewise.
4997 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4998 (rs6000_expand_ternop_builtin): Likewise.
4999 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5001 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5003 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5004 when copying loop nest with only one inner loop.
5006 2017-10-10 Richard Biener <rguenther@suse.de>
5008 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5009 blocks if SCEV is active.
5010 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5012 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5013 (scev_initialize): Assert we are not yet initialized.
5015 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5017 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5018 inner loop's exit stmt by making it always exit the loop, otherwise
5019 we would generate an infinite empty loop.
5021 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5023 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5024 renaming variables in new preheader if it's deleted.
5026 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5028 * tree-loop-distribution.c (struct partition): Remove unused field
5029 loops of the structure.
5030 (partition_alloc, partition_free): Ditto.
5031 (build_rdg_partition_for_vertex): Ditto.
5033 2017-10-09 Jeff Law <law@redhat.com>
5035 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5036 return type to match prototype and documentation.
5038 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5040 * config/rs6000/rs6000.c (processor_costs): Move to ...
5041 * config/rs6000/rs6000.h: ... here.
5042 (rs6000_cost): Declare.
5044 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5046 * except.c (setjmp_fn): New global variable.
5047 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5048 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5049 if DONT_USE_BUILTIN_SETJMP is defined.
5051 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5053 * target.def (insn_cost): New hook.
5054 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5055 * doc/tm.texi: Regenerate.
5056 * rtlanal.c (insn_cost): Use the new hook.
5058 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5060 * combine.c (combine_validate_cost): Compute the new insn_cost,
5061 not just pattern_cost.
5062 (try_combine): Adjust comment.
5064 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5066 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5068 * combine.c (uid_insn_cost): Adjust comment.
5069 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5071 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5072 * dse.c (find_shift_sequence): Ditto.
5073 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5074 (bb_valid_for_noce_process_p): Use pattern_cost.
5075 * rtl.h (insn_rtx_cost): Delete.
5076 (pattern_cost): New prototype.
5077 (insn_cost): New prototype.
5078 * rtlanal.c (insn_rtx_cost): Rename to...
5079 (pattern_cost): ... this.
5082 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5084 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5085 (*jcc<mode>_0_r_i387): Ditto.
5086 (*jccxf_r_i387): Ditto.
5087 (*jcc<mode>_r_i387): Ditto.
5088 (*jccu<mode>_r_i387): Ditto.
5089 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5090 (*jcc): Rename from *jcc_1.
5092 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5094 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5095 deferred rescans after the lvx/stvx recombination pre-pass.
5097 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5099 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5100 memory operation instruction support.
5101 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5102 (rs6000-ibm-aix[789]*): Likewise.
5103 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5104 Document new functions.
5106 2017-10-09 Richard Biener <rguenther@suse.de>
5108 PR tree-optimization/82397
5109 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5110 equality only for semantically equal trees.
5112 2017-10-09 Richard Biener <rguenther@suse.de>
5114 PR tree-optimization/82449
5115 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5116 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5117 allow constant addresses.
5118 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5121 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5123 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5126 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5129 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5132 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5135 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5137 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5140 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5141 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5143 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5145 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5146 (WI_BINARY_PREDICATE_RESULT): Likewise.
5147 (wi::binary_traits::operator_result): New type.
5148 (wi::binary_traits::predicate_result): Likewise.
5149 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5150 (generic_wide_int::operator==, generic_wide_int::operator!=)
5151 (generic_wide_int::operator&, generic_wide_int::and_not)
5152 (generic_wide_int::operator|, generic_wide_int::or_not)
5153 (generic_wide_int::operator^, generic_wide_int::operator+
5154 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5156 (operator~, unary operator-, operator==, operator!=, operator&)
5157 (operator|, operator^, operator+, binary operator-, operator*): New
5159 * expr.c (get_inner_reference): Use wi::bit_and_not.
5160 * fold-const.c (fold_binary_loc): Likewise.
5161 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5162 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5163 (bit_value_binop): Likewise.
5164 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5165 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5166 (extract_range_from_binary_expr_1): Likewise.
5167 (masked_increment): Likewise.
5168 (simplify_bit_ops_using_ranges): Likewise.
5170 2017-10-09 Martin Jambor <mjambor@suse.cz>
5173 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5174 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5175 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5177 (hsa_fixup_mov_insn_type): New function.
5178 (hsa_op_with_type::get_in_type): Use it.
5179 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5180 immediates in an assert.
5181 (hsa_op_with_type::extend_int_to_32bit): New method.
5182 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5183 types. Convert to dest type if necessary.
5184 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5185 (reg_for_gimple_ssa): Pass false as min32int to
5186 hsa_type_for_scalar_tree_type.
5187 (gen_hsa_addr): Fixup type when creating addresable temporary.
5188 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5189 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5190 necessary. Call hsa_fixup_mov_insn_type.
5191 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5192 extend operands and convert to dest type if necessary.
5193 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5194 to dest type if necessary.
5195 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5196 if conversion nt necessary and size matches.
5197 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5198 to dest type if necessary.
5199 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5200 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5202 (gen_hsa_clrsb): Likewise.
5203 (gen_hsa_ffs): Likewise.
5204 (gen_hsa_divmod): Extend operands and convert to dest type if
5206 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5208 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5210 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5211 Remove empty default arguments. Use a brace block as output
5213 (conditional return): Ditto.
5215 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5216 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5218 (group_ending_nop): Ditto.
5219 (doloop_end): Ditto.
5220 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5221 (splitters for those): Ditto.
5223 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5225 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5226 a conditional jump (and the compare for it) so that pc_rtx is the
5228 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5229 for the deleted and renamed ctr<mode>_internal[234] patterns.
5230 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5231 Delete second conditional return pattern.
5232 (ctr<mode>_internal2): Delete this second bdnz pattern.
5233 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5234 (ctr<mode>_internal4): Delete this second bdz pattern.
5236 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5238 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5239 (always_initialized_rtx_for_ssa_name_p): New predicate.
5240 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5241 (finish_out_of_ssa): Free new field of SA.
5242 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5243 * tree-ssa-coalesce.c: Include tree-ssa.h.
5244 (get_parm_default_def_partitions): Remove extern keyword.
5245 (get_undefined_value_partitions): New function.
5246 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5247 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5249 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5251 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5253 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5255 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5257 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5258 for targets that preffer 128bit.
5260 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5262 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5264 2017-10-08 Olivier Hainque <hainque@adacore.com>
5266 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5267 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5269 (thumb_set_return_address): Likewise.
5271 2017-10-08 Olivier Hainque <hainque@adacore.com>
5273 * common/config/arm/arm-common.c (arm_except_unwind_info):
5274 Handle DWARF2_UNWIND_INFO.
5276 2017-10-07 Michael Collison <michael.collison@arm.com>
5278 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5281 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5283 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5284 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5285 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5286 defined, force the creation of a new block for a dispatch label.
5288 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5290 * invoke.texi (Wsuggest-attribute=cold): Document.
5291 * common.opt (Wsuggest-attribute=cold): New
5292 * ipa-pure-const.c (warn_function_cold): New function.
5293 * predict.c (compute_function_frequency): Use it.
5294 * predict.h (warn_function_cold): Declare.
5296 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5298 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5301 2017-10-06 Martin Liska <mliska@suse.cz>
5303 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5304 keyword for member functions.
5305 (struct sanopt_tree_couple): New struct.
5306 (struct sanopt_tree_couple_hash): New function.
5307 (struct sanopt_ctx): Add new hash_map.
5308 (has_dominating_ubsan_ptr_check): New function.
5309 (record_ubsan_ptr_check_stmt): Likewise.
5310 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5311 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5312 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5314 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5317 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5318 aarch64_simd_valid_immediate on CONST_VECTORs.
5319 (aarch64_reg_or_bic_imm): Likewise.
5321 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5323 PR rtl-optimization/82396
5324 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5326 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5328 * graphite-dependences.c (scop_get_reads): Move code to...
5329 (scop_get_must_writes): Move code to...
5330 (scop_get_may_writes): Move code to...
5331 (scop_get_reads_and_writes): ... here.
5332 (scop_get_dependences): Call scop_get_reads_and_writes.
5334 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5336 PR tree-optimization/82434
5337 * fold-const.h (can_native_encode_type_p,
5338 can_native_encode_string_p): Remove.
5339 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5340 don't encode anything, just return what would be otherwise returned.
5341 (native_encode_fixed, native_encode_complex, native_encode_vector):
5343 (native_encode_string): Likewise. Inline by hand
5344 can_native_encode_string_p.
5345 (can_native_encode_type_p): Remove.
5346 (can_native_encode_string_p): Remove.
5347 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5348 STRING_CSTs using can_native_encode_string_p, test all
5349 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5350 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5351 argument from native_encode_expr.
5352 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5353 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5356 2017-10-06 Richard Biener <rguenther@suse.de>
5358 PR tree-optimization/82397
5359 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5360 operand_equal_p but rely on data_ref_compare_tree for detecting
5362 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5363 to match up with dr_group_sort_cmp.
5365 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5368 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5370 * config/s390/s390-builtin-types.def: Regenerate.
5372 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5375 * config/s390/s390-builtin-types.def: Regenerate.
5376 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5377 Change flag from B_VXE to B_VX.
5378 (s390_vec_min_dbl): Remove B_VXE flag.
5380 2017-10-06 Richard Biener <rguenther@suse.de>
5382 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5383 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5384 translate_isl_ast_to_gimple::is_valid_rename,
5385 translate_isl_ast_to_gimple::get_rename,
5386 translate_isl_ast_to_gimple::get_def_bb_for_const,
5387 translate_isl_ast_to_gimple::get_new_name,
5388 translate_isl_ast_to_gimple::collect_all_ssa_names,
5389 translate_isl_ast_to_gimple::copy_loop_phi_args,
5390 translate_isl_ast_to_gimple::collect_all_ssa_names,
5391 translate_isl_ast_to_gimple::copy_loop_phi_args,
5392 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5393 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5394 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5395 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5396 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5397 translate_isl_ast_to_gimple::copy_cond_phi_args,
5398 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5399 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5400 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5401 translate_isl_ast_to_gimple::rename_uses,
5402 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5403 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5404 (set_rename_for_each_def): Likewise.
5405 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5406 here. Handle rewriting SCEV analyzable uses here.
5407 (copy_bb_and_scalar_dependences): Generate code for PHI
5409 (graphite_regenerate_ast_isl): Adjust.
5410 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5411 (add_write, add_read): New functions.
5412 (build_cross_bb_scalars_def): Use it and simplify.
5413 (build_cross_bb_scalars_use): Likewise.
5414 (graphite_find_cross_bb_scalar_vars): Inline into...
5415 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5416 simulating out-of-SSA. Compute liveout and add dependencies.
5417 (build_scops): Force an empty entry block.
5418 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5420 (sese_build_liveouts): Declare.
5421 (sese_trivially_empty_bb_p): Likewise.
5422 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5423 compute liveout and debug_liveout.
5424 (sese_bad_liveouts_use): Remove.
5425 (sese_reset_debug_liveouts_bb): Likewise.
5426 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5427 (sese_build_liveouts): Build liveout and debug_liveout and store
5429 (new_sese_info): Adjust.
5430 (free_sese_info): Likewise.
5431 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5432 do not build liveout here.
5433 (move_sese_in_condition): Adjust region entry.
5434 (scev_analyzable_p): Match up with chrec_apply requirements.
5435 (sese_trivially_empty_bb_p): New.
5436 * tree-into-ssa.c (get_reaching_def): Properly support generating
5437 default-defs for incremental rewrite of anonymous names.
5439 2017-10-06 Richard Biener <rguenther@suse.de>
5441 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5442 precision do not perform modulo reduction.
5444 2017-10-06 Richard Biener <rguenther@suse.de>
5446 PR tree-optimization/82436
5447 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5448 conservatively choose the vectorization factor when checking
5449 whether we can perform the required load permutation.
5450 (vect_transform_slp_perm_load): Assert when we may not fail.
5452 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5454 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5455 message for incompatible -msdata=* and -mcall-* options.
5457 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5459 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5460 rate for post-reload scheduling.
5462 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5464 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5466 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5468 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5469 to improve monte carlo in scimark.
5471 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5473 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5474 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5475 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5476 pentium4_cost, nocona_cost): Set reassociation width to 1.
5477 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5478 width to 2 for fp operations and 1 otherwise.
5479 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5481 (atom_cost): Set reassociation width to 2.
5482 (slm_cost, generic_cost): Set fp reassociation width
5483 to 2 and 1 otherwise.
5484 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5485 (core_cost): Set fp reassociation width to 4 and vector to 2.
5486 (ix86_reassociation_width): Rewrite using cost table; special case
5487 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5488 and TARGET_AVX128_OPTIMAL.
5489 * config/i386/i386.h (processor_costs): Add
5490 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5491 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5492 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5493 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5494 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5495 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5497 2017-10-05 Nathan Sidwell <nathan@acm.org>
5499 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5501 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5503 * config/arm/arm.c (arm_test_fpu_data): New.
5504 (arm_run_selftests): Call arm_test_fpu_data.
5506 2017-10-04 Nathan Sidwell <nathan@acm.org>
5508 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5510 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5512 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5514 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5515 check type for aarch64_simd_valid_immediate.
5516 (aarch64_output_simd_mov_immediate): Update prototype.
5517 (aarch64_simd_valid_immediate): Update prototype.
5518 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5519 support for ORR-immediate.
5520 (and<mode>3): modified pattern to add support for BIC-immediate.
5521 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5522 now checks for valid immediate for BIC and ORR based on new enum
5524 (aarch64_output_simd_mov_immediate): Function now used to output
5525 BIC/ORR imm as well based on new enum argument.
5526 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5528 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5529 (aarch64_reg_or_bic_imm): Likewise.
5531 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5533 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5534 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5536 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5540 PR rtl-optimization/82396
5541 * haifa-sched.c (autopref_multipass_init): Simplify
5543 (autopref_rank_data): Simplify sort order.
5544 * sched-int.h (autopref_multipass_data_): Remove
5545 multi_mem_insn_p, min_offset and max_offset.
5547 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5549 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5551 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5553 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5556 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5558 * doc/sourcebuild.texi: Document vect_long_mult.
5560 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5562 PR tree-optimization/82413
5563 * fold-const.c (build_range_check): Use widest_int when comparing
5564 the maximum ETYPE value with HIGH.
5566 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5568 PR rtl-optimization/82396
5569 * haifa-sched.c (autopref_multipass_init): Simplify
5571 (autopref_rank_data): Simplify sort order.
5572 * sched-int.h (autopref_multipass_data_): Remove
5573 multi_mem_insn_p, min_offset and max_offset.
5575 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5577 PR tree-optimization/82381
5578 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5579 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5582 PR tree-optimization/82374
5583 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5584 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5585 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5586 current_function_decl to the new decl.
5588 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5590 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5591 helper macro for IEEE float128 hardware built-in functions.
5592 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5594 (TRUNCF128_ODD): Likewise.
5595 (ADDF128_ODD): Likewise.
5596 (SUBF128_ODD): Likewise.
5597 (MULF128_ODD): Likewise.
5598 (DIVF128_ODD): Likewise.
5599 (FMAF128_ODD): Likewise.
5600 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5601 UNSPEC_TRUNC_ROUND_TO_ODD.
5602 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5603 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5604 floating point round to odd instructions.
5605 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5606 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5607 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5608 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5609 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5610 (trunc<mode>sf2_hw): Change the truncate with round to odd
5611 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5612 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5613 to odd hardware instructions.
5614 (sub<mode>3_odd): Likewise.
5615 (mul<mode>3_odd): Likewise.
5616 (div<mode>3_odd): Likewise.
5617 (sqrt<mode>2_odd): Likewise.
5618 (fma<mode>4_odd): Likewise.
5619 (fms<mode>4_odd): Likewise.
5620 (nfma<mode>4_odd): Likewise.
5621 (nfms<mode>4_odd): Likewise.
5622 (trunc<mode>df2_odd): Change the truncate with round to odd
5623 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5625 * doc/extend.texi (PowerPC built-in functions): Update documentation
5626 for existing IEEE float128-bit built-in functions. Add built-in
5627 functions that generate the IEEE 128-bit floating point round to
5630 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5632 PR rtl-optimization/77729
5633 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5634 to (X&(C1&~C2))|C2 transformations.
5636 2017-10-03 Martin Jambor <mjambor@suse.cz>
5638 PR tree-optimization/82363
5639 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5640 mismatch, mark lacc written regardless of racc.
5642 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5644 PR tree-optimization/82381
5645 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5646 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5647 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5648 fallthrough into reassoc_stmt_dominates_stmt_p.
5651 * combine.c (combine_instructions): Don't combine in unreachable
5654 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5657 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5658 function to not use the have_cpu variable. Do not set cpu_index,
5659 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5661 (rs6000_valid_attribute_p): Remove duplicate initializations of
5662 old_optimize and func_optimize.
5663 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5664 (rs6000_activate_target_options): Make global.
5665 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5668 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5670 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5671 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5672 Set *poffset to 0 and *psize and *pmax_size to -1 if
5673 *poffset + *psize overflows in HOST_WIDE_INT.
5675 PR tree-optimization/82387
5676 PR tree-optimization/82388
5677 PR tree-optimization/82389
5678 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5679 instead of live_bytes non-NULL.
5681 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5684 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5687 2017-10-02 Richard Biener <rguenther@suse.de>
5689 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5690 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5691 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5693 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5695 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5696 requested precision matches the type's.
5697 * calls.c (alloc_max_size): Calculate the new candidate size as
5698 a widest_int and use wi::to_widest when comparing it with the
5699 current candidate size.
5700 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5701 zero rather than integer_zero_node.
5702 * match.pd: Check for a no-op conversion before using wi::add
5703 rather than after. Use tree_to_uhwi when summing small shift
5704 counts into an unsigned int.
5706 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5707 Alan Hayward <alan.hayward@arm.com>
5708 David Sherwood <david.sherwood@arm.com>
5711 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5712 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5713 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5714 POINTER_AND_FP_REGS.
5716 2017-10-02 Richard Biener <rguenther@suse.de>
5718 PR tree-optimization/82355
5719 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5720 a mapping for the enclosing loop but avoid generating one for
5722 (copy_bb_and_scalar_dependences): Remove premature codegen
5723 error on PHIs in blocks duplicated into multiple places.
5724 * graphite-scop-detection.c
5725 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5726 in the region use it as loop and nest to analyze the DR in.
5727 (try_generate_gimple_bb): Likewise.
5728 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5729 (add_loop_constraints): For blocks in a loop not in the region
5730 create a dimension with a single iteration.
5731 * sese.h (gbb_loop_at_index): Remove assert.
5733 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5735 * omp-expand.c (adjust_context_scope): New function.
5736 (expand_parallel_call): Call adjust_context_scope.
5738 2017-10-01 Jeff Law <law@redhat.com>
5740 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5741 dom_opt_dom_walker class with direct access to private members.
5742 Add comments. Call test_for_singularity.
5743 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5744 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5745 m_dummy_cond anymore.
5746 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5748 (pass_dominator:execute): Build the dummy_cond here and pass it
5749 to the dom_opt_dom_walker ctor.
5750 (test_for_singularity): New function.
5752 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5753 Maya Rashish <coypu@sdf.org>
5755 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5756 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5757 (alpha*-*-netbsd*) Use nbsd_tm_file.
5758 (arm*-*-netbsdelf*) Likewise.
5759 (i[34567]86-*-netbsdelf*) Likewise.
5760 (x86_64-*-netbsd*) Likewise.
5761 (mips*-*-netbsd*) Likewise.
5762 (powerpc-*-netbsd*) Likewise.
5763 (sh*-*-netbsd*) Likewise.
5764 (sparc-*-netbsdelf*) Likewise.
5765 (sparc64-*-netbsd*) Likewise.
5766 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5768 (vax-*-netbsdelf*) Likewise.
5769 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5770 (UINT_FAST8_TYPE) Likewise.
5771 (INT_FAST16_TYPE) Check CHAR_FAST16.
5772 (UINT_FAST16_TYPE) Likewise.
5774 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5777 * config/i386/i386.md
5778 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5779 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5780 *divmodsi4_zext_2): New define_insn_and_split.
5781 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5782 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5783 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5784 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5785 New define_insn_and_split.
5786 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5787 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5788 operands[1] having DImode when mode is SImode.
5790 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5791 always SImode for DIV and MOD in REG_EQUAL notes.
5793 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5796 * match.pd: Fix handling of NaNs in pattern.
5798 2017-09-29 Jeff Law <law@redhat.com>
5800 * sbitmap.c (bitmap_bit_in_range_p): New function.
5801 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5802 * tree-ssa-dse.c (live_bytes_read): New function.
5803 (dse_classify_store): Ignore reads of dead bytes.
5805 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5806 typos and whitespace errors.
5807 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5808 * config/s390/s390.c (s390_emit_prologue): Likewise.
5810 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5813 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5814 with a symbol for LRA.
5816 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5818 PR rtl-optimization/82338
5819 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5821 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5823 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5824 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5825 (qsort_chk): Declare.
5826 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5827 (qsort_chk): New function.
5829 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5831 PR tree-optimization/82337
5832 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5833 phi definition if the PHI result appears in an abnormal PHI.
5834 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5835 basis appears in an abnormal PHI.
5837 2017-09-29 Richard Biener <rguenther@suse.de>
5839 * graphite-isl-ast-to-gimple.c
5840 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5841 (binary_op_to_tree): Use it.
5842 (get_rename_from_scev): Likewise.
5843 (copy_loop_phi_nodes): Likewise.
5844 (copy_bb_and_scalar_dependences): Likewise.
5845 (translate_pending_phi_nodes): Likewise.
5847 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5850 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5851 for movabsq $(i32 << shift), r64.
5853 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5855 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5856 index when encoding %esp as %rsp to avoid 0x67 prefix.
5858 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5860 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5861 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5863 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5865 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5866 Extensions with more than 16 double VFP registers.
5867 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5868 to_clear_mask and all code related to it. Replace the remaining
5869 entry by a sbitmap and adapt code accordingly.
5871 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5873 * brig-builtins.def: Change pure attributes to const.
5875 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5877 * config.gcc (default_gnu_indirect_function): Default to yes for
5878 sparc*-*-linux* with glibc.
5880 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5882 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5883 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5884 when creating .init_array and .fini_array sections with priority
5887 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5890 * config/aarch64/aarch64.c
5891 (aarch64_builtin_support_vector_misalignment): Always return false
5892 when misalignment is unknown.
5894 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5896 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5897 this function to return false if the definition used by the swap
5898 instruction is artificial, or if the memory address from which the
5899 constant value is loaded is not represented by a base address held
5900 in a register or if the base address register is a frame or stack
5901 pointer. Additionally, return false if the base address of the
5902 loaded constant is a SYMBOL_REF but is not considered to be a
5904 (replace_swapped_load_constant): New function.
5905 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5906 loaded constant vector with a load of a swapped constant vector.
5908 2017-09-27 Carl Love <cel@us.ibm.com>
5910 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5911 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5912 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5915 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5917 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5918 first, always call autopref_rank_data otherwise.
5920 2017-09-27 Richard Biener <rguenther@suse.de>
5922 * graphite-scop-detection.c (find_scop_parameters): Move
5923 loop bound handling ...
5924 (gather_bbs::before_dom_children): ... here, avoiding the need
5925 to build scop_info->loop_nest.
5926 (record_loop_in_sese): Remove.
5927 * sese.h (sese_info_t::loop_nest): Remove.
5928 * sese.c (new_sese_info): Do not allocate loop_nest.
5929 (free_sese_info): Do not free loop_nest.
5931 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5934 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5935 lhs from calls if the lhs has addressable type.
5937 2017-09-27 Richard Biener <rguenther@suse.de>
5939 * graphite.h (scop::max_alias_set): New member.
5940 * graphite-scop-detection.c: Remove references to non-existing
5941 --param in comments.
5942 (build_alias_sets): Record the maximum alias set used for drs.
5943 (build_scops): Support zero as unlimited for
5944 --param graphite-max-arrays-per-scop.
5945 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5947 (build_poly_sr_1): ... here. Compute alias set based on the
5948 maximum alias set used for drs rather than
5949 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5951 2017-09-27 Richard Biener <rguenther@suse.de>
5953 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5954 --param loop-block-tile-size=0 to disable tiling.
5956 2017-09-27 Richard Biener <rguenther@suse.de>
5958 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5959 (graphite-max-nb-scop-params): Document special value zero.
5960 * domwalk.h (dom_walker::STOP): New symbolical constant.
5961 (dom_walker::dom_walker): Add optional parameter for bb to
5963 (dom_walker::~dom_walker): Declare.
5964 (dom_walker::before_dom_children): Document STOP return value.
5965 (dom_walker::m_user_bb_to_rpo): New member.
5966 (dom_walker::m_bb_to_rpo): Likewise.
5967 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5968 mapping here if not provided by the user.
5969 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5970 provided by the user.
5971 (dom_walker::STOP): Define.
5972 (dom_walker::walk): Do not compute bb to RPO mapping here.
5973 Support STOP return value from before_dom_children to stop
5975 * graphite-optimize-isl.c (optimize_isl): If the schedule
5976 is the same still generate code if -fgraphite-identity
5977 or -floop-parallelize-all are given.
5978 * graphite-scop-detection.c: Include cfganal.h.
5979 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5981 (gather_bbs::before_dom_children): Return STOP for BBs
5983 (build_scops): Compute bb to RPO mapping and pass it to
5984 the domwalk. Treat --param graphite-max-nb-scop-params=0
5985 as not limiting the number of params.
5986 * graphite.c (graphite_initialize): Remove limit on the
5987 number of basic-blocks in a function.
5988 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5989 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5990 default value of 10.
5992 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5994 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5995 Adjust code to eliminate needing to do the shift right 32-bits
5996 operation after XSCVDPSPN.
5998 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6000 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6001 ((X / Y) != 0 -> X >= Y): Likewise.
6003 2017-09-26 Carl Love <cel@us.ibm.com>
6005 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6006 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6007 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6008 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6009 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6010 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6011 definitions and overloading.
6012 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6013 statement for P9V_BUILTIN_XST_LEN_R.
6014 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6015 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6016 define_expand and define_insn for the instructions and builtins.
6017 * doc/extend.texi: Update the built-in documentation file for the new
6019 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6020 define_insn for the instructions
6022 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6025 * gcc/config/netbsd-protos.h: New file.
6026 * gcc/config/netbsd.c: New file.
6027 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6028 * gcc/config/t-netbsd: New file.
6029 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6030 (tmake_file) Add t-netbsd.
6031 (extra_objs) Add netbsd.o.
6033 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6037 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6039 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6041 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6042 sign extension from a vector register to a GPR by doing a 32-bit
6043 direct move and then an EXTSW.
6044 (extendsi<mode>2 splitter): Likewise.
6045 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6046 right or vector extract after doing XSCVDPSPN. Use
6047 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6049 (movdi_from_sf_zero_ext): Likewise.
6050 (reload_gpr_from_vsxsf): Likewise.
6051 (p8_mfvsrd_4_disf): Delete, no longer used.
6052 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6053 then needing to move the SFmode to a GPR to use the XSCVDPSP
6054 instruction instead of FRSP and XSCVDPSPN.
6055 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6056 it is adjacent to the other XSCVSPDP insns.
6057 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6058 SFmode to be in traditional Altivec registers.
6059 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6060 (vsx_xscvspdpn): Likewise.
6061 (vsx_xscvspdpn_scalar): Likewise.
6063 2017-09-26 Martin Jambor <mjambor@suse.cz>
6065 * tree-sra.c (compare_access_positions): Put integral types first,
6066 stabilize sorting of integral types, remove conditions putting
6067 non-full-precision integers last.
6068 (sort_and_splice_var_accesses): Disable scalarization if a
6069 non-integert would be represented by a non-full-precision integer.
6071 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6073 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6074 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6075 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6076 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6077 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6078 conditionals inside the function instead of around it. Call
6079 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6080 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6082 2017-09-26 Richard Biener <rguenther@suse.de>
6084 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6086 (scop_detection::build_scop_breadth): ... this. Removed.
6087 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6088 (scop_detection::harmful_stmt_in_bb): Likewise.
6089 (scop_detection::graphite_can_represent_stmt): Likewise.
6090 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6091 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6092 (scop_detection::can_represent_loop_1): ... this. Removed.
6093 (scop_detection::harmful_loop_in_region): Simplify after inlining
6094 the above and remove more quadraticness.
6095 (build_scops): Adjust.
6096 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6099 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6102 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6103 REGNO (base) == SP_REG if base is a REG.
6106 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6107 if it is different SSA_NAME.
6108 (optimize_range_tests_cmp_bitwise): New function.
6109 (optimize_range_tests): Call it.
6111 2017-09-26 Richard Biener <rguenther@suse.de>
6113 PR tree-optimization/82321
6114 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6115 for the def being inside the loop.
6117 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6119 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6121 * config/s390/s390-builtins.def: Fix constraint on op4.
6123 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6125 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6126 independent expanders.
6127 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6128 ("vec_ordered", "vec_unordered"): New expanders.
6130 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6132 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6135 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6137 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6138 vec_unpacks_lo_v16qi.
6139 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6141 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6143 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6144 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6145 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6147 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6149 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6151 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6153 ("*vec_slb<mode>"): New insn pattern.
6154 ("vec_shr_<mode>"): New expander.
6155 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6156 and force the shift count operand to V16QImode.
6157 ("vec_srb<mode>"): Set shift count mode to V16QI.
6159 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6161 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6162 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6163 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6165 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6168 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6170 2017-09-26 Richard Biener <rguenther@suse.de>
6172 PR tree-optimization/82320
6173 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6176 2017-09-25 Jeff Law <law@redhat.com>
6178 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6179 prototype for new argument.
6180 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6181 mostly extracted from rs6000_emit_allocate_stack.
6182 (rs6000_emit_probe_stack_range_stack_clash): New function.
6183 (rs6000_emit_allocate_stack): Call
6184 rs6000_emit_probe_stack_range_stack_clash as needed.
6185 (rs6000_emit_probe_stack_range): Add additional argument
6186 to call to gen_probe_stack_range{si,di}.
6187 (output_probe_stack_range): New.
6188 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6189 (output_probe_stack_range_stack_clash): New.
6190 (rs6000_emit_prologue): Emit notes into dump file as requested.
6191 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6192 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6193 Add additional operand and pass it to output_probe_stack_range.
6195 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6197 PR tree-optimization/82163
6198 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6199 (checking_verify_loop_closed_ssa): New parameter.
6200 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6201 (check_loop_closed_ssa_stmt): Delete.
6202 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6203 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6204 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6207 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6209 * brig-builtins.def: Treat HSAIL barrier builtins as
6210 setjmp/longjump style functions.
6212 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6214 * target.def (constant_alignment): New hook.
6215 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6216 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6217 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6218 * doc/tm.texi: Regenerate.
6219 * targhooks.h (default_constant_alignment): Declare.
6220 (constant_alignment_word_strings): Likewise.
6221 * targhooks.c (default_constant_alignment): New function.
6222 (constant_alignment_word_strings): Likewise.
6223 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6224 instead of CONSTANT_ALIGNMENT.
6225 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6226 (force_const_mem): Likewise.
6227 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6228 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6229 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6230 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6231 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6233 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6234 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6235 constant_alignment_word_strings.
6236 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6237 (CONSTANT_ALIGNMENT): Likewise.
6238 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6239 (arm_constant_alignment): New function.
6240 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6241 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6242 constant_alignment_word_strings.
6243 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6244 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6245 constant_alignment_word_strings.
6246 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6247 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6248 (cris_constant_alignment): New function.
6249 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6250 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6251 (epiphany_constant_alignment): New function.
6252 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6253 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6254 constant_alignment_word_strings.
6255 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6256 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6257 constant_alignment_word_strings.
6258 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6259 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6260 constant_alignment_word_strings.
6261 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6262 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6263 * config/i386/i386.c (ix86_constant_alignment): Make static.
6264 Use the same interface as the target hook.
6265 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6266 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6267 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6268 constant_alignment_word_strings.
6269 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6270 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6271 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6272 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6273 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6274 constant_alignment_word_strings.
6275 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6276 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6277 constant_alignment_word_strings.
6278 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6279 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6280 constant_alignment_word_strings.
6281 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6282 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6284 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6285 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6286 * config/mips/mips.c (mips_constant_alignment): New function.
6287 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6288 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6289 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6290 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6291 (mmix_constant_alignment): Make static. Use the same interface
6293 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6294 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6295 constant_alignment_word_strings.
6296 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6297 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6298 constant_alignment_word_strings.
6299 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6300 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6301 constant_alignment_word_strings.
6302 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6303 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6304 (rs6000_constant_alignment): New function.
6305 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6306 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6307 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6308 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6309 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6310 (rs6000_constant_alignment): New function.
6311 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6312 * config/s390/s390.c (s390_constant_alignment): New function.
6313 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6314 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6315 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6316 constant_alignment_word_strings.
6317 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6318 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6319 (sparc_constant_alignment): New function.
6320 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6321 * config/spu/spu.c (spu_constant_alignment): New function.
6322 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6323 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6324 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6325 constant_alignment_word_strings.
6326 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6327 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6328 constant_alignment_word_strings.
6329 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6330 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6331 constant_alignment_word_strings.
6332 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6333 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6334 (visium_constant_alignment): New function.
6335 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6336 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6337 (xtensa_constant_alignment): New function.
6338 * system.h (CONSTANT_ALIGNMENT): Poison.
6340 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6342 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6343 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6344 (rs6000_builtin_valid_without_lhs): New helper function.
6345 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6346 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6348 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6350 * target.h (vec_perm_indices): Use unsigned short rather than
6352 (auto_vec_perm_indices): Likewise.
6353 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6354 Use unsigned int rather than unsigned char.
6355 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6357 2017-09-25 Richard Biener <rguenther@suse.de>
6359 * cfgloop.h (sort_sibling_loops): Declare.
6360 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6361 (sort_sibling_loops): New function sorting the sibling loop list
6363 * graphite.c (graphite_transform_loops): Sort sibling loops.
6365 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6367 * target.def (vec_perm_const_ok): Change sel parameter to
6369 * optabs-query.c (can_vec_perm_p): Update accordingly.
6370 * doc/tm.texi: Regenerate.
6371 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6372 auto_vec_perm_indices and remove separate nelt field.
6373 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6374 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6375 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6376 (aarch64_expand_vec_perm_const): Update accordingly.
6377 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6378 to vec_perm_indices.
6379 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6380 auto_vec_perm_indices and remove separate nelt field.
6381 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6382 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6383 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6385 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6386 to vec_perm_indices.
6387 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6388 sel to vec_perm_indices.
6389 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6390 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6391 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6393 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6396 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6399 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6400 on the FUNCTION_DECL function context if it has a DIE that is a
6403 2017-09-25 Richard Biener <rguenther@suse.de>
6405 PR tree-optimization/82285
6406 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6409 2017-09-25 Tom de Vries <tom@codesourcery.com>
6413 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6416 2017-09-25 Richard Biener <rguenther@suse.de>
6418 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6419 ISL errors other than isl_error_quota happen. Dump if the
6420 schedule is the same.
6421 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6422 errors instead of aborting inside ISL.
6424 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6427 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6428 of libgcc_eh for m64.
6429 * config/i386/darwin64.h: Likewise.
6431 2017-09-25 Richard Biener <rguenther@suse.de>
6434 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6435 attribute for incomplete types nor twice for complete ones.
6437 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6440 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6441 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6444 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6447 * config/i386/i386.opt (mprefer-avx256): Use
6448 ix86_target_flags variable.
6449 * config/i386/i386.c (ix86_target_string): Move
6450 -mprefer-avx256 to flag2_opts.
6452 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6455 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6456 and x != -1 | y != -1 into (x & y) != -1.
6458 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6460 * config.gcc: Add new case statement to set
6461 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6462 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6463 s390x-*-linux* case statements. Added aarch64 to the list of
6464 supported architectures.
6466 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6468 PR tree-optimization/82289
6469 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6470 STMT_VINFO_RELEVANT_P.
6472 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6473 Alan Hayward <alan.hayward@arm.com>
6474 David Sherwood <david.sherwood@arm.com>
6476 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6477 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6478 (extract_range_from_binary_expr_1): Don't call
6479 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6481 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6482 Alan Hayward <alan.hayward@arm.com>
6483 David Sherwood <david.sherwood@arm.com>
6485 * target.def (preferred_vector_alignment): New hook.
6486 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6488 * doc/tm.texi: Regenerate.
6489 * targhooks.h (default_preferred_vector_alignment): Declare.
6490 * targhooks.c (default_preferred_vector_alignment): New function.
6491 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6493 (DR_TARGET_ALIGNMENT): New macro.
6494 (aligned_access_p): Update commentary.
6495 (vect_known_alignment_in_bytes): New function.
6496 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6498 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6499 Calculate the misalignment based on the target alignment rather than
6501 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6502 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6503 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6504 the target alignment, rather than masking the element misalignment
6505 with the number of elements in a vector. Also use the target
6506 alignment when calculating the maximum number of peels.
6507 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6508 instead of TYPE_ALIGN_UNIT.
6509 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6510 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6511 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6512 (vect_create_data_ref_ptr): Likewise.
6513 (vect_setup_realignment): Realign by ANDing with
6514 -DR_TARGET_MISALIGNMENT.
6515 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6516 the number of peels based on DR_TARGET_ALIGNMENT.
6517 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6518 with the guaranteed alignment boundary when deciding whether
6520 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6521 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6522 (ensure_base_align): Remove stmt_info parameter. Get the
6523 target base alignment from DR_TARGET_ALIGNMENT.
6524 (vectorizable_store): Update call accordingly. Interpret
6525 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6527 (vectorizable_load): Likewise.
6529 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6530 Alan Hayward <alan.hayward@arm.com>
6531 David Sherwood <david.sherwood@arm.com>
6533 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6534 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6535 (vect_enhance_data_refs_alignment): Likewise.
6537 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6539 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6540 error. Only quit immediately if parsing is complete.
6541 (BEGIN): Initialize fatal_err and parse_done.
6542 (begin fpu, end fpu): Check number of arguments.
6543 (begin arch, end arch): Likewise.
6544 (begin cpu, end cpu): Likewise.
6545 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6546 (optalias): Likewise.
6548 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6550 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6551 * config/arm/arm-isa.h: Delete. Move definitions to ...
6552 * arm-cpus.in: ... here. Use new feature and fgroup values.
6553 * config/arm/arm.c (arm_option_override): Use lower case for feature
6555 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6556 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6557 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6559 (print_isa_bits_for): New function.
6560 (gen_isa): New function.
6561 (gen_comm_data): Use print_isa_bits_for.
6562 (define feature): New keyword.
6563 (define fgroup): New keyword.
6564 * config/arm/t-arm (TM_H): Remove.
6565 (GTM_H): Add arm-isa.h.
6566 (arm-isa.h): Add rule to generate file.
6567 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6568 case for feature bit names.
6570 2017-09-22 Richard Biener <rguenther@suse.de>
6572 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6574 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6575 print of no dependency loops ...
6576 * graphite.c (graphite_transform_loops): ... here.
6577 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6579 (same_close_phi_node, remove_duplicate_close_phi,
6580 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6581 (canonicalize_loop_closed_ssa): ... here and simplify.
6582 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6583 (optimize_isl): Use dump_printf_loc to tell when we stopped
6584 optimizing because of an ISL timeout.
6586 2017-09-22 Richard Biener <rguenther@suse.de>
6588 PR tree-optimization/82291
6589 * tree-if-conv.c (predicate_mem_writes): Make sure to
6590 remove writes in blocks predicated with false.
6592 2017-09-22 Richard Biener <rguenther@suse.de>
6594 * sese.c: Include cfganal.h.
6595 (if_region_set_false_region): Remove.
6596 (create_if_region_on_edge): Likewise.
6597 (move_sese_in_condition): Re-implement without destroying
6600 2017-09-22 Richard Biener <rguenther@suse.de>
6602 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6603 Verify both BBs contain loop PHI nodes before dispatching to
6605 (graphite_regenerate_ast_isl): Do not recompute dominators,
6606 do not verify three times. Restructure for clarity.
6607 * graphite-scop-detection.c (same_close_phi_node,
6608 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6609 defined_in_loop_p, canonicalize_loop_closed_ssa,
6610 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6611 checking and SSA rewrite, move to ...
6612 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6613 (graphite_initialize): Do not pass in ctx, do not reset the
6614 SCEV cache, compute only dominators.
6615 (graphite_transform_loops): Allocate ISL ctx after
6616 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6617 Maintain post-dominators only around build_scops.
6618 * sese.c (if_region_set_false_region): Make static. Free
6619 and recompute dominators.
6620 (move_sese_in_condition): Assert we don't get called with
6621 post-dominators computed.
6622 * sese.h (if_region_set_false_region): Remove.
6624 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6626 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6627 mode attribute for TARGET_AVX512VL.
6629 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6631 * config/i386/i386.opt (mprefer-avx256): New option.
6632 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6634 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6635 for TARGET_PREFER_AVX256.
6636 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6638 2017-09-21 Jeff Law <law@redhat.com>
6640 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6641 Fix dump output if the only stack space is for pushed registers.
6643 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6645 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6648 2017-09-21 Martin Sebor <msebor@redhat.com>
6651 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6652 code (in C++) or code that triggers warnings.
6654 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6656 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6658 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6660 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6661 * sched-rgn.c: Ditto.
6662 * sel-sched-ir.c: Ditto.
6664 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6666 * toplev.h (set_random_seed): Adjust return type.
6667 * toplev.c (init_local_tick): Move eager initialization of random_seed
6668 to get_random_seed. Adjust comment.
6669 (init_random_seed): Inline to get_random_seed, delete.
6670 (get_random_seed): Initialize random_seed lazily.
6671 (set_random_seed): Do not return previous value.
6672 (print_switch_value): Do not call get_random_seed.
6674 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6676 * cgraph.c (delete_function_version): New, broken out from...
6677 (cgraph_node::delete_function_version): ...here. Rename to
6678 cgraph_node::delete_function_version_by_decl. Update all uses.
6679 (cgraph_node::remove): Call delete_function_version.
6681 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6684 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6685 VAR_DECLs to which addressable non-volatile parameters are mapped
6686 and for id->retvar after the return value assignment. Clear
6687 id->retval and id->retbnd after inlining.
6689 2017-09-21 Richard Biener <rguenther@suse.de>
6691 PR tree-optimization/82276
6692 PR tree-optimization/82244
6693 * tree-vrp.c (build_assert_expr_for): Set
6694 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6696 (remove_range_assertions): Revert earlier change.
6698 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6701 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6703 2017-09-21 Richard Biener <rguenther@suse.de>
6705 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6706 Restore valid IL after code generation errors.
6707 * graphite.c (graphite_transform_loops): Diagnose code
6708 generation issues as MSG_MISSED_OPTIMIZATION and continue
6709 with processing SCOPs.
6711 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6712 Alan Hayward <alan.hayward@arm.com>
6713 David Sherwood <david.sherwood@arm.com>
6715 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6716 rather than choosing between plus_constant and gen_rtx_<CODE>.
6717 * expr.c (emit_push_insn): Likewise.
6718 (expand_expr_real_2): Likewise.
6720 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6721 Alan Hayward <alan.hayward@arm.com>
6722 David Sherwood <david.sherwood@arm.com>
6724 * loop-unroll.c (split_iv): Call copy_rtx on the step.
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 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6731 calling tree_to_uhwi.
6733 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6734 Alan Hayward <alan.hayward@arm.com>
6735 David Sherwood <david.sherwood@arm.com>
6737 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6738 INTEGER_CST rather than a negative test for ADDR_EXPR.
6740 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6741 Alan Hayward <alan.hayward@arm.com>
6742 David Sherwood <david.sherwood@arm.com>
6744 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6745 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6747 2017-09-21 Richard Biener <rguenther@suse.de>
6749 PR tree-optimization/71351
6750 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6751 graphite_create_new_loop_guard): Remove, fold remaining parts
6753 (translate_isl_ast_node_for): ... here and simplify.
6755 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6758 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6759 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6760 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6761 alternative always use QI mode, for -Os imov (=R,R) alternative
6762 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6765 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6766 Jeff Law <law@redhat.com>
6768 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6769 (allocate_stack_space): New function, partially extracted from
6771 (s390_emit_prologue): Track offset to most recent stack probe.
6772 Code to allocate space moved into allocate_stack_space.
6773 Dump actions when no stack is allocated.
6774 (s390_prologue_plus_offset): New function.
6775 (s390_emit_stack_probe): Likewise.
6777 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6779 * common.opt (Wa, Wl, Wp, g, gz=): Add
6781 (gno-column-info): Remove.
6782 (gcolumn-info): Drop RejectNegative.
6784 (gno-record-gcc-switches): Remove.
6785 (grecord-gcc-switches): Drop RejectNegative.
6786 (gno-split-dwarf): Remove.
6787 (gsplit-dwarf): Drop RejectNegative.
6788 (gno-strict-dwarf): Remove.
6789 (gstrict-dwarf): Drop RejectNegative.
6790 * config/darwin.opt (gfull, gused): Add RejectNegative.
6791 * dwarf2out.c (gen_producer_string): Drop
6792 gno-record-gcc-switches handler.
6793 * optc-gen.awk: Add g to prefixes with negative forms.
6794 * opts-common.c (remapping_prefix_p): New.
6795 (find_opt): Check it.
6796 (generate_canonical_option): Test g prefix.
6797 (option_map): Add -gno- mapping.
6798 (add_misspelling_candidates): Check remapping_prefix_p.
6800 2017-09-20 Jeff Law <law@redhat.com>
6802 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6803 thinko in stack clash protection support.
6805 * explow.c (compute_stack_clash_protection_loop_data): Use
6806 CONST_INT_P instead of explicit test. Verify object is a
6807 CONST_INT_P before looking at INTVAL.
6808 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6809 instead of explicit test.
6811 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6814 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6815 address instead of to r1 and r11.
6817 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6819 * config.gcc: Support "knm".
6820 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6821 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6823 * config/i386/i386.c (m_KNM): Define.
6824 (processor_target_table): Add "knm".
6826 (ix86_option_override_internal): Add "knm".
6827 (ix86_issue_rate): Add PROCESSOR_KNM.
6828 (ix86_adjust_cost): Ditto.
6829 (ia32_multipass_dfa_lookahead): Ditto.
6830 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6831 (fold_builtin_cpu): Add M_INTEL_KNM.
6832 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6833 (processor_type): Add PROCESSOR_KNM.
6834 * config/i386/x86-tune.def: Add m_KNM.
6835 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6837 2017-09-20 Richard Biener <rguenther@suse.de>
6839 PR tree-optimization/80213
6840 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6841 are allowed in empty BBs as well.
6842 (canonicalize_loop_closed_ssa): Also look for other complex
6844 (scop_detection::get_sese): Include the loop-closed PHI block
6846 (scop_detection::merge_sese): Remove code adding extra blocks.
6847 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6848 (build_scops): Assert the final returned scop is invalid.
6850 2017-09-20 Richard Biener <rguenther@suse.de>
6852 PR tree-optimization/82264
6853 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6855 (vn_phi_lookup): Likewise.
6856 (vn_phi_insert): Likewise.
6858 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6860 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6861 that fits into uhwi or shwi, add DW_AT_const_value regardless
6862 of early_dwarf without going through RTL, using add_AT_unsigned
6865 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6866 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6867 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6868 (DEBUG_MACRO_SECTION): Likewise.
6869 (DEBUG_LTO_MACRO_SECTION): Likewise.
6870 (DEBUG_STR_DWO_SECTION): Likewise.
6871 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6872 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6873 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6874 (DEBUG_LTO_LINE_STR_SECTION): Define.
6875 (init_sections_and_labels): Initialize debug_line_str_section
6876 variable. Initialize debug_loc_section for -gdwarf-5 to
6877 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6879 2017-09-20 Richard Biener <rguenther@suse.de>
6881 * graphite-sese-to-poly.c (extract_affine): Properly handle
6882 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6884 2017-09-20 Richard Biener <rguenther@suse.de>
6886 PR tree-optimization/81373
6887 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6888 Force SESE live-out defs to be handled even if they are
6891 2017-09-19 Jeff Law <law@redhat.com>
6893 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6894 nothing for stack adjustments with REG_STACK_CHECK.
6895 * sched-deps.c (parse_add_or_inc): Reject insns with
6896 REG_STACK_CHECK from dependency breaking.
6897 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6898 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6899 * reg-notes.def (STACK_CHECK): New note.
6901 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6902 (ix86_expand_prologue): Dump stack clash info as needed.
6903 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6905 * function.c (dump_stack_clash_frame_info): New function.
6906 * function.h (dump_stack_clash_frame_info): Prototype.
6907 (enum stack_clash_probes): New enum.
6909 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6910 flag_stack_clash_protection.
6911 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6912 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6913 (arm_frame_pointer_required): Likewise.
6914 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6915 (ia64_expand_prologue): Likewise.
6916 * config/mips/mips.c (mips_expand_prologue): Likewise.
6917 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6918 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6919 (sparc_flat_expand_prologue): Likewise.
6920 * config/spu/spu.c (spu_expand_prologue): Likewise.
6922 * explow.c: Include "params.h".
6923 (anti_adjust_stack_and_probe_stack_clash): New function.
6924 (get_stack_check_protect): Likewise.
6925 (compute_stack_clash_protection_loop_data): Likewise.
6926 (emit_stack_clash_protection_loop_start): Likewise.
6927 (emit_stack_clash_protection_loop_end): Likewise.
6928 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6929 Use anti_adjust_stack_and_probe_stack_clash.
6930 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6931 (emit_stack_clash_protection_loop_start): Likewise.
6932 (emit_stack_clash_protection_loop_end): Likewise.
6933 * rtl.h (get_stack_check_protect): Prototype.
6934 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6935 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6936 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6938 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6940 * doc/tm.texi: Rebuilt.
6941 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6942 get_stack_check_protect.
6943 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6944 * config/arm/arm.c (arm_expand_prologue): Likewise.
6945 (arm_frame_pointer_required): Likewise.
6946 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6947 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6948 * config/mips/mips.c (mips_expand_prologue): Likewise.
6949 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6950 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6951 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6952 (sparc_flat_expand_prologue): Likewise.
6954 * common.opt (-fstack-clash-protection): New option.
6955 * flag-types.h (enum stack_check_type): Note difference between
6956 -fstack-check= and -fstack-clash-protection.
6957 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6958 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6959 * toplev.c (process_options): Issue warnings/errors for cases
6960 not handled with -fstack-clash-protection.
6961 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6962 (-fstack-check): Note additional problem with -fstack-check=generic.
6963 Note that -fstack-check is primarily for Ada and refer users
6964 to -fstack-clash-protection for stack-clash-protection.
6965 Document new params for stack clash protection.
6967 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6969 * config/i386/i386.c (ix86_split_long_move): Do not handle
6970 address used for LEA in a special way.
6972 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6974 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6976 2017-09-19 Martin Sebor <msebor@redhat.com>
6979 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6980 of incompatible types.
6982 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6984 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6985 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6986 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6987 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6989 2017-09-19 Richard Biener <rguenther@suse.de>
6991 PR tree-optimization/82244
6992 * tree-vrp.c (remove_range_assertions): Do not propagate
6993 a constant to abnormals but replace the assert with a copy.
6995 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6997 PR rtl-optimization/57878
6998 PR rtl-optimization/68988
6999 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7000 avoidance test involving non_reload_pseudos. Move frequency test
7001 below the general fragmentation avoidance test.
7003 2017-09-19 Richard Biener <rguenther@suse.de>
7005 PR tree-optimization/69728
7006 * graphite-sese-to-poly.c (schedule_error): New global.
7007 (add_loop_schedule): Handle empty domain by failing the
7009 (build_original_schedule): Handle schedule_error.
7011 2017-09-19 Richard Biener <rguenther@suse.de>
7013 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7014 Do not iterate to sibling loops but only to siblings of inner
7017 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7020 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7021 that the registers are different.
7023 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7025 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7026 to processor_model and "amdfam17h" to arch_names_table.
7027 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7029 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7032 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7034 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7035 Alan Hayward <alan.hayward@arm.com>
7036 David Sherwood <david.sherwood@arm.com>
7038 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7040 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7042 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7043 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7044 vect_schedule_slp_instance.
7045 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7046 Update call to vect_slp_analyze_node_operations. Simplify return
7048 (vect_slp_analyze_bb_1): Update call accordingly.
7049 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7050 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7051 (vect_schedule_slp): Update call accordingly.
7053 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7054 Alan Hayward <alan.hayward@arm.com>
7055 David Sherwood <david.sherwood@arm.com>
7057 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7058 with types that aren't in fact scalar.
7060 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7062 * tree-vect-slp.c (vect_record_max_nunits): New function,
7064 (vect_build_slp_tree_1): ...here.
7065 (vect_build_slp_tree_2): Call it for phis too.
7067 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7069 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7070 to vect_get_vec_def_for_operand when getting the mask operand.
7072 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7073 Alan Hayward <alan.hayward@arm.com>
7074 David Sherwood <david.sherwood@arm.com>
7076 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7079 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7080 Alan Hayward <alan.hayward@arm.com>
7081 David Sherwood <david.sherwood@arm.com>
7083 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7084 calculation for vector booleans.
7086 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7087 Alan Hayward <alan.hayward@arm.com>
7088 David Sherwood <david.sherwood@arm.com>
7090 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7092 (vect_transform_stmt): ...here.
7093 (vect_analyze_stmt): Use it instead of calling
7094 vectorizable_live_operation directly.
7096 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7098 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7099 non-SIMT targets in acc vector loops.
7101 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7103 * configure.ac: Add arc and check if assembler supports gdwarf2.
7104 * configure: Regenerate.
7106 2017-09-18 Richard Biener <rguenther@suse.de>
7108 PR tree-optimization/82220
7109 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7110 epilogue niters from the min_profitable_iters compute.
7112 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7115 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7116 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7117 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7119 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7122 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7123 switching to a new text section.
7125 2017-09-18 Richard Biener <rguenther@suse.de>
7127 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7129 (build_alias_set): Reject aliases with no access function.
7131 2017-09-18 Richard Biener <rguenther@suse.de>
7133 PR tree-optimization/79622
7134 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7136 (build_cross_bb_scalars_use): Likewise.
7138 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7140 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7142 2017-09-18 Alan Modra <amodra@gmail.com>
7145 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7146 stack_pointer_rtx for count 0. Update comments. Break up
7147 large rtl expression.
7149 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7151 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7152 Increase to 20 bytes.
7153 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7154 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7155 or avx version of the stub.
7157 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7160 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7161 compute the minimum stack alignment. Also update preferred stack
7162 boundary for leaf functions.
7164 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7166 PR tree-optimization/82228
7167 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7170 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7172 * common/config/nds32/nds32-common.c
7173 (nds32_option_optimization_table): Refine formatting.
7174 (nds32_option_optimization_table): Use -fsched-pressure and
7175 -fomit-frame-pointer for specific optimization level.
7177 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7179 * config/nds32/nds32.c: Refine formatting and comments.
7180 * config/nds32/nds32.h: Likewise.
7181 * config/nds32/nds32.md: Likewise.
7182 * config/nds32/nds32-cost.c: Likewise.
7183 * config/nds32/nds32-isr.c: Likewise.
7184 * config/nds32/nds32-md-auxiliary.c: Likewise.
7185 * config/nds32/nds32-multiple.md: Likewise.
7186 * config/nds32/nds32-predicates.c: Likewise.
7188 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7189 Jakub Jelinek <jakub@redhat.com>
7191 Add support for -std=c++2a.
7192 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7194 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7196 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7199 * doc/extend.texi (Common Function Attributes): Add
7200 references to ARM, AArch64, and S/390 specific attributes.
7201 (Function Specific Option Pragmas): Add AArch64 and S/390
7202 to list of back ends that support the target pragma.
7204 2017-09-15 Nathan Sidwell <nathan@acm.org>
7206 * doc/standards.texi: Fix C++17 description. Update URLs for
7209 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7211 * common.opt (Wcast-align=strict): New warning option.
7212 * doc/invoke.texi: Document -Wcast-align=strict.
7214 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7216 * cgraph.h (cgraph_thunk_info): Add comments.
7217 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7218 assert for VIRTUAL_* arguments stricter.
7220 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7222 PR tree-optimization/71026
7223 * match.pd: Move RDIV patterns from fold-const.c
7224 * fold-const.c (distribute_real_division): Removed.
7225 (fold_binary_loc): Remove calls to distribute_real_divison.
7227 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7229 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7230 c++1z and gnu++1z as deprecated. Change other references to
7231 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7232 Change -Wc++1z-compat to -Wc++17-compat.
7233 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7234 * dwarf2out.c (highest_c_language): Handle C++17.
7235 (gen_compile_unit_die): Likewise.
7237 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7239 PR rtl-optimization/82192
7240 * combine.c (make_extraction): Don't look through non-paradoxical
7241 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7244 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7245 Alan Hayward <alan.hayward@arm.com>
7246 David Sherwood <david.sherwood@arm.com>
7248 * target.def (function_arg_offset): New hook.
7249 * targhooks.h (default_function_arg_offset): Declare.
7250 * targhooks.c (default_function_arg_offset): New function.
7251 * function.c (locate_and_pad_parm): Use
7252 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7253 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7254 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7255 * doc/tm.texi: Regenerate.
7256 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7257 * config/spu/spu.c (spu_function_arg_offset): New function.
7258 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7259 * system.h (FUNCTION_ARG_OFFSET): Poison.
7261 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7262 Alan Hayard <alan.hayward@arm.com>
7263 David Sherwood <david.sherwood@arm.com>
7265 * target.def (truly_noop_truncation): New hook.
7266 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7267 than TRULY_NOOP_TRUNCATION.
7268 * hooks.h (hook_bool_uint_uint_true): Declare.
7269 * hooks.c (hook_bool_uint_uint_true): New function.
7270 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7271 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7272 * doc/tm.texi: Regenerate.
7273 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7274 rather than TRULY_NOOP_TRUNCATION in comments.
7275 (simplify_comparison): Likewise.
7276 (record_truncated_value): Likewise.
7277 * expmed.c (extract_bit_field_1): Likewise.
7278 (extract_split_bit_field): Likewise.
7279 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7280 instead of TRULY_NOOP_TRUNCATION.
7281 * function.c (assign_parm_setup_block): Likewise.
7282 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7283 * rtlhooks.c: Include target.h.
7284 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7285 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7286 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7287 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7288 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7289 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7290 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7291 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7292 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7293 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7294 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7295 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7296 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7297 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7298 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7299 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7300 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7301 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7302 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7303 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7304 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7305 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7306 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7307 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7308 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7309 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7310 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7311 rather than TRULY_NOOP_TRUNCATION in comments.
7312 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7313 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7314 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7315 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7316 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7317 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7318 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7319 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7320 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7321 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7322 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7323 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7324 rather than TRULY_NOOP_TRUNCATION in comments.
7325 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7326 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7327 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7328 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7329 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7330 TRULY_NOOP_TRUNCATION condition.
7331 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7332 (TRULY_NOOP_TRUNCATION): Delete.
7333 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7334 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7335 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7336 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7337 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7338 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7339 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7340 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7341 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7342 rather than TRULY_NOOP_TRUNCATION in comments.
7343 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7344 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7345 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7346 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7347 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7348 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7350 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7353 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7354 (*cmp_ior): Likewise.
7355 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7356 (*ior_scc_scc_cmp): Likewise.
7357 (*and_scc_scc): Likewise.
7358 (*and_scc_scc_cmp): Likewise.
7360 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7361 Alan Hayard <alan.hayward@arm.com>
7362 David Sherwood <david.sherwood@arm.com>
7364 * target.def (can_change_mode_class): New hook.
7365 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7366 (hard_regno_nregs): Likewise.
7367 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7368 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7369 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7370 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7371 (LOAD_EXTEND_OP): Update accordingly.
7372 * doc/tm.texi: Regenerate.
7373 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7374 CANNOT_CHANGE_MODE_CLASS.
7375 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7376 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7377 * combine.c (simplify_set): Update accordingly.
7378 * emit-rtl.c (validate_subreg): Likewise.
7379 * recog.c (general_operand): Likewise.
7380 * regcprop.c (mode_change_ok): Likewise.
7381 * reload1.c (choose_reload_regs): Likewise.
7382 (inherit_piecemeal_p): Likewise.
7383 * rtlanal.c (simplify_subreg_regno): Likewise.
7384 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7385 instead of CANNOT_CHANGE_MODE_CLASS.
7386 (reload_cse_simplify_operands): Likewise.
7387 * reload.c (push_reload): Use targetm.can_change_mode_class
7388 instead of CANNOT_CHANGE_MODE_CLASS.
7389 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7390 REG_CANNOT_CHANGE_MODE_P.
7391 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7392 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7393 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7394 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7395 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7396 (arm_can_change_mode_class): New function.
7397 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7398 than CANNOT_CHANGE_MODE_CLASS in comments.
7399 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7400 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7401 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7402 (ix86_can_change_mode_class): ...this new function, inverting the
7403 sense of the return value.
7404 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7405 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7406 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7407 (ia64_can_change_mode_class): New function.
7408 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7409 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7410 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7411 (m32c_can_change_mode_class): ...this new function, inverting the
7412 sense of the return value.
7413 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7414 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7415 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7416 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7417 (mips_can_change_mode_class): ...this new function, inverting the
7418 sense of the return value.
7419 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7420 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7421 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7422 (msp430_can_change_mode_class): New function.
7423 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7424 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7425 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7426 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7427 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7428 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7429 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7430 (pa_cannot_change_mode_class): Replace with...
7431 (pa_can_change_mode_class): ...this new function, inverting the
7432 sense of the return value.
7433 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7434 than CANNOT_CHANGE_MODE_CLASS in comments.
7435 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7436 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7437 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7438 (pdp11_cannot_change_mode_class): Replace with...
7439 (pdp11_can_change_mode_class): ...this new function, inverting the
7440 sense of the return value.
7441 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7442 * config/powerpcspe/powerpcspe-protos.h
7443 (rs6000_cannot_change_mode_class_ptr): Delete.
7444 * config/powerpcspe/powerpcspe.c
7445 (rs6000_cannot_change_mode_class_ptr): Delete.
7446 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7447 (rs6000_option_override_internal): Assign to
7448 targetm.can_change_mode_class instead of
7449 rs6000_cannot_change_mode_class_ptr.
7450 (rs6000_cannot_change_mode_class): Replace with...
7451 (rs6000_can_change_mode_class): ...this new function, inverting the
7452 sense of the return value.
7453 (rs6000_debug_cannot_change_mode_class): Replace with...
7454 (rs6000_debug_can_change_mode_class): ...this new function.
7455 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7456 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7457 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7458 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7459 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7461 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7462 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7463 (rs6000_option_override_internal): Assign to
7464 targetm.can_change_mode_class instead of
7465 rs6000_cannot_change_mode_class_ptr.
7466 (rs6000_cannot_change_mode_class): Replace with...
7467 (rs6000_can_change_mode_class): ...this new function, inverting the
7468 sense of the return value.
7469 (rs6000_debug_cannot_change_mode_class): Replace with...
7470 (rs6000_debug_can_change_mode_class): ...this new function.
7471 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7472 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7473 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7474 (s390_can_change_mode_class): ...this new function, inverting the
7475 sense of the return value.
7476 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7477 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7478 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7479 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7480 (sh_cannot_change_mode_class): Replace with...
7481 (sh_can_change_mode_class): ...this new function, inverting the
7482 sense of the return value.
7483 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7484 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7485 (sparc_can_change_mode_class): New function.
7486 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7487 * config/spu/spu.c (spu_can_change_mode_class): New function.
7488 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7489 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7490 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7491 (visium_can_change_mode_class): New function.
7492 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7494 2017-09-15 Richard Biener <rguenther@suse.de>
7496 PR tree-optimization/82217
7497 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7498 but not undefined case.
7500 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7503 * postreload.c (reload_cse_simplify_operands): Skip
7504 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7506 2017-09-15 Richard Biener <rguenther@suse.de>
7508 PR tree-optimization/68823
7509 * graphite-scop-detection.c (build_alias_set): If we have a
7510 possible dependence check whether we can handle them by just
7511 looking at the DRs DR_ACCESS_FNs.
7512 (build_scops): If build_alias_set fails, fail the SCOP.
7514 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7516 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7517 to support float128 built-in functions that require the ISA 3.0
7519 (BU_FLOAT128_3_HW): Likewise.
7520 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7522 (FMAF128): Likewise.
7524 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7525 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7526 floating point instructions.
7527 (rs6000_invalid_builtin): Likewise.
7528 (rs6000_builtin_mask_names): Likewise.
7529 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7530 (RS6000_BTM_FLOAT128_HW): Likewise.
7531 (RS6000_BTM_COMMON): Likewise.
7532 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7534 * doc/extend.texi (RS/6000 built-in functions): Document the
7535 IEEE 128-bit floating point square root and fused multiply-add
7538 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7540 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7541 reg (r2) isn't in the set of registers defined in the prologue.
7543 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7544 Alan Hayward <alan.hayward@arm.com>
7545 David Sherwood <david.sherwood@arm.com>
7547 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7548 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7549 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7550 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7551 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7553 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7554 max_vectorization_factor.
7555 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7557 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7558 Alan Hayward <alan.hayward@arm.com>
7559 David Sherwood <david.sherwood@arm.com>
7561 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7562 (vect_worthwhile_without_simd_p): Declare.
7563 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7564 (vectorizable_reduction): Use it.
7565 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7566 (vectorizable_operation): Likewise.
7568 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7569 Alan Hayward <alan.hayward@arm.com>
7570 David Sherwood <david.sherwood@arm.com>
7572 * tree-vectorizer.h (vect_get_num_copies): New function.
7573 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7574 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7575 (vectorizable_induction): Likewise.
7576 (vectorizable_live_operation): Likewise.
7577 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7578 (vectorizable_bswap): Likewise.
7579 (vectorizable_call): Likewise.
7580 (vectorizable_conversion): Likewise.
7581 (vectorizable_assignment): Likewise.
7582 (vectorizable_shift): Likewise.
7583 (vectorizable_operation): Likewise.
7584 (vectorizable_store): Likewise.
7585 (vectorizable_load): Likewise.
7586 (vectorizable_condition): Likewise.
7587 (vectorizable_comparison): Likewise.
7588 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7590 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7591 Alan Hayward <alan.hayward@arm.com>
7592 David Sherwood <david.sherwood@arm.com>
7594 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7595 of vect_init_vector.
7597 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7598 Alan Hayward <alan.hayward@arm.com>
7599 David Sherwood <david.sherwood@arm.com>
7601 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7602 an inline wrapper that provides a location.
7603 (gimple_build_vector): Likewise.
7604 * gimple-fold.c (gimple_build_vector_from_val): New function.
7605 (gimple_build_vector): Likewise.
7606 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7607 functions to build the initial value. Always return a gimple value.
7608 (get_initial_defs_for_reduction): Likewise. Only compute
7610 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7611 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7612 (vectorizable_induction): Use gimple_build_vector rather than
7615 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7616 Alan Hayward <alan.hayward@arm.com>
7617 David Sherwood <david.sherwood@arm.com>
7619 * target.h (vec_perm_indices): New typedef.
7620 (auto_vec_perm_indices): Likewise.
7621 * optabs-query.h: Include target.h
7622 (can_vec_perm_p): Take a vec_perm_indices *.
7623 * optabs-query.c (can_vec_perm_p): Likewise.
7624 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7625 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7626 * tree-vect-generic.c (lower_vec_perm): Likewise.
7627 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7628 (vect_grouped_load_supported): Likewise.
7629 (vect_shift_permute_load_chain): Likewise.
7630 (vect_permute_store_chain): Use auto_vec_perm_indices.
7631 (vect_permute_load_chain): Likewise.
7632 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7633 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7634 Update uses of can_vec_perm_p.
7635 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7636 mode with a number of elements. Take a vec_perm_indices *.
7637 (vect_create_epilog_for_reduction): Update accordingly.
7638 Use auto_vec_perm_indices.
7639 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7640 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7641 (vect_transform_slp_perm_load): Likewise.
7642 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7643 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7644 (vect_gen_perm_mask_checked): Likewise.
7645 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7646 (vect_gen_perm_mask_checked): Likewise.
7647 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7648 (vectorizable_store): Likewise.
7649 (vectorizable_load): Likewise.
7650 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7651 (vectorizable_bswap): Likewise.
7653 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7654 Alan Hayward <alan.hayward@arm.com>
7655 David Sherwood <david.sherwood@arm.com>
7657 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7658 * tree.c (build_vector): Likewise.
7659 (build_vector_from_ctor): Update accordingly.
7660 (build_vector_from_val): Likewise.
7661 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7662 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7663 * tree-vect-generic.c (add_rshift): Likewise.
7664 (expand_vector_divmod): Likewise.
7665 (optimize_vector_constructor): Likewise.
7666 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7667 (vect_transform_slp_perm_load): Likewise.
7668 (vect_schedule_slp_instance): Likewise.
7669 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7670 (vectorizable_call): Likewise.
7671 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7672 * expmed.c (make_tree): Likewise.
7673 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7674 a vector passed to build_vector.
7675 (fold_convert_const): Likewise.
7676 (exact_inverse): Likewise.
7677 (fold_ternary_loc): Likewise.
7678 (fold_relational_const): Likewise.
7679 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7680 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7681 (const_unop): Likewise. Store the reduction accumulator in a
7682 variable rather than an array.
7683 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7684 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7685 the new vector, rather than constructing it after the input arrays.
7686 (native_interpret_vector): Use auto_vec<tree> when building
7687 a vector passed to build_vector. Add elements in order.
7688 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7689 auto_vec<tree> when building a vector passed to build_vector.
7690 (vect_create_epilog_for_reduction): Likewise.
7691 (vectorizable_induction): Likewise.
7692 (get_initial_def_for_reduction): Likewise. Fix indentation of
7694 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7696 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7697 passed to build_vector.
7699 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7700 Alan Hayward <alan.hayward@arm.com>
7701 David Sherwood <david.sherwood@arm.com>
7703 * tree-core.h (tree_base::u): Add an "nelts" field.
7704 (tree_vector): Use VECTOR_CST_NELTS as the length.
7705 * tree.c (tree_size): Likewise.
7706 (make_vector): Initialize VECTOR_CST_NELTS.
7707 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7708 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7709 TYPE_VECTOR_SUBPARTS.
7710 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7711 as the number of units, setting it from VECTOR_CST_NELTS.
7712 (const_vector_from_tree): Likewise.
7713 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7714 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7715 (fold_negate_expr_1): Likewise.
7716 (fold_convert_const): Likewise.
7717 (const_binop): Likewise. Differentiate the number of output and
7719 (const_unop): Likewise.
7720 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7721 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7722 in cases that did the opposite.
7724 2017-09-14 Richard Biener <rguenther@suse.de>
7726 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7729 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7731 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7733 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7736 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7737 if and where to split a bb, except for splitting before debug insn
7738 sequences followed by non-label real insn. Delete debug insns
7739 in between basic blocks.
7741 * combine.c (make_compound_operation_int): Formatting fixes.
7743 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7744 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7745 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7746 * config/sol2.h (LINK_EH_SPEC): Likewise.
7747 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7748 * config/s390/linux.h (LINK_SPEC): Likewise.
7749 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7750 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7751 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7752 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7753 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7754 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7755 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7756 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7758 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7760 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7762 (ENDFILE_LINUX_SPEC): Likewise.
7763 (LINK_EH_SPEC): Likewise.
7764 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7765 (LINK_OS_LINUX_SPEC32): Likewise.
7766 (LINK_OS_LINUX_SPEC64): Likewise.
7767 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7768 (LINK_OS_LINUX_SPEC): Likewise.
7770 2017-09-13 Martin Liska <mliska@suse.cz>
7773 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7774 CASE_HIGH is NULL_TREE.
7776 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7777 Alan Hayward <alan.hayward@arm.com>
7778 David Sherwood <david.sherwood@arm.com>
7780 * target.def (secondary_memory_needed): New hook.
7781 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7782 instead of SECONDARY_MEMORY_NEEDED.
7783 (secondary_memory_needed_mode): Likewise.
7784 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7785 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7786 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7787 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7788 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7789 * doc/tm.texi: Regenerate.
7790 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7791 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7792 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7793 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7794 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7795 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7796 mode argument first and change the reg_class arguments to reg_class_t.
7797 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7798 Make static. Update the call to inline_secondary_memory_needed.
7799 (ix86_register_move_cost): Update the call to
7800 inline_secondary_memory_needed.
7801 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7802 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7804 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7805 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7807 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7808 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7809 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7810 and match hook interface. Add comment from mips.h.
7811 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7812 * config/mmix/mmix.md (truncdfsf2): Refer to
7813 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7815 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7816 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7817 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7818 (pa_secondary_memory_needed): New function.
7819 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7820 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7821 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7822 (pdp11_secondary_memory_needed): Make static and match hook interface.
7823 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7824 * config/powerpcspe/powerpcspe-protos.h
7825 (rs6000_secondary_memory_needed_ptr): Delete.
7826 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7828 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7829 (rs6000_option_override_internal): Assign to
7830 targetm.secondary_memory_needed rather than
7831 rs6000_secondary_memory_needed_ptr.
7832 (rs6000_secondary_memory_needed): Match hook interface.
7833 (rs6000_debug_secondary_memory_needed): Likewise.
7834 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7835 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7836 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7837 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7838 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7839 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7841 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7842 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7843 (rs6000_option_override_internal): Assign to
7844 targetm.secondary_memory_needed rather than
7845 rs6000_secondary_memory_needed_ptr.
7846 (rs6000_secondary_memory_needed): Match hook interface.
7847 (rs6000_debug_secondary_memory_needed): Likewise.
7848 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7849 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7850 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7851 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7852 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7853 (sparc_secondary_memory_needed): New function.
7854 * lra-constraints.c (check_and_process_move): Refer to
7855 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7857 (curr_insn_transform): Likewise.
7858 (process_alt_operands): Use targetm.secondary_memory_needed
7859 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7860 (check_secondary_memory_needed_p): Likewise.
7861 (choose_split_class): Likewise.
7862 * reload.c: Unconditionally include code that was previously
7863 conditional on SECONDARY_MEMORY_NEEDED.
7864 (push_secondary_reload): Use targetm.secondary_memory_needed
7865 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7866 (push_reload): Likewise.
7867 * reload1.c: Unconditionally include code that was previously
7868 conditional on SECONDARY_MEMORY_NEEDED.
7869 (choose_reload_regs): Use targetm.secondary_memory_needed
7870 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7871 (gen_reload): Likewise.
7872 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7874 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7875 Alan Hayward <alan.hayward@arm.com>
7876 David Sherwood <david.sherwood@arm.com>
7878 * target.def (secondary_memory_needed_mode): New hook:
7879 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7880 * targhooks.h (default_secondary_memory_needed_mode): New function.
7881 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7882 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7883 * doc/tm.texi: Regenerate.
7884 * lra-constraints.c (check_and_process_move): Use
7885 targetm.secondary_memory_needed_mode instead of
7886 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7887 (curr_insn_transform): Likewise.
7888 * reload.c (get_secondary_mem): Likewise.
7889 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7890 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7892 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7893 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7894 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7895 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7896 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7898 * config/powerpcspe/powerpcspe-protos.h
7899 (rs6000_secondary_memory_needed_mode): Delete.
7900 * config/powerpcspe/powerpcspe.c
7901 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7902 (rs6000_secondary_memory_needed_mode): Make static.
7903 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7904 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7906 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7908 (rs6000_secondary_memory_needed_mode): Make static.
7909 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7910 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7911 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7912 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7913 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7915 (sparc_secondary_memory_needed_mode): New function.
7916 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7918 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7920 * config/aarch64/constraints.md (Umq): New constraint.
7921 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7923 (mov<mode>): Update condition.
7925 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7927 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7928 when bitposition is the same.
7930 2017-09-13 Richard Biener <rguenther@suse.de>
7932 * dwarf2out.c (output_die_symbol): Remove.
7933 (output_die): Do not output a DIEs symbol.
7935 2017-09-13 Richard Biener <rguenther@suse.de>
7938 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7939 default-def to avoid breaking iterator update with the weird
7940 interaction with cgraph_update_edges_for_call_stmt_node.
7942 2017-09-13 Richard Biener <rguenther@suse.de>
7944 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7945 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7946 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7947 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7949 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7951 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7952 Disable pc relative literal load irrespective of
7953 TARGET_FIX_ERR_A53_84341 for default.
7955 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7957 * config/sparc/sparc.c (output_return): Output the source location of
7958 the insn in the delay slot, if any.
7959 (output_sibcall): Likewise.
7961 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7964 * common.opt (-static-pie): New alias.
7965 (shared): Negate static-pie.
7966 (-no-pie): Update help text.
7968 (static-pie): New option.
7969 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7970 -static-pie support.
7971 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7972 (LINK_EH_SPEC): Likewise.
7973 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7974 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7975 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7976 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7977 (init_gcc_specs): Likewise.
7978 (init_spec): Likewise.
7979 (display_help): Update help message for -pie.
7980 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7983 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7985 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7986 (movdi_aarch64): Likewise.
7987 (movti_aarch64): Likewise.
7989 2017-09-12 Simon Wright <simon@pushface.org>
7992 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7993 calculation of the minor version, always output as 0.
7995 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7998 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7999 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8000 on it early, rather than manual conversion late. For
8001 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8002 instead of performing manual conversion.
8004 2017-09-12 Carl Love <cel@us.ibm.com>
8006 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8007 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8009 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8010 VMULOSW): Add definitions.
8011 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8012 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8013 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8014 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8015 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8017 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8019 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8021 (movti_aarch64): Likewise.
8022 (movdf_aarch64): Likewise.
8023 (movtf_aarch64): Likewise.
8024 (load_pairdi): Likewise.
8025 (store_pairdi): Likewise.
8026 (load_pairdf): Likewise.
8027 (store_pairdf): Likewise.
8028 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8029 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8030 (ldr_got_small_<mode>): Likewise.
8031 (ldr_got_small_28k_<mode>): Likewise.
8032 (ldr_got_tiny): Likewise.
8033 * config/aarch64/iterators.md (ldst_sz): New.
8034 (ldpstp_sz): Likewise.
8035 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8037 (thunderx_load): Split load_8 to load_16.
8038 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8040 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8041 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8042 (xgene1_store_pair): Split store_8 to store_16.
8043 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8044 (falkor_st_0_st_sd): Split store_8 to store_16.
8046 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8048 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8049 and store1/2/3/4 to store_4/8/12/16.
8050 * config/aarch64/aarch64.md: Update for rename.
8051 * config/arm/arm.md: Likewise.: Likewise.
8052 * config/arm/arm.c: Likewise.
8053 * config/arm/thumb1.md: Likewise.
8054 * config/arm/thumb2.md: Likewise.
8055 * config/arm/vfp.md: Likewise.
8056 * config/arm/arm-generic.md: Likewise.
8057 * config/arm/arm1020e.md: Likewise.
8058 * config/arm/arm1026ejs.md: Likewise.
8059 * config/arm/arm1136jfs.md: Likewise.
8060 * config/arm/arm926ejs.md: Likewise.
8061 * config/arm/cortex-a15.md: Likewise.
8062 * config/arm/cortex-a17.md: Likewise.
8063 * config/arm/cortex-a5.md: Likewise.
8064 * config/arm/cortex-a53.md: Likewise.
8065 * config/arm/cortex-a57.md: Likewise.
8066 * config/arm/cortex-a7.md: Likewise.
8067 * config/arm/cortex-a8.md: Likewise.
8068 * config/arm/cortex-a9.md: Likewise.
8069 * config/arm/cortex-m4.md: Likewise.
8070 * config/arm/cortex-m7.md: Likewise.
8071 * config/arm/cortex-r4.md: Likewise.
8072 * config/arm/exynos-m1.md: Likewise.
8073 * config/arm/fa526.md: Likewise.
8074 * config/arm/fa606te.md: Likewise.
8075 * config/arm/fa626te.md: Likewise.
8076 * config/arm/fa726te.md: Likewise.
8077 * config/arm/fmp626.md: Likewise.
8078 * config/arm/iwmmxt.md: Likewise.
8079 * config/arm/ldmstm.md: Likewise.
8080 * config/arm/marvell-pj4.md: Likewise.
8081 * config/arm/xgene1.md: Likewise.
8082 * config/aarch64/thunderx.md: Likewise.
8083 * config/aarch64/thunderx2t99.md: Likewise.
8084 * config/aarch64/falkor.md: Likewise.
8086 2017-09-12 Martin Liska <mliska@suse.cz>
8088 * attribs.c (private_lookup_attribute): New function.
8089 * attribs.h (private_lookup_attribute): Declared here.
8090 (lookup_attribute): Called from this place.
8092 2017-09-12 Richard Biener <rguenther@suse.de>
8094 PR tree-optimization/82157
8095 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8096 stmts with side-effects.
8098 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8099 Alan Hayward <alan.hayward@arm.com>
8100 David Sherwood <david.sherwood@arm.com>
8102 * target.def (hard_regno_nregs): New hook.
8103 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8104 * targhooks.h (default_hard_regno_nregs): Declare.
8105 * targhooks.c (default_hard_regno_nregs): New function.
8106 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8107 (TARGET_HARD_REGNO_NREGS): ...this hook.
8108 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8109 (CLASS_MAX_NREGS): Likewise.
8110 * doc/tm.texi: Regenerate.
8111 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8112 instead of HARD_REGNO_NREGS.
8113 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8114 HARD_REGNO_NREGS in the comment.
8115 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8116 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8117 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8118 Return an unsigned int.
8119 (TARGET_HARD_REGNO_NREGS): Redefine.
8120 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8121 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8122 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8123 (arc_hard_regno_nregs): New function.
8124 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8125 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8126 (arm_hard_regno_nregs): New function.
8127 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8128 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8129 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8130 (TARGET_HARD_REGNO_NREGS): Redefine.
8131 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8132 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8133 (HARD_REGNO_NREGS): Delete.
8134 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8135 (cr16_hard_regno_nregs): New function.
8136 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8137 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8138 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8139 (cris_hard_regno_nregs): New function.
8140 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8141 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8142 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8143 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8144 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8145 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8146 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8147 (frv_hard_regno_nregs): Make static. Take and return an
8149 (frv_class_max_nregs): Remove outdated copy of documentation.
8150 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8151 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8152 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8153 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8154 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8155 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8156 (TARGET_HARD_REGNO_NREGS): Redefine.
8157 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8158 (CLASS_MAX_NREGS): Update comment.
8159 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8160 (ia64_hard_regno_nregs): New function.
8161 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8162 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8163 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8164 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8165 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8167 (m32c_hard_regno_nregs): Likewise. Make static.
8168 (TARGET_HARD_REGNO_NREGS): Redefine.
8169 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8170 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8171 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8172 (m68k_hard_regno_nregs): New function.
8173 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8174 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8175 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8176 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8177 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8178 Take and return an unsigned int.
8179 (TARGET_HARD_REGNO_NREGS): Redefine.
8180 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8181 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8182 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8183 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8184 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8185 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8186 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8187 (msp430_hard_regno_nregs): Make static. Take and return an
8189 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8190 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8191 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8192 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8193 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8194 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8195 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8196 (TARGET_HARD_REGNO_NREGS): Redefine.
8197 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8198 (PA_HARD_REGNO_NREGS): ...this.
8199 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8200 (PA_HARD_REGNO_NREGS): ...this.
8201 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8202 (pa_hard_regno_nregs): New function.
8203 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8204 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8205 (pdp11_hard_regno_nregs): New function.
8206 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8207 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8208 (rs6000_hard_regno_nregs_hook): New function.
8209 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8210 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8211 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8212 Take and return an unsigned int. Move earlier in file.
8213 (TARGET_HARD_REGNO_NREGS): Redefine.
8214 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8215 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8216 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8217 (rl78_hard_regno_nregs): Make static. Take and return an
8219 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8220 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8221 (rs6000_hard_regno_nregs_hook): New function.
8222 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8223 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8224 (TARGET_HARD_REGNO_NREGS): Redefine.
8225 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8226 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8227 instead of HARD_REGNO_NREGS.
8228 (s390_hard_regno_nregs): New function.
8229 (s390_hard_regno_mode_ok): Add comment from s390.h.
8230 (TARGET_HARD_REGNO_NREGS): Redefine.
8231 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8232 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8233 (sh_hard_regno_nregs): New function.
8234 (sh_pass_in_reg_p): Use it.
8235 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8236 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8237 (sparc_hard_regno_nregs): New function.
8238 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8239 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8240 (spu_function_arg_advance): Use it, supplying a valid register number.
8241 (TARGET_HARD_REGNO_NREGS): Redefine.
8242 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8243 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8244 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8245 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8246 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8247 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8248 (CLASS_MAX_NREGS): Remove copy of old documentation.
8249 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8250 (visium_hard_regno_nregs): New function.
8251 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8252 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8253 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8254 xtensa_hard_regno_nregs): New function.
8255 * system.h (HARD_REGNO_NREGS): Poison.
8257 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8259 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8260 hard_regno_nregs instead of HARD_REGNO_NREGS.
8261 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8262 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8263 (c6x_expand_epilogue): Likewise.
8264 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8265 (frv_read_iacc_argument): Likewise.
8266 * config/sh/sh.c: Include regs.h.
8267 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8268 (regs_used): Likewise.
8269 (output_stack_adjust): Likewise.
8270 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8271 * expmed.c: Include regs.h.
8272 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8273 * ree.c: Include regs.h.
8274 (combine_reaching_defs): Use hard_regno_nregs instead of
8276 (add_removable_extension): Likewise.
8278 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8280 * regs.h (hard_regno_nregs): Turn into a function.
8281 (end_hard_regno): Update accordingly.
8282 * caller-save.c (setup_save_areas): Likewise.
8283 (save_call_clobbered_regs): Likewise.
8284 (replace_reg_with_saved_mem): Likewise.
8285 (insert_restore): Likewise.
8286 (insert_save): Likewise.
8287 * combine.c (can_change_dest_mode): Likewise.
8288 (move_deaths): Likewise.
8289 (distribute_notes): Likewise.
8290 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8291 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8292 (rs6000_split_multireg_move): Likewise.
8293 (rs6000_register_move_cost): Likewise.
8294 (rs6000_memory_move_cost): Likewise.
8295 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8296 (rs6000_split_multireg_move): Likewise.
8297 (rs6000_register_move_cost): Likewise.
8298 (rs6000_memory_move_cost): Likewise.
8299 * cselib.c (cselib_reset_table): Likewise.
8300 (cselib_lookup_1): Likewise.
8301 * emit-rtl.c (set_mode_and_regno): Likewise.
8302 * function.c (aggregate_value_p): Likewise.
8303 * ira-color.c (setup_profitable_hard_regs): Likewise.
8304 (check_hard_reg_p): Likewise.
8305 (calculate_saved_nregs): Likewise.
8306 (assign_hard_reg): Likewise.
8307 (improve_allocation): Likewise.
8308 (calculate_spill_cost): Likewise.
8309 * ira-emit.c (modify_move_list): Likewise.
8310 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8311 (ira_hard_reg_in_set_p): Likewise.
8312 * ira.c (setup_reg_mode_hard_regset): Likewise.
8313 (clarify_prohibited_class_mode_regs): Likewise.
8314 (check_allocation): Likewise.
8315 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8316 (lra_setup_reg_renumber): Likewise.
8317 (setup_try_hard_regno_pseudos): Likewise.
8318 (spill_for): Likewise.
8319 (assign_hard_regno): Likewise.
8320 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8321 * lra-constraints.c (in_class_p): Likewise.
8322 (lra_constraint_offset): Likewise.
8323 (simplify_operand_subreg): Likewise.
8324 (lra_constraints): Likewise.
8325 (split_reg): Likewise.
8326 (split_if_necessary): Likewise.
8327 (invariant_p): Likewise.
8328 (inherit_in_ebb): Likewise.
8329 * lra-lives.c (process_bb_lives): Likewise.
8330 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8331 (get_hard_regs): Likewise.
8332 (do_remat): Likewise.
8333 * lra-spills.c (assign_spill_hard_regs): Likewise.
8334 * mode-switching.c (create_pre_exit): Likewise.
8335 * postreload.c (reload_combine_recognize_pattern): Likewise.
8336 * recog.c (peep2_find_free_register): Likewise.
8337 * regcprop.c (kill_value_regno): Likewise.
8338 (set_value_regno): Likewise.
8339 (copy_value): Likewise.
8340 (maybe_mode_change): Likewise.
8341 (find_oldest_value_reg): Likewise.
8342 (copyprop_hardreg_forward_1): Likewise.
8343 * regrename.c (check_new_reg_p): Likewise.
8344 (regrename_do_replace): Likewise.
8345 * reload.c (push_reload): Likewise.
8346 (combine_reloads): Likewise.
8347 (find_dummy_reload): Likewise.
8348 (operands_match_p): Likewise.
8349 (find_reloads): Likewise.
8350 (find_equiv_reg): Likewise.
8351 (reload_adjust_reg_for_mode): Likewise.
8352 * reload1.c (count_pseudo): Likewise.
8353 (count_spilled_pseudo): Likewise.
8354 (find_reg): Likewise.
8355 (clear_reload_reg_in_use): Likewise.
8356 (free_for_value_p): Likewise.
8357 (allocate_reload_reg): Likewise.
8358 (choose_reload_regs): Likewise.
8359 (reload_adjust_reg_for_temp): Likewise.
8360 (emit_reload_insns): Likewise.
8361 (delete_output_reload): Likewise.
8362 * rtlanal.c (subreg_get_info): Likewise.
8363 * sched-deps.c (sched_analyze_reg): Likewise.
8364 * sel-sched.c (init_regs_for_mode): Likewise.
8365 (mark_unavailable_hard_regs): Likewise.
8366 (choose_best_reg_1): Likewise.
8367 (verify_target_availability): Likewise.
8368 * valtrack.c (dead_debug_insert_temp): Likewise.
8369 * var-tracking.c (track_loc_p): Likewise.
8370 (emit_note_insn_var_location): Likewise.
8371 * varasm.c (make_decl_rtl): Likewise.
8372 * reginfo.c (choose_hard_reg_mode): Likewise.
8373 (init_reg_modes_target): Refer directly to
8374 this_target_regs->x_hard_regno_nregs.
8376 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8378 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8379 instead of hard_regno_nregs.
8381 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8383 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8384 end_hard_regno instead of hard_regno_nregs.
8385 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8386 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8387 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8388 * ira-color.c (improve_allocation): Likewise.
8389 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8390 * lra-lives.c (mark_regno_live): Likewise.
8391 (mark_regno_dead): Likewise.
8392 * lra-remat.c (operand_to_remat): Likewise.
8393 * lra.c (collect_non_operand_hard_regs): Likewise.
8394 * postreload.c (reload_combine_note_store): Likewise.
8395 (move2add_valid_value_p): Likewise.
8396 * reload.c (regno_clobbered_p): Likewise.
8398 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8400 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8402 * config/v850/v850.c (v850_reorg): Likewise.
8403 * reload.c (refers_to_regno_for_reload_p): Likewise.
8404 (find_equiv_reg): Likewise.
8405 * reload1.c (reload_reg_reaches_end_p): Likewise.
8407 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8409 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8411 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8412 * config/arm/arm.c (output_move_neon): Likewise.
8413 (arm_attr_length_move_neon): Likewise.
8414 (neon_split_vcombine): Likewise.
8415 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8416 (c6x_mark_reg_written): Likewise.
8417 (c6x_dwarf_register_span): Likewise.
8418 * config/i386/i386.c (ix86_save_reg): Likewise.
8419 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8420 (rws_access_reg): Likewise.
8421 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8422 * mode-switching.c (create_pre_exit): Likewise.
8423 * ree.c (combine_reaching_defs): Likewise.
8424 (add_removable_extension): Likewise.
8425 * regcprop.c (find_oldest_value_reg): Likewise.
8426 (copyprop_hardreg_forward_1): Likewise.
8427 * reload.c (reload_inner_reg_of_subreg): Likewise.
8428 (push_reload): Likewise.
8429 (combine_reloads): Likewise.
8430 (find_dummy_reload): Likewise.
8431 (reload_adjust_reg_for_mode): Likewise.
8432 * reload1.c (find_reload_regs): Likewise.
8433 (forget_old_reloads_1): Likewise.
8434 (reload_reg_free_for_value_p): Likewise.
8435 (reload_adjust_reg_for_temp): Likewise.
8436 (emit_reload_insns): Likewise.
8437 (delete_output_reload): Likewise.
8438 * sel-sched.c (choose_best_reg_1): Likewise.
8439 (choose_best_pseudo_reg): Likewise.
8441 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8442 Alan Hayward <alan.hayward@arm.com>
8443 David Sherwood <david.sherwood@arm.com>
8445 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8446 * target.def (slow_unaligned_access): New hook.
8447 * targhooks.h (default_slow_unaligned_access): Declare.
8448 * targhooks.c (default_slow_unaligned_access): New function.
8449 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8450 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8451 * doc/tm.texi: Regenerate.
8452 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8453 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8454 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8456 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8457 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8459 (rs6000_slow_unaligned_access): New function.
8460 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8461 (expand_block_compare): Likewise.
8462 (expand_strn_compare): Likewise.
8463 (rs6000_rtx_costs): Likewise.
8464 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8465 (riscv_slow_unaligned_access): Likewise.
8466 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8467 (riscv_slow_unaligned_access_p): ...this and make static.
8468 (riscv_option_override): Update accordingly.
8469 (riscv_slow_unaligned_access): New function.
8470 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8471 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8472 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8473 (rs6000_slow_unaligned_access): New function.
8474 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8475 (rs6000_rtx_costs): Likewise.
8476 * config/rs6000/rs6000-string.c (expand_block_compare)
8477 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8478 of SLOW_UNALIGNED_ACCESS.
8479 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8480 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8481 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8482 of SLOW_UNALIGNED_ACCESS.
8483 * expmed.c (simple_mem_bitfield_p): Likewise.
8484 * expr.c (alignment_for_piecewise_move): Likewise.
8485 (emit_group_load_1): Likewise.
8486 (emit_group_store): Likewise.
8487 (copy_blkmode_from_reg): Likewise.
8488 (emit_push_insn): Likewise.
8489 (expand_assignment): Likewise.
8490 (store_field): Likewise.
8491 (expand_expr_real_1): Likewise.
8492 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8493 * lra-constraints.c (simplify_operand_subreg): Likewise.
8494 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8495 * gimple-ssa-store-merging.c: Likewise in block comment at start
8497 * tree-ssa-strlen.c: Include target.h.
8498 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8499 of SLOW_UNALIGNED_ACCESS.
8500 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8502 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8504 PR rtl-optimization/82185
8505 * expmed.c (emit_store_flag_int): Only test tem if it has been
8508 2017-09-12 Richard Biener <rguenther@suse.de>
8511 * match.pd ((FTYPE) N CMP CST): Fix typo.
8513 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8515 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8517 (mips_near_type_p): Add 'short_call' attribute as a synonym
8519 * doc/extend.texi (short_call): Document new function attribute.
8521 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8524 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8525 assertion check that in the condition.
8526 (get_atomic_generic_size): Likewise. Before testing if parameter
8527 has pointer type, if it has array type, call for C++
8528 default_conversion to perform array-to-pointer conversion.
8530 2017-09-12 Richard Biener <rguenther@suse.de>
8532 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8533 for operations we cannot scalarize.
8535 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8537 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8538 vectors heap vectors. Clean up comments.
8539 Make visited_bbs a reference.
8540 (profitable_jump_thread_path): Make GC
8541 vectors heap vectors. Clean up comments.
8543 (convert_and_register_jump_thread_path): Make GC vectors heap
8545 (check_subpath_and_update_thread_path): Same. Clean up comments.
8546 Make visited_bbs a reference.
8547 (handle_phi): Abstract common code to to
8548 register_jump_thread_path_if_profitable.
8549 Rename VAR_BB to DEF_BB.
8551 Make GC vectors heap vectors.
8552 Make visited_bbs a reference.
8553 (handle_assignment): Same.
8554 (register_jump_thread_path_if_profitable): New.
8555 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8557 Make GC vectors heap vectors. Clean up comments.
8558 Make visited_bbs a reference.
8559 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8560 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8563 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8566 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8567 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8569 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8571 Revert r251800 and r251799.
8573 2017-09-11 Martin Jambor <mjambor@suse.cz>
8576 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8577 arguments in advance.
8578 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8579 use it to find predecessor edges.
8580 (naive_outof_ssa): Collect vector of predecessors.
8582 2017-09-08 Jason Merrill <jason@redhat.com>
8584 PR c++/70029 - ICE with ref-qualifier and -flto
8585 * langhooks.h (struct lang_hooks_for_types): Add
8586 copy_lang_qualifiers.
8587 * attribs.c (build_type_attribute_qual_variant): Use it.
8588 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8590 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8591 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8593 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8596 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8597 (*mulsi3_sp64): New instruction.
8598 (mulsi3): New expander.
8600 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8602 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8604 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8606 * sancov.c: Include memmodel.h.
8608 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8611 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8614 2017-09-07 Carl Love <cel@us.ibm.com>
8616 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8617 the sldi instruction.
8619 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8621 * sancov.c: Include tm_p.h.
8623 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8626 * output.h (switch_to_other_text_partition): New declaration.
8627 * varasm.c (switch_to_other_text_partition): New function.
8628 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8629 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8630 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8631 to the other text partition before emitting LCL label and switch back
8632 after emitting the word after it.
8634 2017-09-07 Richard Biener <rguenther@suse.de>
8636 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8637 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8638 critical edges here, after loop init.
8639 (pass_data_pre): Remove PROP_no_crit_edges flags.
8640 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8641 for valueization of call args to avoid leaking VN_TOP.
8642 (visit_use): Assert we do not visit default defs.
8643 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8644 Use error_mark_node to more easily detect leaking VN_TOP.
8645 All default-defs are varying, not VN_TOP. Mark them visited.
8646 (run_scc_vn): Make code match comment.
8648 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8650 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8651 OPTION_MASK_FLOAT128_KEYWORD.
8652 (POWERPC_MASKS): Likewise.
8653 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8654 support for the -mfloat128-type option, and make -mfloat128
8655 default on PowerPC Linux systems. Define or undefine
8656 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8657 Define __float128 to be __ieee128 if IEEE 128-bit support is
8658 enabled, or undefine it.
8659 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8660 Delete defining __FLOAT128_TYPE__.
8661 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8662 -mfloat128-type option and make -mfloat128 default on PowerPC
8664 (TARGET_FLOAT128_TYPE): Likewise.
8665 (-mfloat128-type): Likewise.
8666 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8667 Delete the -mfloat128-type option and make -mfloat128 default on
8668 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8669 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8670 128-bit floating point is enabled. Change tests from using
8671 -mfloat128-type to -mfloat128.
8672 (rs6000_mangle_type): Use the correct mangling for the __float128
8673 type even if normal long double is restricted to 64-bits.
8674 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8676 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8677 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8678 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8680 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8681 documentation for -mfloat128.
8683 2017-09-06 Olivier Hainque <hainque@adacore.com>
8685 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8687 2017-09-06 Wish Wu <wishwu007@gmail.com>
8688 Jakub Jelinek <jakub@redhat.com>
8690 * asan.c (initialize_sanitizer_builtins): Add
8691 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8692 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8693 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8694 BT_FN_VOID_UINT64_PTR variables.
8695 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8696 (BT_FN_VOID_UINT16_UINT16): Likewise.
8697 (BT_FN_VOID_UINT32_UINT32): Likewise.
8698 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8699 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8700 (BT_FN_VOID_UINT64_PTR): Likewise.
8701 * common.opt (flag_sanitize_coverage): New variable.
8702 (fsanitize-coverage=trace-pc): Remove.
8703 (fsanitize-coverage=): Add.
8704 * flag-types.h (enum sanitize_coverage_code): New enum.
8705 * fold-const.c (fold_range_test): Disable non-short-circuit
8706 optimization if flag_sanitize_coverage.
8707 (fold_truth_andor): Likewise.
8708 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8709 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8710 (coverage_sanitizer_opts): New array.
8711 (get_closest_sanitizer_option): Add OPTS argument, handle also
8712 OPT_fsanitize_coverage_.
8713 (parse_sanitizer_options): Adjusted to also handle
8714 OPT_fsanitize_coverage_.
8715 (common_handle_option): Add OPT_fsanitize_coverage_.
8716 * sancov.c (instrument_comparison, instrument_switch): New function.
8717 (sancov_pass): Add trace-cmp support.
8718 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8719 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8720 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8721 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8722 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8723 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8724 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8725 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8726 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8727 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8729 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8731 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8732 error. Only quit immediately if parsing is complete.
8733 (BEGIN): Initialize fatal_err and parse_done.
8734 (begin fpu, end fpu): Check number of arguments.
8735 (begin arch, end arch): Likewise.
8736 (begin cpu, end cpu): Likewise.
8737 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8738 (optalias): Likewise.
8740 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8742 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8743 * config/arm/arm-isa.h: Delete. Move definitions to ...
8744 * arm-cpus.in: ... here. Use new feature and fgroup values.
8745 * config/arm/arm.c (arm_option_override): Use lower case for feature
8747 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8748 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8749 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8751 (print_isa_bits_for): New function.
8752 (gen_isa): New function.
8753 (gen_comm_data): Use print_isa_bits_for.
8754 (define feature): New keyword.
8755 (define fgroup): New keyword.
8756 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8757 (arm-isa.h): Add rule to generate file.
8758 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8759 case for feature bit names.
8761 2017-09-06 Richard Biener <rguenther@suse.de>
8763 * tree-ssa-pre.c (NECESSARY): Remove.
8764 (create_expression_by_pieces): Do not touch pass-local flags.
8765 (insert_into_preds_of_block): Likewise.
8766 (do_pre_regular_insertion): Likewise.
8767 (eliminate_insert): Likewise.
8768 (eliminate_dom_walker::before_dom_children): Likewise.
8769 (fini_eliminate): Do not look at inserted_exprs.
8770 (mark_operand_necessary): Remove.
8771 (remove_dead_inserted_code): Replace with simple work-list
8772 algorithm based on inserted_exprs and SSA uses.
8773 (pass_pre::execute): Re-order fini_eliminate and
8774 remove_dead_inserted_code.
8776 2017-09-06 Olivier Hainque <hainque@adacore.com>
8778 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8779 for VxWorks 7. Adjust surrounding comments.
8781 2017-09-06 Richard Biener <rguenther@suse.de>
8783 * gimple-ssa-strength-reduction.c
8784 (find_candidates_dom_walker::before_dom_children): Also allow
8787 2017-09-06 Richard Biener <rguenther@suse.de>
8789 PR tree-optimization/82108
8790 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8791 for gap in the non-permutation SLP case.
8793 2017-09-06 Martin Jambor <mjambor@suse.cz>
8795 PR tree-optimization/82078
8796 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8797 add_access_to_work_queue...
8798 (build_accesses_from_assign): ...here.
8799 (propagate_all_subaccesses): Make sure racc is the group
8800 representative, if there is one.
8802 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8805 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8808 2017-09-06 Richard Biener <rguenther@suse.de>
8810 * gimple-ssa-strength-reduction.c
8811 (find_candidates_dom_walker::before_doom_children): Use a
8812 type and not a mode check.
8814 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8817 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8819 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8821 2017-09-05 Jeff Law <law@redhat.com>
8823 PR tree-optimization/64910
8824 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8825 cases where we have 3 or more operands.
8827 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8830 * omp-low.c (lower_omp_for): Recompute tree invariant if
8831 gimple_omp_for_initial/final is ADDR_EXPR.
8834 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8835 into gimple val before gimplification fo the COND_EXPR.
8837 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8839 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8840 REGION_COPY argument.
8841 (thread_through_all_blocks): Remove unused argument to
8842 duplicate_thread_path.
8844 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8845 Alan Hayward <alan.hayward@arm.com>
8846 David Sherwood <david.sherwood@arm.com>
8848 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8849 Take a scalar_mode rather than a machine_mode.
8850 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8851 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8852 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8853 (aarch64_gen_adjusted_ldpstp): Likewise.
8854 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8856 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8857 Alan Hayward <alan.hayward@arm.com>
8858 David Sherwood <david.sherwood@arm.com>
8860 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8861 Take a scalar_int_mode instead of a machine_mode.
8862 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8863 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8864 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8865 (aarch64_simd_attr_length_rglist): Delete.
8866 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8867 a scalar_int_mode instead of a machine_mode.
8868 (aarch64_add_offset): Likewise.
8869 (aarch64_internal_mov_immediate): Likewise
8870 (aarch64_add_constant_internal): Likewise.
8871 (aarch64_add_constant): Likewise.
8872 (aarch64_movw_imm): Likewise.
8873 (aarch64_rtx_arith_op_extract_p): Likewise.
8874 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8875 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8876 Remove assert that the mode isn't a vector.
8877 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8878 (aarch64_expand_mov_immediate): Update calls after above changes.
8879 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8880 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8881 (aarch64_move_imm): Likewise.
8882 (aarch64_can_const_movi_rtx_p): Likewise.
8883 (aarch64_strip_extend): Likewise.
8884 (aarch64_extr_rtx_p): Likewise.
8885 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8886 a CONST_INT when the mode parameter is VOIDmode.
8887 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8889 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8891 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8892 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8893 (bitwise_type_for_mode): Update accordingly.
8895 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8897 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8898 * stor-layout.c (mode_for_size_tree): Likewise.
8899 (mode_for_array): Update accordingly.
8900 (layout_decl): Likewise.
8901 (compute_record_mode): Likewise. Only set the mode once.
8903 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8905 * target.def (get_mask_mode): Change return type to opt_mode.
8907 * doc/tm.texi: Regenerate.
8908 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8909 * targhooks.c (default_get_mask_mode): Likewise.
8910 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8911 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8912 targetm.get_mask_mode.
8913 * tree.c (build_truth_vector_type): Likewise.
8915 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8917 * machmode.h (mode_for_vector): Return an opt_mode.
8918 * stor-layout.c (mode_for_vector): Likewise.
8919 (mode_for_int_vector): Update accordingly.
8920 (layout_type): Likewise.
8921 * config/i386/i386.c (emit_memmov): Likewise.
8922 (ix86_expand_set_or_movmem): Likewise.
8923 (ix86_expand_vector_init): Likewise.
8924 (ix86_get_mask_mode): Likewise.
8925 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8927 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8928 * expmed.c (extract_bit_field_1): Likewise.
8929 * expr.c (expand_expr_real_2): Likewise.
8930 * optabs-query.c (can_vec_perm_p): Likewise.
8931 (can_vec_mask_load_store_p): Likewise.
8932 * optabs.c (expand_vec_perm): Likewise.
8933 * targhooks.c (default_get_mask_mode): Likewise.
8934 * tree-vect-stmts.c (vectorizable_store): Likewise.
8935 (vectorizable_load): Likewise.
8936 (get_vectype_for_scalar_type_and_size): Likewise.
8938 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8940 * machmode.h (mode_for_int_vector): New function.
8941 * stor-layout.c (mode_for_int_vector): Likewise.
8942 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8943 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8944 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8945 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8946 (s390_expand_vcond): Likewise.
8948 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8950 * machmode.h (opt_machine_mode): New type.
8951 (opt_mode<T>): Allow construction from anything that can be
8953 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8954 (mode_for_size): Return an opt_machine_mode.
8955 * stor-layout.c (mode_for_size): Likewise.
8956 (mode_for_size_tree): Update call accordingly.
8957 (bitwise_mode_for_mode): Likewise.
8958 (make_fract_type): Likewise.
8959 (make_accum_type): Likewise.
8960 * caller-save.c (replace_reg_with_saved_mem): Update call
8962 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8963 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8964 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8965 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8966 * expmed.c (extract_bit_field_1): Likewise.
8967 * reload.c (get_secondary_mem): Likewise.
8968 * varasm.c (assemble_integer): Likewise.
8969 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8972 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8974 * machmode.h (decimal_float_mode_for_size): New function.
8975 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8976 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8977 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8978 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8979 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8980 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8982 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8984 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8985 (get_builtin_sync_mode): Likewise.
8986 (expand_ifn_atomic_compare_exchange): Likewise.
8987 (expand_builtin_atomic_clear): Likewise.
8988 (expand_builtin_atomic_test_and_set): Likewise.
8989 (fold_builtin_atomic_always_lock_free): Likewise.
8990 * calls.c (compute_argument_addresses): Likewise.
8991 (emit_library_call_value_1): Likewise.
8992 (store_one_arg): Likewise.
8993 * combine.c (combine_instructions): Likewise.
8994 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8995 * config/arm/arm.c (arm_function_value): Likewise.
8996 (aapcs_allocate_return_reg): Likewise.
8997 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8998 * config/i386/i386.c (construct_container): Likewise.
8999 (ix86_gimplify_va_arg): Likewise.
9000 (ix86_expand_sse_cmp): Likewise.
9001 (emit_memmov): Likewise.
9002 (emit_memset): Likewise.
9003 (expand_small_movmem_or_setmem): Likewise.
9004 (ix86_expand_pextr): Likewise.
9005 (ix86_expand_pinsr): Likewise.
9006 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9007 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9009 * config/mips/mips.c (mips_function_value_1) Likewise.
9010 (mips_block_move_straight): Likewise.
9011 (mips_expand_ins_as_unaligned_store): Likewise.
9012 * config/powerpcspe/powerpcspe.c
9013 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9014 (rs6000_darwin64_record_arg_flush): Likewise.
9015 * config/rs6000/rs6000.c
9016 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9017 (rs6000_darwin64_record_arg_flush): Likewise.
9018 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9019 (sparc_function_value_1): Likewise.
9020 * config/spu/spu.c (adjust_operand): Likewise.
9021 (spu_emit_branch_or_set): Likewise.
9022 (arith_immediate_p): Likewise.
9023 * emit-rtl.c (gen_lowpart_common): Likewise.
9024 * expr.c (expand_expr_real_1): Likewise.
9025 * function.c (assign_parm_setup_block): Likewise.
9026 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9027 * reload1.c (alter_reg): Likewise.
9028 * stor-layout.c (mode_for_vector): Likewise.
9029 (layout_type): Likewise.
9031 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9033 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9034 (spu_convert_move): Likewise.
9035 * lower-subreg.c (resolve_simple_move): Likewise.
9037 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9040 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9041 define_insn to a define_expand.
9042 (altivec_vsum2sws_direct): New define_insn.
9043 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9045 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9047 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9050 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9055 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9056 realign stack if stack alignment needed is less than incoming
9059 2017-09-05 Marek Polacek <polacek@redhat.com>
9062 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9065 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9067 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9069 2017-09-05 Richard Biener <rguenther@suse.de>
9071 PR tree-optimization/82084
9072 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9074 2017-09-05 Richard Biener <rguenther@suse.de>
9076 PR tree-optimization/82102
9077 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9079 2017-09-05 Martin Liska <mliska@suse.cz>
9081 PR tree-optimization/82032
9082 * tree-cfg.c (generate_range_test): New function.
9083 * tree-cfg.h (generate_range_test): Declared here.
9084 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9085 (cleanup_control_expr_graph): Use it.
9086 * tree-switch-conversion.c (try_switch_expansion): Remove
9088 (emit_case_nodes): Use generate_range_test.
9090 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9093 * config/i386/i386.md (*<btsc><mode>_mask): Add
9094 TARGET_USE_BT to insn constraint.
9095 (*btr<mode>_mask): Ditto.
9097 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9099 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9100 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9102 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9105 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9106 TARGET_NEON and TARGET_IWMMXT.
9107 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9108 TARGET_NEON and TARGET_IWMMXT.
9109 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9111 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9113 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9114 (ix86_rewrite_tls_address): Ditto.
9115 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9116 (ix86_rewrite_tls_address_1): Ditto.
9117 (ix86_rewrite_tls_address): Ditto.
9118 * config/i386/predicates.md (tls_address_pattern): New predicate.
9119 * config/i386/i386.md (TLS address splitter): New splitter.
9121 2017-09-04 Richard Biener <rguenther@suse.de>
9123 PR tree-optimization/82084
9124 * fold-const.h (can_native_encode_string_p): Declare.
9125 * fold-const.c (can_native_encode_string_p): Factor out from ...
9126 (native_encode_string): ... here.
9127 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9128 vectorizing stores from constants we later cannot handle.
9130 2017-09-04 Marek Polacek <polacek@redhat.com>
9133 * doc/invoke.texi: Update -Wtautological-compare documentation.
9135 2017-09-04 Jeff Law <law@redhat.com>
9137 PR tree-optimization/64910
9138 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9139 swap the first and last operand if the last is a constant.
9141 2017-09-04 Marek Polacek <polacek@redhat.com>
9144 * convert.c (do_narrow): When sanitizing signed integer overflows,
9145 bail out for signed types.
9146 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9148 2017-09-04 Richard Biener <rguenther@suse.de>
9150 PR tree-optimization/82060
9151 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9152 Move devirtualization after stmt folding and before EH/AB/noreturn
9153 cleanup to get the stmt refs canonicalized. Use a bool instead
9154 of gimple_modified_p since that doesn't work for NOPs. Schedule
9155 NOPs generated by folding for removal.
9157 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9158 Alan Hayward <alan.hayward@arm.com>
9159 David Sherwood <david.sherwood@arm.com>
9161 * coretypes.h (pad_direction): New enum.
9162 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9163 (FUNCTION_ARG_PADDING): Likewise.
9164 * target.def (function_arg_padding): New hook.
9165 * targhooks.h (default_function_arg_padding): Declare.
9166 * targhooks.c (default_function_arg_padding): New function.
9167 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9168 (TARGET_FUNCTION_ARG_PADDING): ...this.
9169 * doc/tm.texi: Regenerate.
9170 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9171 instead of direction.
9172 (compute_argument_addresses): Likewise.
9173 (load_register_parameters): Likewise.
9174 (emit_library_call_value_1): Likewise.
9175 (store_one_arg): Use targetm.calls.function_arg_padding instead
9176 of FUNCTION_ARG_PADDING.
9177 (must_pass_in_stack_var_size_or_pad): Likewise.
9178 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9179 (emit_group_store): Likewise.
9180 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9181 instead of FUNCTION_ARG_PADDING.
9182 (emit_push_insn): Likewise, and propagate enum change throughout
9184 * function.h (direction): Delete.
9185 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9187 * function.c (assign_parm_find_stack_rtl): Likewise.
9188 (assign_parm_setup_block_p): Likewise.
9189 (assign_parm_setup_block): Likewise.
9190 (gimplify_parameters): Likewise.
9191 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9192 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9194 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9195 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9196 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9197 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9198 (aarch64_function_arg_padding): ...this new function.
9199 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9200 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9201 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9202 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9203 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9204 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9205 (arm_pad_arg_upward): Replace with...
9206 (arm_function_arg_padding): ...this new function.
9207 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9209 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9210 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9211 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9212 (ia64_hpux_function_arg_padding): Replace with...
9213 (ia64_function_arg_padding): ...this new function. Use pad_direction
9214 instead of direction. Check for TARGET_HPUX.
9215 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9216 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9217 (iq2000_function_arg_padding): New function.
9218 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9219 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9220 (mips_function_arg_padding): ...this new function.
9221 (mips_pad_reg_upward): Update accordingly.
9222 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9223 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9224 targetm.calls.function_arg_padding.
9225 (FUNCTION_ARG_PADDING): Delete.
9226 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9227 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9228 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9229 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9230 (nios2_block_reg_padding): Return pad_direction instead of direction.
9231 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9232 instead of direction.
9233 (nios2_function_arg_padding): Likewise. Make static.
9234 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9235 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9236 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9237 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9238 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9239 (pa_function_arg_padding): Make static. Return pad_direction instead
9241 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9242 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9243 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9244 instead of direction. Use targetm.calls.function_arg_padding.
9245 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9246 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9247 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9248 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9249 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9251 (function_arg_padding): Rename to...
9252 (rs6000_function_arg_padding): ...this. Make static. Return
9253 pad_direction instead of direction.
9254 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9255 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9256 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9257 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9258 instead of direction. Use targetm.calls.function_arg_padding.
9259 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9260 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9261 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9262 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9263 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9264 (function_arg_padding): Rename to...
9265 (rs6000_function_arg_padding): ...this. Make static. Return
9266 pad_direction instead of direction.
9267 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9268 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9269 * config/s390/s390.c (s390_function_arg_padding): New function.
9270 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9271 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9272 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9273 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9274 (function_arg_padding): Rename to...
9275 (sparc_function_arg_padding): ...this. Make static. Return
9276 pad_direction instead of direction.
9277 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9278 * config/spu/spu.c (spu_function_arg_padding): New function.
9279 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9280 * system.h (FUNCTION_ARG_PADDING): Poison.
9282 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9283 Alan Hayward <alan.hayward@arm.com>
9284 David Sherwood <david.sherwood@arm.com>
9286 * target.def (modes_tieable_p): New hook.
9287 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9288 (TARGET_MODES_TIEABLE_P): ...this.
9289 * doc/tm.texi.in: Regenerate.
9290 * hooks.h (hook_bool_mode_mode_true): Declare.
9291 * hooks.c (hook_bool_mode_mode_true): New function.
9292 * combine.c (subst): Use targetm.modes_tieable_p instead of
9294 * dse.c (find_shift_sequence): Likewise.
9295 * expmed.c (extract_low_bits): Likewise.
9296 * lower-subreg.c: Include target.h.
9297 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9299 * rtlanal.c (rtx_cost): Likewise.
9300 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9301 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9302 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9303 (TARGET_MODES_TIEABLE_P): Redefine.
9304 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9305 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9306 (TARGET_MODES_TIEABLE_P): Redefine.
9307 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9308 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9309 (arc_modes_tieable_p): New function.
9310 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9311 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9312 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9313 (arm_modes_tieable_p): Make static.
9314 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9315 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9316 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9317 (TARGET_MODES_TIEABLE_P): Redefine.
9318 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9319 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9320 (TARGET_MODES_TIEABLE_P): Redefine.
9321 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9322 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9323 (cr16_modes_tieable_p): New function.
9324 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9325 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9326 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9327 (TRULY_NOOP_TRUNCATION): Update comment.
9328 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9329 (TRULY_NOOP_TRUNCATION): Update comment.
9330 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9331 (frv_modes_tieable_p): New function.
9332 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9333 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9334 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9335 (TARGET_MODES_TIEABLE_P): Redefine.
9336 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9337 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9338 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9339 (TARGET_MODES_TIEABLE_P): Redefine.
9340 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9341 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9342 (ia64_modes_tieable_p): New function.
9343 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9344 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9345 (iq2000_modes_tieable_p): New function.
9346 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9347 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9348 (lm32_modes_tieable_p): New function.
9349 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9350 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9351 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9352 (TARGET_MODES_TIEABLE_P): Redefine.
9353 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9354 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9355 (m32r_modes_tieable_p): New function.
9356 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9357 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9358 (m68k_modes_tieable_p): New function.
9359 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9360 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9361 (mcore_modes_tieable_p): New function.
9362 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9363 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9365 (TARGET_MODES_TIEABLE_P): Redefine.
9366 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9367 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9368 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9369 (TARGET_MODES_TIEABLE_P): Redefine.
9370 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9371 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9372 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9373 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9374 (mn10300_modes_tieable_p): ...this and make static.
9375 (TARGET_MODES_TIEABLE_P): Redefine.
9376 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9377 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9378 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9379 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9380 (msp430_modes_tieable_p): Make static.
9381 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9382 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9383 (TARGET_MODES_TIEABLE_P): Redefine.
9384 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9385 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9386 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9387 (TARGET_MODES_TIEABLE_P): Redefine.
9388 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9389 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9390 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9391 (TARGET_MODES_TIEABLE_P): Redefine.
9392 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9393 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9394 (pdp11_modes_tieable_p): New function.
9395 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9396 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9397 (rs6000_modes_tieable_p): New function.
9398 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9399 * config/powerpcspe/powerpcspe.md: Update comment.
9400 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9401 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9402 (TARGET_MODES_TIEABLE_P): Redefine.
9403 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9404 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9405 (rl78_modes_tieable_p): New function.
9406 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9407 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9408 (rs6000_modes_tieable_p): New function.
9409 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9410 * config/rs6000/rs6000.md: Update comment.
9411 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9412 * config/rx/rx.c (rx_modes_tieable_p): New function.
9413 (TARGET_MODES_TIEABLE_P): Redefine.
9414 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9415 * config/s390/s390.c (s390_modes_tieable_p): New function.
9416 (TARGET_MODES_TIEABLE_P): Redefine.
9417 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9418 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9419 (sh_modes_tieable_p): New function.
9420 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9421 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9422 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9423 (sparc_modes_tieable_p): Make static.
9424 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9425 * config/spu/spu.c (spu_modes_tieable_p): New function.
9426 (TARGET_MODES_TIEABLE_P): Redefine.
9427 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9428 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9429 (TARGET_MODES_TIEABLE_P): Redefine.
9430 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9431 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9432 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9433 * config/v850/v850.c (v850_modes_tieable_p): New function.
9434 (TARGET_MODES_TIEABLE_P): Redefine.
9435 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9436 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9437 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9438 (visium_modes_tieable_p): New function.
9439 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9440 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9441 (xtensa_modes_tieable_p): New function.
9442 * system.h (MODES_TIEABLE_P): Poison.
9444 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9445 Alan Hayward <alan.hayward@arm.com>
9446 David Sherwood <david.sherwood@arm.com>
9448 * target.def (hard_regno_mode_ok): New hook.
9449 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9450 (TARGET_HARD_REGNO_MODE_OK): ...this.
9451 * doc/tm.texi.in: Regenerate.
9452 * hooks.h (hook_bool_uint_mode_true): Declare.
9453 * hooks.c (hook_bool_uint_mode_true): New function.
9454 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9456 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9457 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9458 instead of HARD_REGNO_MODE_OK.
9459 * caller-save.c: Include target.h.
9460 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9462 * combine.c (can_combine_p): Likewise.
9463 (combinable_i3pat): Likewise.
9464 (can_change_dest_mode): Likewise.
9465 * expr.c (init_expr_target): Likewise.
9466 (convert_move): Likewise.
9467 (convert_modes): Likewise.
9468 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9469 (setup_prohibited_mode_move_regs): Likewise.
9470 * ira.h (target_ira): Likewise.
9471 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9472 * lra-constraints.c (process_alt_operands): Likewise.
9473 (split_reg): Likewise.
9474 * recog.c (peep2_find_free_register): Likewise.
9475 * ree.c (combine_reaching_defs): Likewise.
9476 * regcprop.c (maybe_mode_change): Likewise.
9477 * reginfo.c (init_reg_sets_1): Likewise.
9478 (choose_hard_reg_mode): Likewise.
9479 (simplifiable_subregs): Likewise.
9480 * regrename.c (check_new_reg_p): Likewise.
9481 * reload.c (find_valid_class): Likewise.
9482 (find_valid_class_1): Likewise.
9483 (reload_inner_reg_of_subreg): Likewise.
9484 (push_reload): Likewise.
9485 (combine_reloads): Likewise.
9486 (find_dummy_reload): Likewise.
9487 (find_reloads): Likewise.
9488 * reload1.c (find_reg): Likewise.
9489 (set_reload_reg): Likewise.
9490 (allocate_reload_reg): Likewise.
9491 (choose_reload_regs): Likewise.
9492 (reload_adjust_reg_for_temp): Likewise.
9493 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9494 (simplify_subreg_regno): Likewise.
9495 * sel-sched.c (init_regs_for_mode): Likewise.
9496 * varasm.c (make_decl_rtl): Likewise.
9497 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9498 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9500 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9501 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9502 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9503 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9504 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9505 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9506 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9507 (arc_mode_class): Delete.
9508 (HARD_REGNO_MODE_OK): Delete.
9509 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9510 (arc_hard_regno_mode_ok): Rename old array to...
9511 (arc_hard_regno_mode_ok_modes): ...this.
9512 (arc_conditional_register_usage): Update accordingly.
9513 (arc_mode_class): Make static.
9514 (arc_hard_regno_mode_ok): New function.
9515 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9516 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9517 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9518 (arm_hard_regno_mode_ok): Make static.
9519 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9521 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9522 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9523 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9525 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9526 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9527 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9528 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9529 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9530 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9531 * config/bfin/predicates.md (valid_reg_operand): Use
9532 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9533 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9534 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9535 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9536 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9537 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9538 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9539 (cr16_hard_regno_mode_ok): Make static and return a bool.
9540 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9541 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9542 (cris_hard_regno_mode_ok): New function.
9543 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9544 (epiphany_mode_class): Delete.
9545 (HARD_REGNO_MODE_OK): Delete.
9546 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9547 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9548 (hard_regno_mode_ok): Rename to...
9549 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9550 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9551 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9553 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9554 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9555 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9556 (frv_hard_regno_mode_ok): Make static and return a bool.
9557 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9559 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9560 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9561 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9562 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9564 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9565 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9566 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9567 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9569 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9570 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9571 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9572 (ia64_hard_regno_mode_ok): New function.
9573 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9574 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9575 (iq2000_hard_regno_mode_ok): New function.
9576 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9577 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9578 (lm32_hard_regno_mode_ok): New function.
9579 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9580 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9581 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9582 instead of HARD_REGNO_MODE_OK.
9583 (m32c_hard_regno_ok): Rename to...
9584 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9585 (m32c_cannot_change_mode_class): Update accordingly.
9586 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9587 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9588 (m32r_mode_class): Delete.
9589 (HARD_REGNO_MODE_OK): Delete.
9590 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9591 (m32r_hard_regno_mode_ok): Rename to...
9592 (m32r_hard_regno_modes): ...this.
9593 (m32r_mode_class): Make static.
9594 (m32r_hard_regno_mode_ok): New function.
9595 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9596 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9597 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9598 (m68k_hard_regno_mode_ok): Make static.
9599 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9600 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9601 (mcore_hard_regno_mode_ok): New function.
9602 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9603 (HARD_REGNO_MODE_OK): Delete.
9604 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9606 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9607 (microblaze_hard_regno_mode_ok): New function.
9608 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9609 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9610 (mips_hard_regno_mode_ok): Delete.
9611 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9612 (mips_hard_regno_mode_ok_p): ...this and make static.
9613 (mips_hard_regno_mode_ok_p): Rename to...
9614 (mips_hard_regno_mode_ok_uncached): ...this.
9615 (mips_hard_regno_mode_ok): New function.
9616 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9617 of HARD_REGNO_MODE_OK.
9618 (mips_option_override): Update after above name changes.
9619 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9620 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9621 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9622 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9623 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9624 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9625 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9626 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9627 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9628 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9629 (msp430_hard_regno_mode_ok): Make static and return a bool.
9630 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9631 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9632 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9634 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9635 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9636 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9637 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9638 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9639 (PA_HARD_REGNO_MODE_OK): ...this
9640 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9641 (PA_HARD_REGNO_MODE_OK): ...this.
9642 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9643 (pa_hard_regno_mode_ok): New function.
9644 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9645 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9646 (pdp11_hard_regno_mode_ok): New function.
9647 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9648 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9650 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9652 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9653 (rs6000_hard_regno_mode_ok): Rename to...
9654 (rs6000_hard_regno_mode_ok_uncached): ...this.
9655 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9656 (rs6000_hard_regno_mode_ok): New function.
9657 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9658 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9659 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9660 (riscv_hard_regno_mode_ok): ...this and make static.
9661 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9662 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9663 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9664 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9665 (rl78_hard_regno_mode_ok): Make static and return bool.
9666 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9667 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9669 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9670 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9671 (rs6000_hard_regno_mode_ok): Rename to...
9672 (rs6000_hard_regno_mode_ok_uncached): ...this.
9673 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9674 (rs6000_hard_regno_mode_ok): New function.
9675 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9676 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9677 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9678 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9679 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9680 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9681 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9682 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9683 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9684 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9685 (sh_hard_regno_mode_ok): Make static.
9686 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9687 instead of HARD_REGNO_MODE_OK.
9688 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9689 (sparc_mode_class): Delete.
9690 (HARD_REGNO_MODE_OK): Delete.
9691 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9692 (hard_regno_mode_classes): Make static.
9693 (sparc_mode_class): Likewise.
9694 (sparc_hard_regno_mode_ok): New function.
9695 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9696 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9697 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9699 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9700 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9701 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9702 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9703 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9704 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9705 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9706 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9707 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9708 (visium_hard_regno_mode_ok): New function.
9709 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9710 instead of HARD_REGNO_MODE_OK.
9711 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9712 (HARD_REGNO_MODE_OK): Delete.
9713 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9714 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9715 (xtensa_option_override): Update accordingly.
9716 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9717 (xtensa_hard_regno_mode_ok): New function.
9718 * system.h (HARD_REGNO_MODE_OK): Poison.
9720 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9721 Alan Hayward <alan.hayward@arm.com>
9722 David Sherwood <david.sherwood@arm.com>
9724 * target.def (hard_regno_call_part_clobbered): New hook.
9725 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9726 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9727 * doc/tm.texi: Regenerate.
9728 * hooks.h (hook_bool_uint_mode_false): Declare.
9729 * hooks.c (hook_bool_uint_mode_false): New function.
9730 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9731 * cselib.c (cselib_process_insn): Use
9732 targetm.hard_regno_call_part_clobbered instead of
9733 HARD_REGNO_CALL_PART_CLOBBERED.
9734 * ira-conflicts.c (ira_build_conflicts): Likewise.
9735 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9736 * lra-constraints.c (need_for_call_save_p): Likewise.
9737 * lra-lives.c: Include target.h.
9738 (check_pseudos_live_through_calls): Use
9739 targetm.hard_regno_call_part_clobbered instead of
9740 HARD_REGNO_CALL_PART_CLOBBERED.
9741 * regcprop.c: Include target.h.
9742 (copyprop_hardreg_forward_1): Use
9743 targetm.hard_regno_call_part_clobbered instead of
9744 HARD_REGNO_CALL_PART_CLOBBERED.
9745 * reginfo.c (choose_hard_reg_mode): Likewise.
9746 * regrename.c (check_new_reg_p): Likewise.
9747 * reload.c (find_equiv_reg): Likewise.
9748 * reload1.c (emit_reload_insns): Likewise.
9749 * sched-deps.c (deps_analyze_insn): Likewise.
9750 * sel-sched.c (init_regs_for_mode): Likewise.
9751 (mark_unavailable_hard_regs): Likewise.
9752 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9753 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9754 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9756 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9757 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9758 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9760 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9762 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9763 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9764 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9766 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9767 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9768 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9770 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9771 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9773 * config/powerpcspe/powerpcspe.c
9774 (rs6000_hard_regno_call_part_clobbered): New function.
9775 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9776 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9777 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9779 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9780 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9781 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9783 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9784 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9785 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9787 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9788 Alan Hayward <alan.hayward@arm.com>
9789 David Sherwood <david.sherwood@arm.com>
9791 * rtl.h (subreg_memory_offset): Declare.
9792 * emit-rtl.c (subreg_memory_offset): New function.
9793 * expmed.c (store_bit_field_1): Use it.
9794 * expr.c (undefined_operand_subword_p): Likewise.
9795 * simplify-rtx.c (simplify_subreg): Likewise.
9797 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9799 PR rtl-optimization/57448
9802 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9803 using atomic_load pattern.
9804 (expand_atomic_store): Likewise.
9806 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9809 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9810 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9811 handling. Use replace_call_with_value with NULL instead of
9812 gsi_replace, unlink_stmt_vdef and release_defs.
9814 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9817 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9819 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9822 * rtl.h (emit_library_call_value_1): Declare.
9823 (emit_library_call): Replace declaration with a series of overloads.
9824 Remove the parameter count argument.
9825 (emit_library_call_value): Likewise.
9826 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9827 with an "rtx_mode_t *".
9828 (emit_library_call_value): Delete.
9829 (emit_library_call): Likewise.
9830 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9831 (asan_emit_allocas_unpoison): Likewise.
9832 * builtins.c (expand_builtin_powi): Likewise.
9833 (expand_asan_emit_allocas_unpoison): Likewise.
9834 * cfgexpand.c (expand_main_function): Likewise.
9835 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9836 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9837 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9838 * config/arm/arm.c (arm_trampoline_init): Likewise.
9839 (arm_call_tls_get_addr): Likewise.
9840 (arm_expand_divmod_libfunc): Likewise.
9841 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9842 (smulsi3_highpart): Likewise.
9843 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9844 (c6x_expand_compare): Likewise.
9845 (c6x_expand_movmem): Likewise.
9846 * config/frv/frv.c (frv_trampoline_init): Likewise.
9847 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9848 (ix86_expand_divmod_libfunc): Likewise.
9849 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9850 (ia64_expand_compare): Likewise.
9851 (ia64_profile_hook): Likewise.
9852 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9853 (nonlocal_goto): Likewise.
9854 (restore_stack_nonlocal): Likewise.
9855 * config/m32r/m32r.c (block_move_call): Likewise.
9856 (m32r_trampoline_init): Likewise.
9857 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9858 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9859 (m68k_call_m68k_read_tp): Likewise.
9860 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9861 (microblaze_expand_divide): Likewise.
9862 * config/mips/mips.h (mips_args): Likewise.
9863 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9864 (MIPS_ICACHE_SYNC): Likewise.
9865 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9866 (nios2_trampoline_init): Likewise.
9867 * config/pa/pa.c (hppa_tls_call): Likewise.
9868 (pa_trampoline_init): Likewise.
9869 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9870 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9871 (expand_strn_compare): Likewise.
9872 (rs6000_generate_compare): Likewise.
9873 (rs6000_expand_float128_convert): Likewise.
9874 (output_profile_hook): Likewise.
9875 (rs6000_trampoline_init): Likewise.
9876 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9877 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9878 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9879 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9880 (rs6000_generate_compare): Likewise.
9881 (rs6000_expand_float128_convert): Likewise.
9882 (output_profile_hook): Likewise.
9883 (rs6000_trampoline_init): Likewise.
9884 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9885 * config/sh/sh.c (sh_trampoline_init): Likewise.
9886 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9887 (sparc_emit_float_lib_cmp): Likewise.
9888 (sparc32_initialize_trampoline): Likewise.
9889 (sparc64_initialize_trampoline): Likewise.
9890 (sparc_profile_hook): Likewise.
9891 * config/spu/spu.c (ea_load_store): Likewise.
9892 * config/spu/spu.md (floatunssidf2): Likewise.
9893 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9894 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9895 * config/visium/visium.c (expand_block_move_4): Likewise.
9896 (expand_block_move_2): Likewise.
9897 (expand_block_move_1): Likewise.
9898 (expand_block_set_4): Likewise.
9899 (expand_block_set_2): Likewise.
9900 (expand_block_set_1): Likewise.
9901 (visium_trampoline_init): Likewise.
9902 (visium_profile_hook): Likewise.
9903 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9904 (xtensa_setup_frame_addresses): Likewise.
9905 (xtensa_trampoline_init): Likewise.
9906 * except.c (sjlj_emit_function_enter): Likewise.
9907 (sjlj_emit_function_exit): Likewise.
9908 * explow.c (allocate_dynamic_stack_space): Likewise.
9909 (probe_stack_range): Likewise.
9910 * expr.c (convert_mode_scalar): Likewise.
9911 * optabs.c (expand_binop): Likewise.
9912 (expand_twoval_binop_libfunc): Likewise.
9913 (expand_unop): Likewise.
9914 (prepare_cmp_insn): Likewise.
9915 (prepare_float_lib_cmp): Likewise.
9916 (expand_float): Likewise.
9917 (expand_fix): Likewise.
9918 (expand_fixed_convert): Likewise.
9919 (maybe_emit_sync_lock_test_and_set): Likewise.
9920 (expand_atomic_compare_and_swap): Likewise.
9921 (expand_mem_thread_fence): Likewise.
9922 (expand_atomic_fetch_op): Likewise.
9924 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9926 * doc/generic.texi (OpenACC): Adjust URL.
9927 * doc/invoke.texi (C Dialect Options): Ditto.
9929 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9931 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9932 predicate for operand 1. Add (m,<S>) constraint.
9933 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9934 Prevent memory operand 1 with register operand 2.
9936 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9938 PR rtl-optimization/82024
9939 * combine.c (try_combine): If the combination result is a PARALLEL,
9940 and we only need to retain the SET in there that would be placed
9941 at I2, check that we can place that at I3 instead, before doing so.
9943 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9946 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9948 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9949 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9952 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9953 Jeff Law <law@redhat.com>
9955 * varasm.c (bss_initializer_p): Do not put constants into .bss
9956 (categorize_decl_for_section): Handle bss_initializer_p returning
9957 false when DECL_INITIAL is NULL.
9959 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9962 * config/s390/s390.c (s390_can_inline_p): New function.
9964 2017-09-01 Jeff Law <law@redhat.com>
9966 PR tree-optimization/82052
9967 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9968 Always initialize the returned slot after a hash table miss
9969 when INSERT is true.
9971 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9973 * config/s390/s390.md (mem_signal_fence): Remove.
9974 * doc/md.texi (mem_signal_fence): Remove.
9975 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9977 * target-insns.def (mem_signal_fence): Remove.
9979 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9982 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9985 * asan.c (create_odr_indicator): Strip name encoding from assembler
9986 name before appending it after __odr_asan_.
9988 2017-09-01 Martin Liska <mliska@suse.cz>
9990 PR tree-optimization/82059
9991 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9992 frequency only when an edge is redirected.
9994 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9996 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9997 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9998 (arc_conditional_register_usage): Remove ARC600 lp_count
10000 (arc_file_start): Emit Tag_ARC_CPU_variation.
10001 (arc_can_use_doloop_p): New conditions to use ZOLs.
10002 (hwloop_fail): New function.
10003 (hwloop_optimize): Likewise.
10004 (hwloop_pattern_reg): Likewise.
10005 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10006 (arc_reorg_loops): New function, calls reorg_loops.
10007 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10008 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10010 (arc_loop_hazard): Remove function, functionality moved into
10012 (arc_hazard): Remove arc_loop_hazard call.
10013 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10014 into hwloop_optimize.
10015 (arc_label_align): Remove ZOL handling.
10016 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10017 * config/arc/arc.md (doloop_begin): Remove pattern.
10018 (doloop_begin_i): Likewise.
10019 (doloop_end_i): Likewise.
10020 (doloop_fallback): Likewise.
10021 (doloop_fallback_m): Likewise.
10022 (doloop_end): Reimplement expand.
10023 (arc_lp): New pattern for LP instruction.
10024 (loop_end): New pattern.
10025 (loop_fail): Likewise.
10026 (decrement_and_branch_until_zero): Likewise.
10027 * config/arc/arc.opt (mlpc-width): New option.
10028 * doc/invoke.texi (mlpc-width): Document option.
10030 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10032 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10033 (arc_ccfsm_advance): Fix checking for delay slots.
10034 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10036 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10038 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10040 (movhi_insn): Update store instruction constraint which are saving
10041 6-bit short immediates.
10042 (movsi_insn): Consider also short scaled load operations.
10043 (zero_extendhisi2_i): Use Usd constraint instead of T.
10044 (extendhisi2_i): Add q constraint.
10045 (arc_clzsi2): Add type and length attributes.
10046 (arc_ctzsi2): Likewise.
10047 * config/arc/constraints.md (Usc): Update constraint, the
10048 assembler can parse two relocations for a single instruction.
10050 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10052 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10053 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10055 2017-08-31 Olivier Hainque <hainque@adacore.com>
10057 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10058 match as powerpc-wrs-vxworks*.
10060 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10062 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10063 register constraint for by-element operand.
10064 (aarch64_mls_elt_merge<mode>): Likewise.
10066 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10068 * config/arc/arc.c (arc_can_follow_jump): Check for short
10071 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10073 * config.gcc: Use g.opt for arc.
10074 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10075 functionality moved to ...
10076 (legitimate_scaled_address_p): New function, ...here.
10077 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10078 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10079 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10081 (arc_override_options): Handle G option.
10082 (arc_output_pic_addr_const): Correct function definition.
10083 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10084 (arc_decl_anon_ns_mem_p): Delete.
10085 (arc_in_small_data_p): Overhaul this function to take into
10086 consideration the value given via G option.
10087 (arc_rewrite_small_data_1): Renamed and corrected old
10088 arc_rewrite_small_data function.
10089 (arc_rewrite_small_data): New function.
10090 (small_data_pattern): Don't use pic_offset_table_rtx.
10091 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10092 * config/arc/simdext.md (movmisalignv2hi): Use
10093 prepare_move_operands function.
10095 (movmisalign*): Likewise.
10096 * doc/invoke.texi (ARC options): Document -G option.
10098 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10100 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10102 * config/arc/arc.c (arc_print_operand): Output scalled address for
10103 sdata whenever is possible.
10104 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10105 load/stores are available.
10106 (compact_sda_memory_operand): Check for the alignment required by
10107 code density instructions.
10108 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10110 * config/arc/constraints.md (Usd): Update constraint.
10111 (Us0): New constraint.
10112 (Usc): Update constraint.
10114 2017-08-31 Richard Biener <rguenther@suse.de>
10116 PR middle-end/82054
10117 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10118 function only once.
10120 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10122 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10123 Resize type_signature.
10125 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10126 Alan Hayward <alan.hayward@arm.com>
10127 David Sherwood <david.sherwood@arm.com>
10129 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10130 subregs whose inner modes can be stored in GPRs.
10131 (aarch64_classify_index): Likewise.
10133 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10134 Alan Hayward <alan.hayward@arm.com>
10135 David Sherwood <david.sherwood@arm.com>
10137 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10138 (V_INT_EQUIV): ...this.
10139 (v_cmp_result): Rename to...
10140 (v_int_equiv): ...this.
10141 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10142 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10143 (copysign<mode>3): Likewise.
10144 (aarch64_simd_bsl<mode>_internal): Likewise.
10145 (aarch64_simd_bsl<mode>): Likewise.
10146 (vec_cmp<mode><mode>): Likewise.
10147 (vcond<mode><mode>): Likewise.
10148 (vcond<v_cmp_mixed><mode>): Likewise.
10149 (vcondu<mode><v_cmp_mixed>): Likewise.
10150 (aarch64_cm<optab><mode>): Likewise.
10151 (aarch64_cmtst<mode>): Likewise.
10152 (aarch64_fac<optab><mode>): Likewise.
10153 (vec_perm_const<mode>): Likewise.
10154 (vcond_mask_<mode><v_cmp_result>): Rename to...
10155 (vcond_mask_<mode><v_int_equiv>): ...this.
10156 (vec_cmp<mode><v_cmp_result>): Rename to...
10157 (vec_cmp<mode><v_int_equiv>): ...this.
10159 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10160 Alan Hayward <alan.hayward@arm.com>
10161 David Sherwood <david.sherwood@arm.com>
10163 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10165 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10166 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10167 (UNSPEC_LD4_DREG): New unspecs.
10168 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10169 (aarch64_ld2<mode>_dreg_be): Replace with...
10170 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10172 (aarch64_ld3<mode>_dreg_le)
10173 (aarch64_ld3<mode>_dreg_be): Replace with...
10174 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10176 (aarch64_ld4<mode>_dreg_le)
10177 (aarch64_ld4<mode>_dreg_be): Replace with...
10178 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10181 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10183 PR tree-optimization/81987
10184 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10185 insert an initializer in a location not dominated by the stride
10188 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10190 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10191 on the entire header of the finally block in the fallthru case.
10193 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10195 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10197 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10199 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10200 rs6000_emit_move_from_cr and call renamed function.
10201 (rs6000_emit_prologue): Call renamed functions.
10202 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10203 movesi_from_cr, remove volatile CRs.
10205 2017-08-30 Jon Beniston <jon@beniston.com>
10206 Richard Biener <rguenther@suse.de>
10208 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10209 of VECTOR_MODE_P check.
10210 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10211 element vector types.
10213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10215 * df.h (df_read_modify_subreg_p): Remove in favor of...
10216 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10217 const_rtx instead of an rtx.
10218 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10219 * df-problems.c (df_word_lr_mark_ref): Likewise.
10220 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10221 (mark_pseudo_reg_dead): Likewise.
10222 (mark_ref_dead): Likewise.
10223 * reginfo.c (init_subregs_of_mode): Likewise.
10224 * sched-deps.c (sched_analyze_1): Likewise.
10225 * df-scan.c (df_def_record_1): Likewise.
10226 (df_uses_record): Likewise.
10227 (df_read_modify_subreg_p): Remove in favor of...
10228 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10229 const_rtx instead of an rtx.
10231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10232 Alan Hayward <alan.hayward@arm.com>
10233 David Sherwood <david.sherwood@arm.com>
10235 * rtl.h (partial_subreg_p): New function.
10236 * caller-save.c (save_call_clobbered_regs): Use it.
10237 * calls.c (expand_call): Likewise.
10238 * combine.c (combinable_i3pat): Likewise.
10239 (simplify_set): Likewise.
10240 (make_extraction): Likewise.
10241 (make_compound_operation_int): Likewise.
10242 (gen_lowpart_or_truncate): Likewise.
10243 (force_to_mode): Likewise.
10244 (make_field_assignment): Likewise.
10245 (reg_truncated_to_mode): Likewise.
10246 (record_truncated_value): Likewise.
10247 (move_deaths): Likewise.
10248 * cse.c (record_jump_cond): Likewise.
10249 (cse_insn): Likewise.
10250 * cselib.c (cselib_lookup_1): Likewise.
10251 * expmed.c (extract_bit_field_using_extv): Likewise.
10252 * function.c (assign_parm_setup_reg): Likewise.
10253 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10254 * ira-build.c (create_insn_allocnos): Likewise.
10255 * lra-coalesce.c (merge_pseudos): Likewise.
10256 * lra-constraints.c (match_reload): Likewise.
10257 (simplify_operand_subreg): Likewise.
10258 (curr_insn_transform): Likewise.
10259 * lra-lives.c (process_bb_lives): Likewise.
10260 * lra.c (new_insn_reg): Likewise.
10261 (lra_substitute_pseudo): Likewise.
10262 * regcprop.c (mode_change_ok): Likewise.
10263 (maybe_mode_change): Likewise.
10264 (copyprop_hardreg_forward_1): Likewise.
10265 * reload.c (push_reload): Likewise.
10266 (find_reloads): Likewise.
10267 (find_reloads_subreg_address): Likewise.
10268 * reload1.c (alter_reg): Likewise.
10269 (eliminate_regs_1): Likewise.
10270 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10272 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10274 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10275 back to if statements, including unpack.
10277 2017-08-30 Martin Liska <mliska@suse.cz>
10279 PR inline-asm/82001
10280 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10282 (func_checker::compare_asm_inputs_outputs): ... this function.
10283 (func_checker::compare_gimple_asm): Use the function to compare
10284 also ASM constrains.
10285 * ipa-icf-gimple.h: Rename the function.
10287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10288 Alan Hayward <alan.hayward@arm.com>
10289 David Sherwood <david.sherwood@arm.com>
10291 * coretypes.h (complex_mode): New type.
10292 * gdbhooks.py (build_pretty_printer): Handle it.
10293 * machmode.h (complex_mode): New class.
10294 (complex_mode::includes_p): New function.
10295 (is_complex_int_mode): Likewise.
10296 (is_complex_float_mode): Likewise.
10297 * genmodes.c (get_mode_class): Handle complex mode classes.
10298 * function.c (expand_function_end): Use is_complex_int_mode.
10300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10301 Alan Hayward <alan.hayward@arm.com>
10302 David Sherwood <david.sherwood@arm.com>
10304 * coretypes.h (scalar_mode_pod): New typedef.
10305 * gdbhooks.py (build_pretty_printer): Handle it.
10306 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10307 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10308 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10309 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10310 as_a <scalar_mode>.
10312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10313 Alan Hayward <alan.hayward@arm.com>
10314 David Sherwood <david.sherwood@arm.com>
10316 * machmode.h (mode_for_vector): Take a scalar_mode instead
10318 * stor-layout.c (mode_for_vector): Likewise.
10319 * explow.c (promote_mode): Use as_a <scalar_mode>.
10320 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10323 Alan Hayward <alan.hayward@arm.com>
10324 David Sherwood <david.sherwood@arm.com>
10326 * target.def (preferred_simd_mode): Take a scalar_mode
10327 instead of a machine_mode.
10328 * targhooks.h (default_preferred_simd_mode): Likewise.
10329 * targhooks.c (default_preferred_simd_mode): Likewise.
10330 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10331 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10332 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10333 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10334 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10335 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10336 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10337 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10339 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10340 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10341 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10342 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10343 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10344 * doc/tm.texi: Regenerate.
10345 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10349 Alan Hayward <alan.hayward@arm.com>
10350 David Sherwood <david.sherwood@arm.com>
10352 * target.def (scalar_mode_supported_p): Take a scalar_mode
10353 instead of a machine_mode.
10354 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10355 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10356 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10357 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10358 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10359 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10360 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10361 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10362 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10363 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10364 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10365 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10366 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10367 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10369 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10370 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10371 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10372 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10373 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10375 * doc/tm.texi: Regenerate.
10377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10378 Alan Hayward <alan.hayward@arm.com>
10379 David Sherwood <david.sherwood@arm.com>
10381 * coretypes.h (opt_scalar_mode): New typedef.
10382 * gdbhooks.py (build_pretty_printers): Handle it.
10383 * machmode.h (mode_iterator::get_2xwider): Add overload for
10385 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10387 * expr.c (convert_mode_scalar): Likewise.
10388 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10389 * optabs.c (expand_float): Likewise.
10390 (expand_fix): Likewise.
10391 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10394 Alan Hayward <alan.hayward@arm.com>
10395 David Sherwood <david.sherwood@arm.com>
10397 * optabs.c (expand_float): Explicitly check for scalars before
10398 using a branching expansion.
10399 (expand_fix): Likewise.
10401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10402 Alan Hayward <alan.hayward@arm.com>
10403 David Sherwood <david.sherwood@arm.com>
10405 * expr.c (convert_mode): Split scalar handling out into...
10406 (convert_mode_scalar): ...this new function. Treat the modes
10409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10410 Alan Hayward <alan.hayward@arm.com>
10411 David Sherwood <david.sherwood@arm.com>
10413 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10415 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10418 Alan Hayward <alan.hayward@arm.com>
10419 David Sherwood <david.sherwood@arm.com>
10421 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10422 rather than a machine_mode.
10423 (fixed_from_string): Likewise.
10424 (fixed_convert): Likewise.
10425 (fixed_convert_from_int): Likewise.
10426 (fixed_convert_from_real): Likewise.
10427 (real_convert_from_fixed): Likewise.
10428 * fixed-value.c (fixed_from_double_int): Likewise.
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 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10437 Alan Hayward <alan.hayward@arm.com>
10438 David Sherwood <david.sherwood@arm.com>
10440 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10441 of separate mode class checks. Do not allow vector modes here.
10442 (immed_wide_int_const): Use as_a <scalar_mode>.
10443 * explow.c (trunc_int_for_mode): Likewise.
10444 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10445 (wi::shwi): Likewise.
10446 (wi::min_value): Likewise.
10447 (wi::max_value): Likewise.
10448 * dwarf2out.c (loc_descriptor): Likewise.
10449 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10450 for CONST_WIDE_INT.
10452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10453 Alan Hayward <alan.hayward@arm.com>
10454 David Sherwood <david.sherwood@arm.com>
10456 * tree.h (SCALAR_TYPE_MODE): New macro.
10457 * expr.c (expand_expr_addr_expr_1): Use it.
10458 (expand_expr_real_2): Likewise.
10459 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10460 (fold_convert_const_fixed_from_int): Likewise.
10461 (fold_convert_const_fixed_from_real): Likewise.
10462 (native_encode_fixed): Likewise
10463 (native_encode_complex): Likewise
10464 (native_encode_vector): Likewise.
10465 (native_interpret_fixed): Likewise.
10466 (native_interpret_real): Likewise.
10467 (native_interpret_complex): Likewise.
10468 (native_interpret_vector): Likewise.
10469 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10470 (simd_clone_adjust_argument_types): Likewise.
10471 (simd_clone_init_simd_arrays): Likewise.
10472 (simd_clone_adjust): Likewise.
10473 * stor-layout.c (layout_type): Likewise.
10474 * tree.c (build_minus_one_cst): Likewise.
10475 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10476 * tree-inline.c (estimate_move_cost): Likewise.
10477 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10478 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10479 (vectorizable_reduction): Likewise.
10480 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10481 (vect_recog_mixed_size_cond_pattern): Likewise.
10482 (check_bool_pattern): Likewise.
10483 (adjust_bool_pattern): Likewise.
10484 (search_type_for_mask_1): Likewise.
10485 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10486 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10487 (vectorizable_load): Likewise.
10488 (vectorizable_store): Likewise.
10489 * ubsan.c (ubsan_encode_value): Likewise.
10490 * varasm.c (output_constant): Likewise.
10492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10493 Alan Hayward <alan.hayward@arm.com>
10494 David Sherwood <david.sherwood@arm.com>
10496 * coretypes.h (scalar_mode): New class.
10497 * machmode.h (scalar_mode): Likewise.
10498 (scalar_mode::includes_p): New function.
10499 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10500 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10501 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10502 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10503 * expmed.c (store_bit_field_1): Likewise.
10504 (extract_bit_field_1): Likewise.
10505 * expr.c (write_complex_part): Likewise.
10506 (read_complex_part): Likewise.
10507 (emit_move_complex_push): Likewise.
10508 (expand_expr_real_2): Likewise.
10509 * function.c (assign_parm_setup_reg): Likewise.
10510 (assign_parms_unsplit_complex): Likewise.
10511 * optabs.c (expand_binop): Likewise.
10512 * rtlanal.c (subreg_get_info): Likewise.
10513 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10514 * varasm.c (output_constant_pool_2): Likewise.
10516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10517 Alan Hayward <alan.hayward@arm.com>
10518 David Sherwood <david.sherwood@arm.com>
10520 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10522 (expmed_mult_highpart_optab): Likewise.
10523 (expmed_mult_highpart): Likewise.
10525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10526 Alan Hayward <alan.hayward@arm.com>
10527 David Sherwood <david.sherwood@arm.com>
10529 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10530 instead of a machine_mode.
10531 (builtin_memset_read_str): Likewise.
10532 * builtins.c (c_readstr): Likewise.
10533 (builtin_memcpy_read_str): Likewise.
10534 (builtin_strncpy_read_str): Likewise.
10535 (builtin_memset_read_str): Likewise.
10536 (builtin_memset_gen_str): Likewise.
10537 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10538 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10539 instead of a machine_mode.
10540 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10542 (make_extraction): Likewise.
10543 (try_widen_shift_mode): Take and return scalar_int_modes instead
10545 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10546 a scalar_int_mode instead of a machine_mode.
10547 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10548 (avr_addr_space_pointer_mode): Likewise.
10549 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10550 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10551 (msp430_unwind_word_mode): Likewise.
10552 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10553 (spu_addr_space_pointer_mode): Likewise.
10554 (spu_addr_space_address_mode): Likewise.
10555 (spu_libgcc_cmp_return_mode): Likewise.
10556 (spu_libgcc_shift_count_mode): Likewise.
10557 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10558 (rl78_addr_space_pointer_mode): Likewise.
10559 (fl78_unwind_word_mode): Likewise.
10560 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10562 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10563 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10564 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10565 (mips_valid_pointer_mode): Likewise.
10566 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10567 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10568 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10570 (ft32_addr_space_address_mode): Likewise.
10571 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10572 scalar_int_mode instead of a machine_mode.
10573 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10575 (m32c_addr_space_address_mode): Likewise.
10576 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10577 (rs6000_eh_return_filter_mode): Likewise.
10578 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10579 (rs6000_eh_return_filter_mode): Likewise.
10580 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10581 (s390_libgcc_shift_count_mode): Likewise.
10582 (s390_unwind_word_mode): Likewise.
10583 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10585 * target.def (mode_rep_extended): Likewise.
10586 (valid_pointer_mode): Likewise.
10587 (addr_space.valid_pointer_mode): Likewise.
10588 (eh_return_filter_mode): Return a scalar_int_mode rather than
10590 (libgcc_cmp_return_mode): Likewise.
10591 (libgcc_shift_count_mode): Likewise.
10592 (unwind_word_mode): Likewise.
10593 (addr_space.pointer_mode): Likewise.
10594 (addr_space.address_mode): Likewise.
10595 * doc/tm.texi: Regenerate.
10596 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10598 (do_jump): Use scalar_int_mode for local variables.
10599 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10600 rather than a machine_mode.
10601 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10602 (scompare_loc_descriptor_wide): Likewise.
10603 (scompare_loc_descriptor_narrow): Likewise.
10604 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10606 * except.c (sjlj_emit_dispatch_table): Likewise.
10607 (expand_builtin_eh_copy_values): Likewise.
10608 * explow.c (convert_memory_address_addr_space_1): Likewise.
10609 Take a scalar_int_mode rather than a machine_mode.
10610 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10611 than a machine_mode.
10612 (memory_address_addr_space): Use scalar_int_mode for local variables.
10613 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10614 rather than a machine_mode.
10615 * expmed.c (mask_rtx): Likewise.
10616 (init_expmed_one_conv): Likewise.
10617 (expand_mult_highpart_adjust): Likewise.
10618 (extract_high_half): Likewise.
10619 (expmed_mult_highpart_optab): Likewise.
10620 (expmed_mult_highpart): Likewise.
10621 (expand_smod_pow2): Likewise.
10622 (expand_sdiv_pow2): Likewise.
10623 (emit_store_flag_int): Likewise.
10624 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10626 (extract_low_bits): Likewise.
10627 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10629 * expr.c (pieces_addr::adjust): Likewise.
10630 (can_store_by_pieces): Likewise.
10631 (store_by_pieces): Likewise.
10632 (clear_by_pieces_1): Likewise.
10633 (expand_expr_addr_expr_1): Likewise.
10634 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10635 (expand_expr_real_1): Likewise.
10636 (try_casesi): Likewise.
10637 * final.c (shorten_branches): Likewise.
10638 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10639 type of "mode" to machine_mode.
10640 * internal-fn.c (expand_arith_overflow_result_store): Take a
10641 scalar_int_mode rather than a machine_mode.
10642 (expand_mul_overflow): Use scalar_int_mode for local variables.
10643 * loop-doloop.c (doloop_modify): Likewise.
10644 (doloop_optimize): Likewise.
10645 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10646 than a machine_mode.
10647 (expand_doubleword_shift_condmove): Likewise.
10648 (expand_doubleword_shift): Likewise.
10649 (expand_doubleword_clz): Likewise.
10650 (expand_doubleword_popcount): Likewise.
10651 (expand_doubleword_parity): Likewise.
10652 (expand_absneg_bit): Use scalar_int_mode for local variables.
10653 (prepare_float_lib_cmp): Likewise.
10654 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10655 rather than a machine_mode.
10656 (convert_memory_address_addr_space): Likewise.
10657 (get_mode_bounds): Likewise.
10658 (get_address_mode): Return a scalar_int_mode rather than a
10660 * rtlanal.c (get_address_mode): Likewise.
10661 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10662 than a machine_mode.
10663 * targhooks.c (default_mode_rep_extended): Likewise.
10664 (default_valid_pointer_mode): Likewise.
10665 (default_addr_space_valid_pointer_mode): Likewise.
10666 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10667 than a machine_mode.
10668 (default_libgcc_cmp_return_mode): Likewise.
10669 (default_libgcc_shift_count_mode): Likewise.
10670 (default_unwind_word_mode): Likewise.
10671 (default_addr_space_pointer_mode): Likewise.
10672 (default_addr_space_address_mode): Likewise.
10673 * targhooks.h (default_eh_return_filter_mode): Likewise.
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 (default_mode_rep_extended): Take a scalar_int_mode rather than
10681 (default_valid_pointer_mode): Likewise.
10682 (default_addr_space_valid_pointer_mode): Likewise.
10683 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10685 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10686 rather than a machine_mode.
10687 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10688 for local variables.
10689 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10690 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10691 than a machine_mode.
10693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10694 Alan Hayward <alan.hayward@arm.com>
10695 David Sherwood <david.sherwood@arm.com>
10697 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10698 the mode argument to scalar_int_mode.
10699 (do_jump_by_parts_zero_rtx): Likewise.
10700 (do_jump_by_parts_equality_rtx): Likewise.
10701 (do_jump_by_parts_greater): Take a mode argument.
10702 (do_jump_by_parts_equality): Likewise.
10703 (do_jump_1): Update calls accordingly.
10705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10706 Alan Hayward <alan.hayward@arm.com>
10707 David Sherwood <david.sherwood@arm.com>
10709 * is-a.h (safe_dyn_cast): New function.
10710 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10711 (jump_table_for_label): Likewise.
10712 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10713 instead of an rtx_insn *.
10714 (shorten_branches): Use dyn_cast instead of LABEL_P and
10715 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10716 rtx_jump_table_data::get_data_mode.
10717 (final_scan_insn): Likewise.
10719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10720 Alan Hayward <alan.hayward@arm.com>
10721 David Sherwood <david.sherwood@arm.com>
10723 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10724 trying to combine a full-register integer set with a subreg
10727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10728 Alan Hayward <alan.hayward@arm.com>
10729 David Sherwood <david.sherwood@arm.com>
10731 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10732 that is always either address_mode or pointer_mode.
10734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10735 Alan Hayward <alan.hayward@arm.com>
10736 David Sherwood <david.sherwood@arm.com>
10738 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10739 when the two are known to be equal.
10741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10742 Alan Hayward <alan.hayward@arm.com>
10743 David Sherwood <david.sherwood@arm.com>
10745 * simplify-rtx.c (simplify_const_unary_operation): Use
10746 is_a <scalar_int_mode> instead of checking for a nonzero
10747 precision. Forcibly convert op_mode to a scalar_int_mode
10748 in that case. More clearly differentiate the operand and
10749 result modes and use the former when deciding what the value
10750 of a count-bits operation should be. Use is_int_mode instead
10751 of checking for a MODE_INT. Remove redundant check for whether
10752 this mode has a zero precision.
10754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10755 Alan Hayward <alan.hayward@arm.com>
10756 David Sherwood <david.sherwood@arm.com>
10758 * optabs.c (widen_leading): Change the type of the mode argument
10759 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10760 (widen_bswap): Likewise.
10761 (expand_parity): Likewise.
10762 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10763 (expand_ffs): Likewise.
10764 (epand_unop): Check for scalar integer modes before calling the
10767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10768 Alan Hayward <alan.hayward@arm.com>
10769 David Sherwood <david.sherwood@arm.com>
10771 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10773 (expand_expr_real_1): Update call accordingly.
10775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10776 Alan Hayward <alan.hayward@arm.com>
10777 David Sherwood <david.sherwood@arm.com>
10779 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10780 value_mode arguments. Use scalar_int_mode internally.
10781 (store_bit_field_1): Rename the new integer mode from imode
10782 to op0_mode and use it instead of GET_MODE (op0). Update calls
10783 to store_split_bit_field, store_bit_field_using_insv and
10784 store_fixed_bit_field.
10785 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10786 Use scalar_int_mode internally. Use a bit count rather than a mode
10787 when calculating the largest bit size for get_best_mode.
10788 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10789 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10790 Remove assertion that OP0 has a scalar integer mode.
10791 (store_split_bit_field): Add op0_mode and value_mode arguments.
10792 Update calls to extract_fixed_bit_field.
10793 (extract_bit_field_using_extv): Add an op0_mode argument.
10794 Use scalar_int_mode internally.
10795 (extract_bit_field_1): Rename the new integer mode from imode to
10796 op0_mode and use it instead of GET_MODE (op0). Update calls to
10797 extract_split_bit_field, extract_bit_field_using_extv and
10798 extract_fixed_bit_field.
10799 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10800 to extract_split_bit_field and extract_fixed_bit_field_1.
10801 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10802 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10803 on the target mode.
10804 (extract_split_bit_field): Add an op0_mode argument. Update call
10805 to extract_fixed_bit_field.
10807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10808 Alan Hayward <alan.hayward@arm.com>
10809 David Sherwood <david.sherwood@arm.com>
10811 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10812 * explow.c (hard_function_value): Likewise.
10813 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10814 convert_to_mode call outside the loop.
10815 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10816 for the mode iterator. Require the mode specified by max_pieces
10818 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10820 (copy_blkmode_to_reg): Likewise.
10821 (set_storage_via_setmem): Likewise.
10822 * optabs.c (prepare_cmp_insn): Likewise.
10823 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10824 * stor-layout.c (finish_bitfield_representative): Likewise.
10826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10827 Alan Hayward <alan.hayward@arm.com>
10828 David Sherwood <david.sherwood@arm.com>
10830 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10831 * expr.c (convert_move): Use them.
10832 (convert_modes): Likewise.
10833 (store_expr_with_bounds): Likewise.
10835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10836 Alan Hayward <alan.hayward@arm.com>
10837 David Sherwood <david.sherwood@arm.com>
10839 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10840 parameter for the mode of "x". Remove the "known_x", "known_mode"
10841 and "known_ret" arguments. Change the type of the mode argument
10842 to scalar_int_mode.
10843 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10844 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10845 (reg_num_sign_bit_copies_for_combine): Likewise.
10846 * rtlanal.c (nonzero_bits1): Likewise.
10847 (num_sign_bit_copies1): Likewise.
10848 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10849 (reg_num_sign_bit_copies_general): Likewise.
10850 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10851 (reg_nonzero_bits_general): Likewise.
10853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10854 Alan Hayward <alan.hayward@arm.com>
10855 David Sherwood <david.sherwood@arm.com>
10857 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10858 than in subroutines. Return 1 for non-integer modes.
10859 (cached_num_sign_bit_copies): Change the type of the mode parameter
10860 to scalar_int_mode.
10861 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10862 classes. Handle CONST_INT_P first and then check whether X also
10863 has a scalar integer mode. Check the same thing for inner registers
10864 of a SUBREG and for values that are being extended or truncated.
10866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10867 Alan Hayward <alan.hayward@arm.com>
10868 David Sherwood <david.sherwood@arm.com>
10870 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10871 in subroutines. Return the mode mask for non-integer modes.
10872 (cached_nonzero_bits): Change the type of the mode parameter
10873 to scalar_int_mode.
10874 (nonzero_bits1): Likewise. Remove early exit for other mode
10875 classes. Handle CONST_INT_P first and then check whether X
10876 also has a scalar integer mode.
10878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10879 Alan Hayward <alan.hayward@arm.com>
10880 David Sherwood <david.sherwood@arm.com>
10882 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10883 Return a scalar_int_mode and assert that the size is greater than
10885 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10886 (op_by_pieces_d::op_by_pieces_d): Likewise.
10887 (op_by_pieces_d::run): Likewise.
10888 (can_store_by_pieces): Likewise.
10890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10891 Alan Hayward <alan.hayward@arm.com>
10892 David Sherwood <david.sherwood@arm.com>
10894 * combine.c (extract_left_shift): Add a mode argument and update
10896 (make_compound_operation_int): Change the type of the mode parameter
10897 to scalar_int_mode and update the call to extract_left_shift.
10899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10900 Alan Hayward <alan.hayward@arm.com>
10901 David Sherwood <david.sherwood@arm.com>
10903 * combine.c (simplify_and_const_int): Change the type of the mode
10904 parameter to scalar_int_mode.
10905 (simplify_and_const_int_1): Likewise. Update recursive call.
10907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10908 Alan Hayward <alan.hayward@arm.com>
10909 David Sherwood <david.sherwood@arm.com>
10911 * combine.c (simplify_compare_const): Check that the mode is a
10912 scalar_int_mode (rather than VOIDmode) before testing its
10914 (simplify_comparison): Move COMPARISON_P handling out of the
10915 loop and restrict the latter part of the loop to scalar_int_modes.
10916 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10917 and when considering SUBREG_REGs. Use is_int_mode instead of
10918 checking GET_MODE_CLASS against MODE_INT.
10920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10921 Alan Hayward <alan.hayward@arm.com>
10922 David Sherwood <david.sherwood@arm.com>
10924 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10925 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10926 shift_unit_mode and for modes involved in scalar shifts.
10928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10929 Alan Hayward <alan.hayward@arm.com>
10930 David Sherwood <david.sherwood@arm.com>
10932 * combine.c (force_int_to_mode): New function, split out from...
10933 (force_to_mode): ...here. Keep xmode up-to-date and use it
10934 instead of GET_MODE (x).
10936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10937 Alan Hayward <alan.hayward@arm.com>
10938 David Sherwood <david.sherwood@arm.com>
10940 * optabs-query.h (extraction_insn::struct_mode): Change type to
10941 opt_scalar_int_mode and update comment.
10942 (extraction_insn::field_mode): Change type to scalar_int_mode.
10943 (extraction_insn::pos_mode): Likewise.
10944 * combine.c (make_extraction): Update accordingly.
10945 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10946 (get_optab_extraction_insn): Likewise.
10947 * recog.c (simplify_while_replacing): Likewise.
10948 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10949 parameter to opt_scalar_int_mode.
10951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10952 Alan Hayward <alan.hayward@arm.com>
10953 David Sherwood <david.sherwood@arm.com>
10955 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10956 to a scalar_int_mode instead of a machine_mode.
10957 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10958 (get_best_mode): Return a boolean and use a pointer argument to store
10959 the selected mode. Replace the limit mode parameter with a bit limit.
10960 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10961 for the values returned by bit_field_mode_iterator::next_mode.
10962 (store_bit_field): Update call to get_best_mode.
10963 (store_fixed_bit_field): Likewise.
10964 (extract_fixed_bit_field): Likewise.
10965 * expr.c (optimize_bitfield_assignment_op): Likewise.
10966 * fold-const.c (optimize_bit_field_compare): Likewise.
10967 (fold_truth_andor_1): Likewise.
10968 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10969 Update for new type of m_mode.
10970 (get_best_mode): As above.
10972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10973 Alan Hayward <alan.hayward@arm.com>
10974 David Sherwood <david.sherwood@arm.com>
10976 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10977 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10978 (store_bit_field): Check is_a <scalar_int_mode> before calling
10979 strict_volatile_bitfield_p.
10980 (extract_bit_field): Likewise.
10982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10983 Alan Hayward <alan.hayward@arm.com>
10984 David Sherwood <david.sherwood@arm.com>
10986 * target.def (cstore_mode): Return a scalar_int_mode.
10987 * doc/tm.texi: Regenerate.
10988 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10989 * targhooks.h (default_cstore_mode): Likewise.
10990 * targhooks.c (default_cstore_mode): Likewise, using a forced
10992 * expmed.c (emit_cstore): Expect the target of the cstore to be
10995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10996 Alan Hayward <alan.hayward@arm.com>
10997 David Sherwood <david.sherwood@arm.com>
10999 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11001 (niter_desc): Likewise mode.
11002 (iv_analyze): Add a mode parameter.
11004 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11005 and change its type to scalar_int_mode.
11006 * loop-iv.c: Update commentary at head of file.
11007 (iv_constant): Pass the mode paraeter before the rtx it describes
11008 and change its type to scalar_int_mode. Remove VOIDmode handling.
11009 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11010 (iv_extend): Likewise.
11011 (shorten_into_mode): Likewise.
11012 (iv_add): Use scalar_int_mode.
11013 (iv_mult): Likewise.
11014 (iv_shift): Likewise.
11015 (canonicalize_iv_subregs): Likewise.
11016 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11017 it describes and change its mode to scalar_int_mode. Also change
11018 the type of the returned inner_mode to scalar_int_mode.
11019 (get_biv_step): Likewise, turning outer_mode from a pointer
11020 into a direct parameter. Update call to get_biv_step_1.
11021 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11022 iv_constant and get_biv_step.
11023 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11024 and change its type to scalar_int_mode. Don't initialise iv->mode
11025 to VOIDmode and remove later checks for its still being VOIDmode.
11026 Update calls to iv_analyze_op and iv_analyze_expr. Check
11027 is_a <scalar_int_mode> when changing the mode under consideration.
11028 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11029 Update call to iv_analyze_expr.
11030 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11031 inner register is not also a scalar_int_mode. Update call to
11033 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11034 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11035 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11036 separate mode class checks. Update calls to iv_analyze. Remove
11037 fix-up of VOIDmodes after iv_analyze_biv.
11038 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11039 don't have a scalar_int_mode. Update call to biv_p.
11041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11042 Alan Hayward <alan.hayward@arm.com>
11043 David Sherwood <david.sherwood@arm.com>
11045 * cfgexpand.c (convert_debug_memory_address): Use
11046 as_a <scalar_int_mode>.
11047 * combine.c (expand_compound_operation): Likewise.
11048 (make_extraction): Likewise.
11049 (change_zero_ext): Likewise.
11050 (simplify_comparison): Likewise.
11051 * cse.c (cse_insn): Likewise.
11052 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11053 (mem_loc_descriptor): Likewise.
11054 (loc_descriptor): Likewise.
11055 * expmed.c (init_expmed_one_mode): Likewise.
11056 (synth_mult): Likewise.
11057 (emit_store_flag_1): Likewise.
11058 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11059 of a comparison with size.
11060 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11061 (reduce_to_bit_field_precision): Likewise.
11062 * function.c (expand_function_end): Likewise.
11063 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11064 * loop-doloop.c (doloop_modify): Likewise.
11065 * optabs.c (expand_binop): Likewise.
11066 (expand_unop): Likewise.
11067 (expand_copysign_absneg): Likewise.
11068 (prepare_cmp_insn): Likewise.
11069 (maybe_legitimize_operand): Likewise.
11070 * recog.c (const_scalar_int_operand): Likewise.
11071 * rtlanal.c (get_address_mode): Likewise.
11072 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11073 (simplify_cond_clz_ctz): Likewise.
11074 * tree-nested.c (get_nl_goto_field): Likewise.
11075 * tree.c (build_vector_type_for_mode): Likewise.
11076 * var-tracking.c (use_narrower_mode): Likewise.
11078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11079 Alan Hayward <alan.hayward@arm.com>
11080 David Sherwood <david.sherwood@arm.com>
11082 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11083 * builtins.c (expand_builtin_signbit): Use it.
11084 * cfgexpand.c (expand_debug_expr): Likewise.
11085 * dojump.c (do_jump): Likewise.
11086 (do_compare_and_jump): Likewise.
11087 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11088 * expmed.c (make_tree): Likewise.
11089 * expr.c (expand_expr_real_2): Likewise.
11090 (expand_expr_real_1): Likewise.
11091 (try_casesi): Likewise.
11092 * fold-const-call.c (fold_const_call_ss): Likewise.
11093 * fold-const.c (unextend): Likewise.
11094 (extract_muldiv_1): Likewise.
11095 (fold_single_bit_test): Likewise.
11096 (native_encode_int): Likewise.
11097 (native_encode_string): Likewise.
11098 (native_interpret_int): Likewise.
11099 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11100 * internal-fn.c (expand_addsub_overflow): Likewise.
11101 (expand_neg_overflow): Likewise.
11102 (expand_mul_overflow): Likewise.
11103 (expand_arith_overflow): Likewise.
11104 * match.pd: Likewise.
11105 * stor-layout.c (layout_type): Likewise.
11106 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11107 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11108 * tree-ssanames.c (get_range_info): Likewise.
11109 * tree-switch-conversion.c (array_value_type) Likewise.
11110 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11111 (vect_recog_divmod_pattern): Likewise.
11112 (vect_recog_mixed_size_cond_pattern): Likewise.
11113 * tree-vrp.c (extract_range_basic): Likewise.
11114 (simplify_float_conversion_using_ranges): Likewise.
11115 * tree.c (int_fits_type_p): Likewise.
11116 * ubsan.c (instrument_bool_enum_load): Likewise.
11117 * varasm.c (mergeable_string_section): Likewise.
11118 (narrowing_initializer_constant_valid_p): Likewise.
11119 (output_constant): Likewise.
11121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11122 Alan Hayward <alan.hayward@arm.com>
11123 David Sherwood <david.sherwood@arm.com>
11125 * machmode.h (NARROWEST_INT_MODE): New macro.
11126 * expr.c (alignment_for_piecewise_move): Use it instead of
11127 GET_CLASS_NARROWEST_MODE (MODE_INT).
11128 (push_block): Likewise.
11129 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11131 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11134 Alan Hayward <alan.hayward@arm.com>
11135 David Sherwood <david.sherwood@arm.com>
11137 * postreload.c (move2add_valid_value_p): Change the type of the
11138 mode parameter to scalar_int_mode.
11139 (move2add_use_add2_insn): Add a mode parameter and use it instead
11141 (move2add_use_add3_insn): Likewise.
11142 (reload_cse_move2add): Update accordingly.
11144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11145 Alan Hayward <alan.hayward@arm.com>
11146 David Sherwood <david.sherwood@arm.com>
11148 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11150 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11151 * optabs.c (expand_unop): Likewise.
11153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11154 Alan Hayward <alan.hayward@arm.com>
11155 David Sherwood <david.sherwood@arm.com>
11157 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11158 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11159 (popcount_loc_descriptor): Likewise.
11160 (bswap_loc_descriptor): Likewise.
11161 (rotate_loc_descriptor): Likewise.
11162 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11163 calling the functions above.
11165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11166 Alan Hayward <alan.hayward@arm.com>
11167 David Sherwood <david.sherwood@arm.com>
11169 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11171 (try_combine): Likewise.
11172 (simplify_if_then_else): Likewise.
11173 * cse.c (cse_insn): Likewise.
11174 * dwarf2out.c (mem_loc_descriptor): Likewise.
11175 * emit-rtl.c (gen_lowpart_common): Likewise.
11176 * simplify-rtx.c (simplify_truncation): Likewise.
11177 (simplify_binary_operation_1): Likewise.
11178 (simplify_const_relational_operation): Likewise.
11179 (simplify_ternary_operation): Likewise.
11180 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11183 Alan Hayward <alan.hayward@arm.com>
11184 David Sherwood <david.sherwood@arm.com>
11186 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11187 * reload.c (push_reload): Likewise.
11188 (find_reloads): Likewise.
11190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11191 Alan Hayward <alan.hayward@arm.com>
11192 David Sherwood <david.sherwood@arm.com>
11194 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11195 (make_compound_operation_int): Likewise.
11196 (change_zero_ext): Likewise.
11197 * expr.c (convert_move): Likewise.
11198 (convert_modes): Likewise.
11199 * fwprop.c (forward_propagate_subreg): Likewise.
11200 * loop-iv.c (get_biv_step_1): Likewise.
11201 * optabs.c (widen_operand): Likewise.
11202 * postreload.c (move2add_valid_value_p): Likewise.
11203 * recog.c (simplify_while_replacing): Likewise.
11204 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11205 (simplify_binary_operation_1): Likewise. Remove redundant
11206 mode equality check.
11208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11209 Alan Hayward <alan.hayward@arm.com>
11210 David Sherwood <david.sherwood@arm.com>
11212 * combine.c (combine_simplify_rtx): Add checks for
11213 is_a <scalar_int_mode>.
11214 (simplify_if_then_else): Likewise.
11215 (make_field_assignment): Likewise.
11216 (simplify_comparison): Likewise.
11217 * ifcvt.c (noce_try_bitop): Likewise.
11218 * loop-invariant.c (canonicalize_address_mult): Likewise.
11219 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11222 Alan Hayward <alan.hayward@arm.com>
11223 David Sherwood <david.sherwood@arm.com>
11225 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11226 is_a <scalar_int_mode> instead of != BLKmode.
11228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11229 Alan Hayward <alan.hayward@arm.com>
11230 David Sherwood <david.sherwood@arm.com>
11232 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11233 instead of != VOIDmode.
11234 * combine.c (if_then_else_cond): Likewise.
11235 (change_zero_ext): Likewise.
11236 * dwarf2out.c (mem_loc_descriptor): Likewise.
11237 (loc_descriptor): Likewise.
11238 * rtlanal.c (canonicalize_condition): Likewise.
11239 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11242 Alan Hayward <alan.hayward@arm.com>
11243 David Sherwood <david.sherwood@arm.com>
11245 * simplify-rtx.c (simplify_binary_operation_1): Use
11246 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11249 Alan Hayward <alan.hayward@arm.com>
11250 David Sherwood <david.sherwood@arm.com>
11252 * wide-int.h (int_traits<unsigned char>) New class.
11253 (int_traits<unsigned short>) Likewise.
11254 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11255 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11257 * combine.c (set_nonzero_bits_and_sign_copies): Use
11258 is_a <scalar_int_mode>.
11259 (find_split_point): Likewise.
11260 (combine_simplify_rtx): Likewise.
11261 (simplify_logical): Likewise.
11262 (expand_compound_operation): Likewise.
11263 (expand_field_assignment): Likewise.
11264 (make_compound_operation): Likewise.
11265 (extended_count): Likewise.
11266 (change_zero_ext): Likewise.
11267 (simplify_comparison): Likewise.
11268 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11269 (ucompare_loc_descriptor): Likewise.
11270 (minmax_loc_descriptor): Likewise.
11271 (mem_loc_descriptor): Likewise.
11272 (loc_descriptor): Likewise.
11273 * expmed.c (init_expmed_one_mode): Likewise.
11274 * lra-constraints.c (lra_constraint_offset): Likewise.
11275 * optabs.c (prepare_libcall_arg): Likewise.
11276 * postreload.c (move2add_note_store): Likewise.
11277 * reload.c (operands_match_p): Likewise.
11278 * rtl.h (load_extend_op): Likewise.
11279 * rtlhooks.c (gen_lowpart_general): Likewise.
11280 * simplify-rtx.c (simplify_truncation): Likewise.
11281 (simplify_unary_operation_1): Likewise.
11282 (simplify_binary_operation_1): Likewise.
11283 (simplify_const_binary_operation): Likewise.
11284 (simplify_const_relational_operation): Likewise.
11285 (simplify_subreg): Likewise.
11286 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11287 * var-tracking.c (adjust_mems): Likewise.
11288 (prepare_call_arguments): Likewise.
11290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11291 Alan Hayward <alan.hayward@arm.com>
11292 David Sherwood <david.sherwood@arm.com>
11294 * machmode.h (is_int_mode): New fuction.
11295 * combine.c (find_split_point): Use it.
11296 (combine_simplify_rtx): Likewise.
11297 (simplify_if_then_else): Likewise.
11298 (simplify_set): Likewise.
11299 (simplify_shift_const_1): Likewise.
11300 (simplify_comparison): Likewise.
11301 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11302 * cse.c (notreg_cost): Likewise.
11303 (cse_insn): Likewise.
11304 * cselib.c (cselib_lookup_1): Likewise.
11305 * dojump.c (do_jump_1): Likewise.
11306 (do_compare_rtx_and_jump): Likewise.
11307 * dse.c (get_call_args): Likewise.
11308 * dwarf2out.c (rtl_for_decl_init): Likewise.
11309 (native_encode_initializer): Likewise.
11310 * expmed.c (emit_store_flag_1): Likewise.
11311 (emit_store_flag): Likewise.
11312 * expr.c (convert_modes): Likewise.
11313 (store_field): Likewise.
11314 (expand_expr_real_1): Likewise.
11315 * fold-const.c (fold_read_from_constant_string): Likewise.
11316 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11317 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11318 * optabs.c (expand_binop): Likewise.
11319 (expand_unop): Likewise.
11320 (expand_abs_nojump): Likewise.
11321 (expand_one_cmpl_abs_nojump): Likewise.
11322 * simplify-rtx.c (mode_signbit_p): Likewise.
11323 (val_signbit_p): Likewise.
11324 (val_signbit_known_set_p): Likewise.
11325 (val_signbit_known_clear_p): Likewise.
11326 (simplify_relational_operation_1): Likewise.
11327 * tree.c (vector_type_mode): Likewise.
11329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11330 Alan Hayward <alan.hayward@arm.com>
11331 David Sherwood <david.sherwood@arm.com>
11333 * machmode.h (smallest_mode_for_size): Fix formatting.
11334 (smallest_int_mode_for_size): New function.
11335 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11336 instead of smallest_mode_for_size.
11337 * combine.c (make_extraction): Likewise.
11338 * config/arc/arc.c (arc_expand_movmem): Likewise.
11339 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11340 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11341 * config/s390/s390.c (s390_expand_insv): Likewise.
11342 * config/sparc/sparc.c (assign_int_registers): Likewise.
11343 * config/spu/spu.c (spu_function_value): Likewise.
11344 (spu_function_arg): Likewise.
11345 * coverage.c (get_gcov_type): Likewise.
11346 (get_gcov_unsigned_t): Likewise.
11347 * dse.c (find_shift_sequence): Likewise.
11348 * expmed.c (store_bit_field_1): Likewise.
11349 * expr.c (convert_move): Likewise.
11350 (store_field): Likewise.
11351 * internal-fn.c (expand_arith_overflow): Likewise.
11352 * optabs-query.c (get_best_extraction_insn): Likewise.
11353 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11354 * stor-layout.c (layout_type): Likewise.
11355 (initialize_sizetypes): Likewise.
11356 * targhooks.c (default_get_mask_mode): Likewise.
11357 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11360 Alan Hayward <alan.hayward@arm.com>
11361 David Sherwood <david.sherwood@arm.com>
11363 * machmode.h (opt_mode::else_blk): New function.
11364 (int_mode_for_mode): Declare.
11365 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11366 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11368 * cfgexpand.c (expand_debug_expr): Likewise.
11369 * combine.c (gen_lowpart_or_truncate): Likewise.
11370 (gen_lowpart_for_combine): Likewise.
11371 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11372 * config/avr/avr.c (avr_to_int_mode): Likewise.
11373 (avr_out_plus_1): Likewise.
11374 (avr_out_plus): Likewise.
11375 (avr_out_round): Likewise.
11376 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11377 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11378 (s390_expand_vcond): Likewise.
11379 * config/spu/spu.c (spu_split_immediate): Likewise.
11380 (spu_expand_mov): Likewise.
11381 * dse.c (get_stored_val): Likewise.
11382 * expmed.c (store_bit_field_1): Likewise.
11383 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11385 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11386 (extract_low_bits): Likewise.
11387 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11388 handling rather than repeating the check.
11389 (emit_group_store): Likewise.
11390 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11391 * optabs.c (expand_absneg_bit): Likewise.
11392 (expand_copysign_absneg): Likewise.
11393 (expand_copysign_bit): Likewise.
11394 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11395 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11396 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11397 * var-tracking.c (prepare_call_arguments): Likewise.
11398 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11399 int_mode_for_mode instead of mode_for_size.
11400 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11403 Alan Hayward <alan.hayward@arm.com>
11404 David Sherwood <david.sherwood@arm.com>
11406 * machmode.h (int_mode_for_size): New function.
11407 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11408 instead of mode_for_size.
11409 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11411 * combine.c (expand_field_assignment): Use int_mode_for_size
11412 instead of mode_for_size.
11413 (make_extraction): Likewise.
11414 (simplify_shift_const_1): Likewise.
11415 (simplify_comparison): Likewise.
11416 * dojump.c (do_jump): Likewise.
11417 * dwarf2out.c (mem_loc_descriptor): Likewise.
11418 * emit-rtl.c (init_derived_machine_modes): Likewise.
11419 * expmed.c (flip_storage_order): Likewise.
11420 (convert_extracted_bit_field): Likewise.
11421 * expr.c (copy_blkmode_from_reg): Likewise.
11422 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11423 * internal-fn.c (expand_mul_overflow): Likewise.
11424 * lower-subreg.c (simple_move): Likewise.
11425 * optabs-libfuncs.c (init_optabs): Likewise.
11426 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11427 * tree.c (vector_type_mode): Likewise.
11428 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11429 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11430 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11431 * tree-vect-stmts.c (vectorizable_load): Likewise.
11432 (vectorizable_store): Likewise.
11434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11435 Alan Hayward <alan.hayward@arm.com>
11436 David Sherwood <david.sherwood@arm.com>
11438 * coretypes.h (pod_mode): New type.
11439 (scalar_int_mode_pod): New typedef.
11440 * machmode.h (pod_mode): New class.
11441 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11442 * genmodes.c (emit_mode_int_n): Update accordingly.
11443 * lower-subreg.h (target_lower_subreg): Change type to
11444 scalar_int_mode_pod.
11445 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11446 scalar_int_mode_pod.
11448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11449 Alan Hayward <alan.hayward@arm.com>
11450 David Sherwood <david.sherwood@arm.com>
11452 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11453 machine_mode to scalar_int_mode.
11454 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11455 (rs6000_option_override_internal): Remove cast to int.
11456 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11457 machine_mode to scalar_int_mode.
11458 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11459 (rs6000_option_override_internal): Remove cast to int.
11460 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11461 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11463 * config/s390/s390.c (s390_expand_builtin): Likewise.
11464 * coretypes.h (scalar_int_mode): New type.
11465 (opt_scalar_int_mode): New typedef.
11466 * machmode.h (scalar_int_mode): New class.
11467 (scalar_int_mode::includes_p): New function.
11468 (byte_mode): Change type to scalar_int_mode.
11469 (word_mode): Likewise.
11470 (ptr_mode): Likewise.
11471 * emit-rtl.c (byte_mode): Likewise.
11472 (word_mode): Likewise.
11473 (ptr_mode): Likewise.
11474 (init_derived_machine_modes): Update accordingly.
11475 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11476 and MODE_PARTIAL_INT.
11477 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11478 opt_scalar_int_mode.
11480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11481 Alan Hayward <alan.hayward@arm.com>
11482 David Sherwood <david.sherwood@arm.com>
11484 * target.def (libgcc_floating_mode_supported_p): Take a
11486 * doc/tm.texi: Regenerate.
11487 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11489 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11490 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11494 Alan Hayward <alan.hayward@arm.com>
11495 David Sherwood <david.sherwood@arm.com>
11497 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11498 * doc/tm.texi: Regenerate.
11499 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11500 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11501 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11502 * targhooks.h (default_floatn_mode): Likewise.
11503 * targhooks.c (default_floatn_mode): Likewise.
11504 * tree.c (build_common_tree_nodes): Update accordingly.
11506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11507 Alan Hayward <alan.hayward@arm.com>
11508 David Sherwood <david.sherwood@arm.com>
11510 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11511 (mode_iterator::iterate_p): Likewise.
11512 (mode_iterator::get_wider): Likewise.
11513 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11516 Alan Hayward <alan.hayward@arm.com>
11517 David Sherwood <david.sherwood@arm.com>
11519 * coretypes.h (opt_scalar_float_mode): New typedef.
11520 * machmode.h (float_mode_for_size): New function.
11521 * emit-rtl.c (double_mode): Delete.
11522 (init_emit_once): Use float_mode_for_size.
11523 * stor-layout.c (layout_type): Likewise.
11524 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11527 Alan Hayward <alan.hayward@arm.com>
11528 David Sherwood <david.sherwood@arm.com>
11530 * output.h (assemble_real): Take a scalar_float_mode.
11531 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11532 * config/arm/arm.md (consttable_4): Likewise.
11533 (consttable_8): Likewise.
11534 (consttable_16): Likewise.
11535 * config/mips/mips.md (consttable_float): Likewise.
11536 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11537 * varasm.c (assemble_real): Take a scalar_float_mode.
11538 (output_constant_pool_2): Update accordingly.
11539 (output_constant): Likewise.
11541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11542 Alan Hayward <alan.hayward@arm.com>
11543 David Sherwood <david.sherwood@arm.com>
11545 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11546 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11547 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11548 (native_encode_real): Likewise.
11549 (native_interpret_real): Likewise.
11550 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11551 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11554 Alan Hayward <alan.hayward@arm.com>
11555 David Sherwood <david.sherwood@arm.com>
11557 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11558 <scalar_float_mode>. Simplify.
11559 (gen_extend_conv_libfunc): Likewise.
11561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11562 Alan Hayward <alan.hayward@arm.com>
11563 David Sherwood <david.sherwood@arm.com>
11565 * coretypes.h (scalar_float_mode): New type.
11566 * machmode.h (mode_traits::from_int): Use machine_mode if
11567 USE_ENUM_MODES is defined.
11568 (is_a): New function.
11570 (dyn_cast): Likewise.
11571 (scalar_float_mode): New class.
11572 (scalar_float_mode::includes_p): New function.
11573 (is_float_mode): Likewise.
11574 * gdbhooks.py (MachineModePrinter): New class.
11575 (build_pretty_printer): Use it for scalar_float_mode.
11576 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11577 (format_helper::format_helper): Turn into a template.
11578 * genmodes.c (get_mode_class): New function.
11579 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11580 or machine_mode if none.
11581 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11582 as_a <scalar_float_mode>.
11583 * dwarf2out.c (mem_loc_descriptor): Likewise.
11584 (insert_float): Likewise.
11585 (add_const_value_attribute): Likewise.
11586 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11587 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11588 (expand_unop): Update accordingly.
11589 (expand_abs_nojump): Likewise.
11590 (expand_copysign_absneg): Take a scalar_float_mode.
11591 (expand_copysign_bit): Likewise.
11592 (expand_copysign): Update accordingly.
11594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11595 Alan Hayward <alan.hayward@arm.com>
11596 David Sherwood <david.sherwood@arm.com>
11598 * coretypes.h (opt_mode): New class.
11599 * machmode.h (opt_mode): Likewise.
11600 (opt_mode::else_void): New function.
11601 (opt_mode::require): Likewise.
11602 (opt_mode::exists): Likewise.
11603 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11604 (GET_MODE_2XWIDER_MODE): Likewise.
11605 (mode_iterator::get_wider): Update accordingly.
11606 (mode_iterator::get_2xwider): Likewise.
11607 (mode_iterator::get_known_wider): Likewise, turning into a template.
11608 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11609 forcing a wider mode to exist.
11610 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11611 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11612 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11613 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11614 * lower-subreg.c (init_lower_subreg): Likewise.
11615 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11616 on the final iteration.
11617 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11618 a wider mode exists before asking for a move pattern.
11619 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11620 forcing a wider mode to exist.
11621 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11622 returning false if no such mode exists.
11623 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11624 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11625 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11626 Avoid checking for a MODE_INT if we already know the mode is not a
11628 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11629 forcing a wider mode to exist.
11630 (expmed_mult_highpart_optab): Likewise.
11631 (expmed_mult_highpart): Likewise.
11632 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11634 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11635 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11636 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11637 * internal-fn.c (expand_mul_overflow): Update use of
11638 GET_MODE_2XWIDER_MODE.
11639 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11640 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11641 GET_MODE_WIDER_MODE.
11642 (convert_plusminus_to_widen): Likewise.
11643 * tree-switch-conversion.c (array_value_type): Likewise.
11644 * var-tracking.c (emit_note_insn_var_location): Likewise.
11645 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11646 Return false inside rather than outside the loop if no wider mode
11648 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11649 and GET_MODE_2XWIDER_MODE
11650 (can_compare_p): Use else_void.
11651 * gdbhooks.py (OptMachineModePrinter): New class.
11652 (build_pretty_printer): Use it for opt_mode.
11654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11655 Alan Hayward <alan.hayward@arm.com>
11656 David Sherwood <david.sherwood@arm.com>
11658 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11659 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11662 Alan Hayward <alan.hayward@arm.com>
11663 David Sherwood <david.sherwood@arm.com>
11665 * machmode.h (mode_traits): New structure.
11666 (get_narrowest_mode): New function.
11667 (mode_iterator::start): Likewise.
11668 (mode_iterator::iterate_p): Likewise.
11669 (mode_iterator::get_wider): Likewise.
11670 (mode_iterator::get_known_wider): Likewise.
11671 (mode_iterator::get_2xwider): Likewise.
11672 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11673 (FOR_EACH_MODE): Likewise.
11674 (FOR_EACH_MODE_FROM): Likewise.
11675 (FOR_EACH_MODE_UNTIL): Likewise.
11676 (FOR_EACH_WIDER_MODE): Likewise.
11677 (FOR_EACH_2XWIDER_MODE): Likewise.
11678 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11679 * combine.c (simplify_comparison): Likewise
11680 * config/i386/i386.c (type_natural_mode): Likewise.
11681 * cse.c (cse_insn): Likewise.
11682 * dse.c (find_shift_sequence): Likewise.
11683 * emit-rtl.c (init_derived_machine_modes): Likewise.
11684 (init_emit_once): Likewise.
11685 * explow.c (hard_function_value): Likewise.
11686 * expmed.c (extract_fixed_bit_field_1): Likewise.
11687 (extract_bit_field_1): Likewise.
11688 (expand_divmod): Likewise.
11689 (emit_store_flag_1): Likewise.
11690 * expr.c (init_expr_target): Likewise.
11691 (convert_move): Likewise.
11692 (alignment_for_piecewise_move): Likewise.
11693 (widest_int_mode_for_size): Likewise.
11694 (emit_block_move_via_movmem): Likewise.
11695 (copy_blkmode_to_reg): Likewise.
11696 (set_storage_via_setmem): Likewise.
11697 (compress_float_constant): Likewise.
11698 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11699 * optabs-query.c (get_best_extraction_insn): Likewise.
11700 * optabs.c (expand_binop): Likewise.
11701 (expand_twoval_unop): Likewise.
11702 (expand_twoval_binop): Likewise.
11703 (widen_leading): Likewise.
11704 (widen_bswap): Likewise.
11705 (expand_parity): Likewise.
11706 (expand_unop): Likewise.
11707 (prepare_cmp_insn): Likewise.
11708 (prepare_float_lib_cmp): Likewise.
11709 (expand_float): Likewise.
11710 (expand_fix): Likewise.
11711 (expand_sfix_optab): Likewise.
11712 * postreload.c (move2add_use_add2_insn): Likewise.
11713 * reg-stack.c (reg_to_stack): Likewise.
11714 * reginfo.c (choose_hard_reg_mode): Likewise.
11715 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11716 * stor-layout.c (mode_for_size): Likewise.
11717 (smallest_mode_for_size): Likewise.
11718 (mode_for_vector): Likewise.
11719 (finish_bitfield_representative): Likewise.
11720 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11721 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11722 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11723 * var-tracking.c (prepare_call_arguments): Likewise.
11725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11726 Alan Hayward <alan.hayward@arm.com>
11727 David Sherwood <david.sherwood@arm.com>
11729 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11730 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11731 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11732 * machmode.h (mode_size): Move earlier in file.
11733 (mode_precision): Likewise.
11734 (mode_inner): Likewise.
11735 (mode_nunits): Likewise.
11736 (mode_unit_size): Likewise.
11737 (unit_unit_precision): Likewise.
11738 (mode_wider): Likewise.
11739 (mode_2xwider): Likewise.
11740 (machine_mode): New class.
11741 (mode_to_bytes): New function.
11742 (mode_to_bits): Likewise.
11743 (mode_to_precision): Likewise.
11744 (mode_to_inner): Likewise.
11745 (mode_to_unit_size): Likewise.
11746 (mode_to_unit_precision): Likewise.
11747 (mode_to_nunits): Likewise.
11748 (GET_MODE_SIZE): Use mode_to_bytes.
11749 (GET_MODE_BITSIZE): Use mode_to_bits.
11750 (GET_MODE_PRECISION): Use mode_to_precision.
11751 (GET_MODE_INNER): Use mode_to_inner.
11752 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11753 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11754 (GET_MODE_NUNITS): Use mode_to_nunits.
11755 * system.h (ALWAYS_INLINE): New macro.
11756 * config/powerpcspe/powerpcspe-c.c
11757 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11758 int for arg1_mode and arg2_mode.
11760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11761 Alan Hayward <alan.hayward@arm.com>
11762 David Sherwood <david.sherwood@arm.com>
11764 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11765 Prefix mode names with E_ in case statements.
11766 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11767 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11768 (aarch64_split_simd_move): Likewise.
11769 (aarch64_gen_storewb_pair): Likewise.
11770 (aarch64_gen_loadwb_pair): Likewise.
11771 (aarch64_gen_store_pair): Likewise.
11772 (aarch64_gen_load_pair): Likewise.
11773 (aarch64_get_condition_code_1): Likewise.
11774 (aarch64_constant_pool_reload_icode): Likewise.
11775 (get_rsqrte_type): Likewise.
11776 (get_rsqrts_type): Likewise.
11777 (get_recpe_type): Likewise.
11778 (get_recps_type): Likewise.
11779 (aarch64_gimplify_va_arg_expr): Likewise.
11780 (aarch64_simd_container_mode): Likewise.
11781 (aarch64_emit_load_exclusive): Likewise.
11782 (aarch64_emit_store_exclusive): Likewise.
11783 (aarch64_expand_compare_and_swap): Likewise.
11784 (aarch64_gen_atomic_cas): Likewise.
11785 (aarch64_emit_bic): Likewise.
11786 (aarch64_emit_atomic_swap): Likewise.
11787 (aarch64_emit_atomic_load_op): Likewise.
11788 (aarch64_evpc_trn): Likewise.
11789 (aarch64_evpc_uzp): Likewise.
11790 (aarch64_evpc_zip): Likewise.
11791 (aarch64_evpc_ext): Likewise.
11792 (aarch64_evpc_rev): Likewise.
11793 (aarch64_evpc_dup): Likewise.
11794 (aarch64_gen_ccmp_first): Likewise.
11795 (aarch64_gen_ccmp_next): Likewise.
11796 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11797 (alpha_emit_xfloating_libcall): Likewise.
11798 (emit_insxl): Likewise.
11799 (alpha_arg_type): Likewise.
11800 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11801 (arc_preferred_simd_mode): Likewise.
11802 (arc_secondary_reload): Likewise.
11803 (get_arc_condition_code): Likewise.
11804 (arc_print_operand): Likewise.
11805 (arc_legitimate_constant_p): Likewise.
11806 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11807 * config/arc/arc.md (casesi_load): Likewise.
11808 (casesi_compact_jump): Likewise.
11809 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11810 (cc_use_register): Likewise.
11811 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11812 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11813 (arm_init_iwmmxt_builtins): Likewise.
11814 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11815 (neon_expand_vector_init): Likewise.
11816 (arm_attr_length_move_neon): Likewise.
11817 (maybe_get_arm_condition_code): Likewise.
11818 (arm_emit_vector_const): Likewise.
11819 (arm_preferred_simd_mode): Likewise.
11820 (arm_output_iwmmxt_tinsr): Likewise.
11821 (thumb1_output_casesi): Likewise.
11822 (thumb2_output_casesi): Likewise.
11823 (arm_emit_load_exclusive): Likewise.
11824 (arm_emit_store_exclusive): Likewise.
11825 (arm_expand_compare_and_swap): Likewise.
11826 (arm_evpc_neon_vuzp): Likewise.
11827 (arm_evpc_neon_vzip): Likewise.
11828 (arm_evpc_neon_vrev): Likewise.
11829 (arm_evpc_neon_vtrn): Likewise.
11830 (arm_evpc_neon_vext): Likewise.
11831 (arm_validize_comparison): Likewise.
11832 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11833 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11834 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11835 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11836 (c6x_preferred_simd_mode): Likewise.
11837 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11838 (epiphany_rtx_costs): Likewise.
11839 * config/epiphany/predicates.md (proper_comparison_operator):
11841 * config/frv/frv.c (condexec_memory_operand): Likewise.
11842 (frv_emit_move): Likewise.
11843 (output_move_single): Likewise.
11844 (output_condmove_single): Likewise.
11845 (frv_hard_regno_mode_ok): Likewise.
11846 (frv_matching_accg_mode): Likewise.
11847 * config/h8300/h8300.c (split_adds_subs): Likewise.
11848 (h8300_rtx_costs): Likewise.
11849 (h8300_print_operand): Likewise.
11850 (compute_mov_length): Likewise.
11851 (output_logical_op): Likewise.
11852 (compute_logical_op_length): Likewise.
11853 (compute_logical_op_cc): Likewise.
11854 (h8300_shift_needs_scratch_p): Likewise.
11855 (output_a_shift): Likewise.
11856 (compute_a_shift_length): Likewise.
11857 (compute_a_shift_cc): Likewise.
11858 (expand_a_rotate): Likewise.
11859 (output_a_rotate): Likewise.
11860 * config/i386/i386.c (classify_argument): Likewise.
11861 (function_arg_advance_32): Likewise.
11862 (function_arg_32): Likewise.
11863 (function_arg_64): Likewise.
11864 (function_value_64): Likewise.
11865 (ix86_gimplify_va_arg): Likewise.
11866 (ix86_legitimate_constant_p): Likewise.
11867 (put_condition_code): Likewise.
11868 (split_double_mode): Likewise.
11869 (ix86_avx256_split_vector_move_misalign): Likewise.
11870 (ix86_expand_vector_logical_operator): Likewise.
11871 (ix86_split_idivmod): Likewise.
11872 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11873 (ix86_build_const_vector): Likewise.
11874 (ix86_build_signbit_mask): Likewise.
11875 (ix86_match_ccmode): Likewise.
11876 (ix86_cc_modes_compatible): Likewise.
11877 (ix86_expand_branch): Likewise.
11878 (ix86_expand_sse_cmp): Likewise.
11879 (ix86_expand_sse_movcc): Likewise.
11880 (ix86_expand_int_sse_cmp): Likewise.
11881 (ix86_expand_vec_perm_vpermi2): Likewise.
11882 (ix86_expand_vec_perm): Likewise.
11883 (ix86_expand_sse_unpack): Likewise.
11884 (ix86_expand_int_addcc): Likewise.
11885 (ix86_split_to_parts): Likewise.
11886 (ix86_vectorize_builtin_gather): Likewise.
11887 (ix86_vectorize_builtin_scatter): Likewise.
11888 (avx_vpermilp_parallel): Likewise.
11889 (inline_memory_move_cost): Likewise.
11890 (ix86_tieable_integer_mode_p): Likewise.
11891 (x86_maybe_negate_const_int): Likewise.
11892 (ix86_expand_vector_init_duplicate): Likewise.
11893 (ix86_expand_vector_init_one_nonzero): Likewise.
11894 (ix86_expand_vector_init_one_var): Likewise.
11895 (ix86_expand_vector_init_concat): Likewise.
11896 (ix86_expand_vector_init_interleave): Likewise.
11897 (ix86_expand_vector_init_general): Likewise.
11898 (ix86_expand_vector_set): Likewise.
11899 (ix86_expand_vector_extract): Likewise.
11900 (emit_reduc_half): Likewise.
11901 (ix86_emit_i387_round): Likewise.
11902 (ix86_mangle_type): Likewise.
11903 (ix86_expand_round_sse4): Likewise.
11904 (expand_vec_perm_blend): Likewise.
11905 (canonicalize_vector_int_perm): Likewise.
11906 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11907 (expand_vec_perm_1): Likewise.
11908 (expand_vec_perm_interleave3): Likewise.
11909 (expand_vec_perm_even_odd_pack): Likewise.
11910 (expand_vec_perm_even_odd_1): Likewise.
11911 (expand_vec_perm_broadcast_1): Likewise.
11912 (ix86_vectorize_vec_perm_const_ok): Likewise.
11913 (ix86_expand_vecop_qihi): Likewise.
11914 (ix86_expand_mul_widen_hilo): Likewise.
11915 (ix86_expand_sse2_abs): Likewise.
11916 (ix86_expand_pextr): Likewise.
11917 (ix86_expand_pinsr): Likewise.
11918 (ix86_preferred_simd_mode): Likewise.
11919 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11920 * config/i386/sse.md (*andnot<mode>3): Likewise.
11921 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11922 (*<code><mode>3): Likewise.
11923 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11924 (ia64_expand_atomic_op): Likewise.
11925 (ia64_arg_type): Likewise.
11926 (ia64_mode_to_int): Likewise.
11927 (ia64_scalar_mode_supported_p): Likewise.
11928 (ia64_vector_mode_supported_p): Likewise.
11929 (expand_vec_perm_broadcast): Likewise.
11930 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11931 (iq2000_function_arg_advance): Likewise.
11932 (iq2000_function_arg): Likewise.
11933 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11934 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11935 (m68k_libcall_value): Likewise.
11936 (m68k_function_value): Likewise.
11937 (sched_attr_op_type): Likewise.
11938 * config/mcore/mcore.c (mcore_output_move): Likewise.
11939 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11941 (microblaze_function_arg): Likewise.
11942 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11943 (mips_print_operand): Likewise.
11944 (mips_mode_ok_for_mov_fmt_p): Likewise.
11945 (mips_vector_mode_supported_p): Likewise.
11946 (mips_preferred_simd_mode): Likewise.
11947 (mips_expand_vpc_loongson_even_odd): Likewise.
11948 (mips_expand_vec_unpack): Likewise.
11949 (mips_expand_vi_broadcast): Likewise.
11950 (mips_expand_vector_init): Likewise.
11951 (mips_expand_vec_reduc): Likewise.
11952 (mips_expand_msa_cmp): Likewise.
11953 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11954 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11955 (cc_flags_for_mode): Likewise.
11956 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11957 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11958 (nds32_output_casesi_pc_relative): Likewise.
11959 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11960 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11961 (nvptx_gen_unpack): Likewise.
11962 (nvptx_gen_pack): Likewise.
11963 (nvptx_gen_shuffle): Likewise.
11964 (nvptx_gen_wcast): Likewise.
11965 (nvptx_preferred_simd_mode): Likewise.
11966 * config/pa/pa.c (pa_secondary_reload): Likewise.
11967 * config/pa/predicates.md (base14_operand): Likewise.
11968 * config/powerpcspe/powerpcspe-c.c
11969 (altivec_resolve_overloaded_builtin): Likewise.
11970 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11972 (rs6000_preferred_simd_mode): Likewise.
11973 (output_vec_const_move): Likewise.
11974 (rs6000_expand_vector_extract): Likewise.
11975 (rs6000_split_vec_extract_var): Likewise.
11976 (reg_offset_addressing_ok_p): Likewise.
11977 (rs6000_legitimate_offset_address_p): Likewise.
11978 (rs6000_legitimize_address): Likewise.
11979 (rs6000_emit_set_const): Likewise.
11980 (rs6000_const_vec): Likewise.
11981 (rs6000_emit_move): Likewise.
11982 (spe_build_register_parallel): Likewise.
11983 (rs6000_darwin64_record_arg_recurse): Likewise.
11984 (swap_selector_for_mode): Likewise.
11985 (spe_init_builtins): Likewise.
11986 (paired_init_builtins): Likewise.
11987 (altivec_init_builtins): Likewise.
11988 (do_load_for_compare): Likewise.
11989 (rs6000_generate_compare): Likewise.
11990 (rs6000_expand_float128_convert): Likewise.
11991 (emit_load_locked): Likewise.
11992 (emit_store_conditional): Likewise.
11993 (rs6000_output_function_epilogue): Likewise.
11994 (rs6000_handle_altivec_attribute): Likewise.
11995 (rs6000_function_value): Likewise.
11996 (emit_fusion_gpr_load): Likewise.
11997 (emit_fusion_p9_load): Likewise.
11998 (emit_fusion_p9_store): Likewise.
11999 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12000 (fusion_gpr_mem_load): Likewise.
12001 (fusion_addis_mem_combo_load): Likewise.
12002 (fusion_addis_mem_combo_store): Likewise.
12003 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12004 (fusion_gpr_mem_load): Likewise.
12005 (fusion_addis_mem_combo_load): Likewise.
12006 (fusion_addis_mem_combo_store): Likewise.
12007 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12009 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12010 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12011 (rs6000_preferred_simd_mode): Likewise.
12012 (output_vec_const_move): Likewise.
12013 (rs6000_expand_vector_extract): Likewise.
12014 (rs6000_split_vec_extract_var): Likewise.
12015 (reg_offset_addressing_ok_p): Likewise.
12016 (rs6000_legitimate_offset_address_p): Likewise.
12017 (rs6000_legitimize_address): Likewise.
12018 (rs6000_emit_set_const): Likewise.
12019 (rs6000_const_vec): Likewise.
12020 (rs6000_emit_move): Likewise.
12021 (rs6000_darwin64_record_arg_recurse): Likewise.
12022 (swap_selector_for_mode): Likewise.
12023 (paired_init_builtins): Likewise.
12024 (altivec_init_builtins): Likewise.
12025 (rs6000_expand_float128_convert): Likewise.
12026 (emit_load_locked): Likewise.
12027 (emit_store_conditional): Likewise.
12028 (rs6000_output_function_epilogue): Likewise.
12029 (rs6000_handle_altivec_attribute): Likewise.
12030 (rs6000_function_value): Likewise.
12031 (emit_fusion_gpr_load): Likewise.
12032 (emit_fusion_p9_load): Likewise.
12033 (emit_fusion_p9_store): Likewise.
12034 * config/rx/rx.c (rx_gen_move_template): Likewise.
12035 (flags_from_mode): Likewise.
12036 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12037 (s390_slb_comparison): Likewise.
12038 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12039 (s390_vector_mode_supported_p): Likewise.
12040 (s390_cc_modes_compatible): Likewise.
12041 (s390_match_ccmode_set): Likewise.
12042 (s390_canonicalize_comparison): Likewise.
12043 (s390_emit_compare_and_swap): Likewise.
12044 (s390_branch_condition_mask): Likewise.
12045 (s390_rtx_costs): Likewise.
12046 (s390_secondary_reload): Likewise.
12047 (__SECONDARY_RELOAD_CASE): Likewise.
12048 (s390_expand_cs): Likewise.
12049 (s390_preferred_simd_mode): Likewise.
12050 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12051 * config/sh/sh.c (sh_print_operand): Likewise.
12052 (dump_table): Likewise.
12053 (sh_secondary_reload): Likewise.
12054 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12055 * config/sh/sh.md (casesi_worker_1): Likewise.
12056 (casesi_worker_2): Likewise.
12057 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12058 (fcc_comparison_operator): Likewise.
12059 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12060 (emit_soft_tfmode_cvt): Likewise.
12061 (sparc_preferred_simd_mode): Likewise.
12062 (output_cbranch): Likewise.
12063 (sparc_print_operand): Likewise.
12064 (sparc_expand_vec_perm_bmask): Likewise.
12065 (vector_init_bshuffle): Likewise.
12066 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12067 (spu_vector_mode_supported_p): Likewise.
12068 (spu_expand_insv): Likewise.
12069 (spu_emit_branch_or_set): Likewise.
12070 (spu_handle_vector_attribute): Likewise.
12071 (spu_builtin_splats): Likewise.
12072 (spu_builtin_extract): Likewise.
12073 (spu_builtin_promote): Likewise.
12074 (spu_expand_sign_extend): Likewise.
12075 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12076 (tilegx_simd_int): Likewise.
12077 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12078 (tilepro_simd_int): Likewise.
12079 * config/v850/v850.c (const_double_split): Likewise.
12080 (v850_print_operand): Likewise.
12081 (ep_memory_offset): Likewise.
12082 * config/vax/vax.c (vax_rtx_costs): Likewise.
12083 (vax_output_int_move): Likewise.
12084 (vax_output_int_add): Likewise.
12085 (vax_output_int_subtract): Likewise.
12086 * config/visium/predicates.md (visium_branch_operator): Likewise.
12087 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12088 (visium_print_operand_address): Likewise.
12089 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12090 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12091 (xtensa_expand_conditional_branch): Likewise.
12092 (xtensa_copy_incoming_a7): Likewise.
12093 (xtensa_output_literal): Likewise.
12094 * dfp.c (decimal_real_maxval): Likewise.
12095 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12098 Alan Hayward <alan.hayward@arm.com>
12099 David Sherwood <david.sherwood@arm.com>
12101 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12102 (mode_nunits_inline): Likewise.
12103 (mode_inner_inline): Likewise.
12104 (mode_unit_size_inline): Likewise.
12105 (mode_unit_precision_inline): Likewise.
12106 (emit_insn_modes_h): Likewise. Also emit a #define of the
12108 (emit_mode_wider): Add an E_ prefix to mode names.
12109 (emit_mode_complex): Likewise.
12110 (emit_mode_inner): Likewise.
12111 (emit_mode_adjustments): Likewise.
12112 (emit_mode_int_n): Likewise.
12113 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12114 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12115 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12116 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12117 (CRC32_BUILTIN, ENTRY): Likewise.
12118 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12119 (aarch64_pop_regs): Likewise.
12120 (aarch64_process_components): Likewise.
12121 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12122 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12123 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12124 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12125 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12126 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12128 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12129 (pcmpistr): Likewise.
12130 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12131 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12132 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12134 * config/rl78/rl78.c (mduc_regs): Likewise.
12135 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12136 (htm_expand_builtin): Likewise.
12137 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12138 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12139 * config/xtensa/xtensa.c (print_operand): Likewise.
12140 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12141 (NUM_MODE_VECTOR_INT): Likewise.
12142 * genoutput.c (null_operand): Likewise.
12143 (output_operand_data): Likewise.
12144 * genrecog.c (print_parameter_value): Likewise.
12145 * lra.c (debug_operand_data): Likewise.
12147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12149 * dwarf2out.c (scompare_loc_descriptor_wide)
12150 (scompare_loc_descriptor_narrow): New functions, split out from...
12151 (scompare_loc_descriptor): ...here.
12152 * expmed.c (emit_store_flag_int): New function, split out from...
12153 (emit_store_flag): ...here.
12155 2017-08-30 Richard Biener <rguenther@suse.de>
12157 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12158 (dwarf2out_early_finish): Move setting of AT_pubnames from
12159 early debug output to early finish.
12161 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12163 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12164 and -mdata-region to the assembler.
12166 2017-08-30 Richard Biener <rguenther@suse.de>
12168 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12170 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12171 (dwarf2out_early_global_decl): Only generate a DIE for the
12172 abstract origin if it doesn't already exist or is a declaration DIE.
12173 (resolve_addr): Do not add the linkage name twice when
12174 generating a stub DIE for the DW_TAG_GNU_call_site target.
12176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12178 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12179 Use machine_mode rather than int for arg1_mode.
12181 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12184 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12185 that the second argument of the built-in functions to unpack
12186 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12187 switch statement instead a lot of if statements.
12188 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12189 Allow 64-bit values to be in Altivec registers as well as
12190 traditional floating point registers.
12191 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12193 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12195 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12196 MAX_REGS_PER_ADDRESS == 1.
12198 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12200 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12201 * config/i386/i386.c (ix86_option_override_internal): Simplify
12202 setting of opts->x_flag_entry.
12204 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12205 Jakub Jelinek <jakub@redhat.com>
12206 Richard Biener <rguenther@suse.de>
12208 PR tree-optimization/81503
12209 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12210 folded constant fits in the target type; reorder tests for clarity.
12212 2017-08-29 Martin Liska <mliska@suse.cz>
12214 * passes.def: Include pass_lower_switch.
12215 * stmt.c (dump_case_nodes): Remove and move to
12216 tree-switch-conversion.
12217 (case_values_threshold): Likewise.
12218 (expand_switch_as_decision_tree_p): Likewise.
12219 (emit_case_decision_tree): Likewise.
12220 (expand_case): Likewise.
12221 (balance_case_nodes): Likewise.
12222 (node_has_low_bound): Likewise.
12223 (node_has_high_bound): Likewise.
12224 (node_is_bounded): Likewise.
12225 (emit_case_nodes): Likewise.
12226 (struct simple_case_node): New struct.
12227 (add_case_node): Remove.
12228 (emit_case_dispatch_table): Use vector instead of case_list.
12229 (reset_out_edges_aux): Remove.
12230 (compute_cases_per_edge): Likewise.
12231 (expand_case): Build list of simple_case_node.
12232 (expand_sjlj_dispatch_table): Use it.
12233 * tree-switch-conversion.c (struct case_node): Moved from
12234 stmt.c and adjusted.
12235 (emit_case_nodes): Likewise.
12236 (node_has_low_bound): Likewise.
12237 (node_has_high_bound): Likewise.
12238 (node_is_bounded): Likewise.
12239 (case_values_threshold): Likewise.
12240 (reset_out_edges_aux): Likewise.
12241 (compute_cases_per_edge): Likewise.
12242 (add_case_node): Likewise.
12243 (dump_case_nodes): Likewise.
12244 (balance_case_nodes): Likewise.
12245 (expand_switch_as_decision_tree_p): Likewise.
12246 (emit_jump): Likewise.
12247 (emit_case_decision_tree): Likewise.
12248 (try_switch_expansion): Likewise.
12249 (do_jump_if_equal): Likewise.
12250 (emit_cmp_and_jump_insns): Likewise.
12251 (fix_phi_operands_for_edge): New function.
12252 (record_phi_operand_mapping): Likewise.
12253 (class pass_lower_switch): New pass.
12254 (pass_lower_switch::execute): New function.
12255 (make_pass_lower_switch): Likewise.
12256 (conditional_probability):
12257 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12258 * tree-pass.h: Add make_pass_lower_switch.
12260 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12263 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12266 2017-08-29 Richard Biener <rguenther@suse.de>
12268 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12269 we do not add a DW_AT_inline attribute twice.
12270 (gen_subprogram_die): Remove code setting DW_AT_inline on
12271 DECL_ABSTRACT_P nodes.
12273 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12275 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12276 calls to internal functions.
12277 (gimplify_modify_expr): Likewise.
12278 * tree-call-cdce.c (use_internal_fn): Likewise.
12279 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12280 (convert_to_divmod): Set the nothrow flag.
12281 * tree-if-conv.c (predicate_mem_writes): Likewise.
12282 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12283 (vectorizable_call): Likewise.
12284 (vectorizable_store): Likewise.
12285 (vectorizable_load): Likewise.
12286 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12287 (vect_recog_mask_conversion_pattern): Likewise.
12289 2017-08-29 Martin Liska <mliska@suse.cz>
12292 * gcc.c (driver_handle_option): Add new argument.
12293 * opts-common.c (handle_option): Pass
12294 target_option_override_hook.
12295 * opts-global.c (lang_handle_option): Add new option.
12296 (set_default_handlers): Add new argument.
12297 (decode_options): Likewise.
12298 * opts.c (target_handle_option): Likewise.
12299 (common_handle_option): Call target_option_override_hook.
12300 * opts.h (struct cl_option_handler_func): Add hook for
12301 target option override.
12302 (struct cl_option_handlers): Likewise.
12303 (set_default_handlers): Add new argument.
12304 (decode_options): Likewise.
12305 (common_handle_option): Likewise.
12306 (target_handle_option): Likewise.
12307 * toplev.c (toplev::main): Pass targetm.target_option.override
12310 2017-08-29 Richard Biener <rguenther@suse.de>
12311 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12313 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12314 life to the active subtree.
12316 2017-08-28 Jeff Law <law@redhat.com>
12318 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12319 derive_equivalences.
12320 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12321 Code moved into....
12322 (edge_info::derive_equivalences): New private member function
12324 * tree-ssa-dom.c (class edge_info): Changed from a struct
12325 to a class. Add ctor/dtor, methods and data members.
12326 (edge_info::edge_info): Renamed from allocate_edge_info.
12327 Initialize additional members.
12328 (edge_info::~edge_info): New.
12329 (free_dom_edge_info): Delete the edge info.
12330 (record_edge_info): Use new class & associated member functions.
12331 Tighten forms for testing for edge equivalences.
12332 (record_temporary_equivalences): Iterate over the simple
12333 equivalences rather than assuming there's only one per edge.
12334 (cprop_into_successor_phis): Iterate over the simple
12335 equivalences rather than assuming there's only one per edge.
12336 (optimize_stmt): Use operand_equal_p rather than pointer
12337 equality for mini-DSE code.
12339 2017-08-28 Nathan Sidwell <nathan@acm.org>
12341 * gcc.c (execute): Fold SIGPIPE handling into switch
12342 statement. Adjust internal error message.
12344 2017-08-28 Richard Biener <rguenther@suse.de>
12347 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12348 Do nothing for removed DIEs.
12350 2017-08-28 Richard Biener <rguenther@suse.de>
12352 PR tree-optimization/81977
12353 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12356 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12359 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12360 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12361 using targetm.gen_mem_thread_fence.
12363 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12366 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12367 predicate to register_operand. Reorder operands.
12368 (*btr<mode>): Ditto.
12369 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12370 (*btr<mode>_mask): Ditto.
12372 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12374 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12375 * config/rs6000/xmmintrin.h: New file.
12376 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12378 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12381 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12382 parameter and_insn and return it.
12383 (recombine_lvx_pattern): Insert a copy to ensure availability of
12384 the base register of the copied masking operation at the point of
12385 the instruction replacement.
12386 (recombine_stvx_pattern): Likewise.
12388 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12390 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12391 undocumented switches.
12392 (-mpower9-dform-vector): Likewise.
12393 (-mpower9-dform): Likewise.
12394 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12395 comments to delete references to -mpower9-dform* switches.
12396 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12397 Delete reference to -mpower9-dform* switches, test for
12398 -mpower9-vector instead.
12399 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12400 (OTHER_P9_VECTOR_MASKS): Likewise.
12401 (POWERPC_MASKS): Likewise.
12402 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12403 tests against -mpower9-dform* to -mpower9-vector. Delete code
12404 that checked for -mpower9-dform* consistancy with other options.
12405 Add test for -mpower9-misc to enable other power9 switches.
12406 (rs6000_init_hard_regno_mode_ok): Likewise.
12407 (rs6000_option_override_internal): Likewise.
12408 (rs6000_emit_prologue): Likewise.
12409 (rs6000_emit_epilogue): Likewise.
12410 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12411 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12412 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12414 (emit_fusion_p9_store): Likewise.
12415 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12416 resetting these macros if the assembler does not support ISA 3.0
12418 (TARGET_P9_DFORM_VECTOR): Likewise.
12419 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12420 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12422 2017-08-25 Alan Modra <amodra@gmail.com>
12424 PR rtl-optimization/81747
12425 * cse.c (cse_extended_basic_block): Don't attempt to record
12426 equivalences for degenerate conditional branches that branch
12427 to their fall-through.
12429 2017-08-24 Martin Sebor <msebor@redhat.com>
12431 PR middle-end/81908
12432 * gimple-fold.c (size_must_be_zero_p): New function.
12433 (gimple_fold_builtin_memory_op): Call it.
12435 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12437 * config/rs6000/mm_malloc.h: New file.
12439 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12441 PR tree-optimization/81913
12442 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12443 analysis when either IVs in condition can wrap.
12445 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12447 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12448 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12450 2017-08-24 Richard Biener <rguenther@suse.de>
12453 * targhooks.c (default_target_can_inline_p): Properly
12454 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12455 is present and always compare.
12456 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12457 infer -mfpmath=sse from TARGET_SSE_P.
12458 (ix86_can_inline_p): Properly use target_option_default_node when
12459 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12461 2017-08-24 Richard Biener <rguenther@suse.de>
12464 * dwarf2out.c (output_die): Handle flag_generate_offload like
12466 (output_comp_unit): Likewise.
12467 (gen_array_type_die): Likewise.
12468 (dwarf2out_early_finish): Likewise.
12469 (note_variable_value_in_expr): Likewise.
12470 (dwarf2out_finish): Likewise. Adjust assert.
12471 * cgraphunit.c (symbol_table::compile): Move setting of
12472 flag_generate_offload earlier ...
12473 (symbol_table::finalize_compilation_unit): ... here, before
12474 early debug finalization.
12476 2017-08-24 Richard Biener <rguenther@suse.de>
12478 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12479 and ipa-fnsummary.h.
12480 (ix86_can_inline_p): When ix86_fpmath flags do not match
12481 check whether the callee uses FP math at all.
12483 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12485 PR middle-end/81931
12486 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12487 instead of TYPE_PRECISION.
12489 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12490 Alan Hayward <alan.hayward@arm.com>
12491 David Sherwood <david.sherwood@arm.com>
12493 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12495 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12497 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12499 (ix86_frame::stack_realign_allocate): New field.
12500 (struct machine_frame_state): Modify comments.
12501 (machine_frame_state::sp_realigned_fp_end): New field.
12502 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12503 layout calculation.
12504 (sp_valid_at): Add assertion to assure no attempt to access invalid
12505 offset of a realigned stack.
12506 (fp_valid_at): Likewise.
12507 (choose_baseaddr): Modify comments.
12508 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12509 ix86_expand_prologue.
12510 (ix86_expand_prologue): Modify stack realignment and allocation.
12511 (ix86_expand_epilogue): Modify comments.
12512 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12513 avx2_runtime, avx512f, and avx512f_runtime.
12515 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12517 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12518 (mstackrealign): Do not init to -1.
12519 * config/i386/i386.c (ix86_option_override_internal):
12520 Check opts_set, not opts when setting default value of
12521 opts->x_ix86_force_align_arg_pointer.
12523 2017-08-23 Richard Biener <rguenther@suse.de>
12525 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12526 lang_hooks.decl_printable_name.
12527 * print-rtl-function.c (print_rtx_function): Likewise.
12528 * tree-pretty-print.c (dump_function_header): Likewise.
12530 2017-08-23 Richard Biener <rguenther@suse.de>
12533 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12534 -g0 at compile-time.
12536 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12538 PR middle-end/19706
12539 * doc/sourcebuild.texi (Other hardware attributes):
12542 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12544 PR middle-end/19706
12545 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12546 Fix single-use check.
12548 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12550 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12552 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12554 * doc/install.texi: Modify to add more details on running selected
12557 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12559 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12560 is combined with -mabi=ms.
12561 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12564 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12566 PR tree-optimization/81488
12567 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12568 and cached_basis fields.
12569 (MAX_SPREAD): New constant.
12570 (alloc_cand_and_find_basis): Initialize new fields.
12571 (clear_visited): New function.
12572 (create_phi_basis_1): Rename from create_phi_basis, set visited
12573 and cached_basis fields.
12574 (create_phi_basis): New wrapper function.
12575 (phi_add_costs_1): Rename from phi_add_costs, add spread
12576 parameter, set visited field, short-circuit when limits reached.
12577 (phi_add_costs): New wrapper function.
12578 (record_phi_increments_1): Rename from record_phi_increments, set
12580 (record_phi_increments): New wrapper function.
12581 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12582 (phi_incr_cost): New wrapper function.
12583 (all_phi_incrs_profitable_1): Rename from
12584 all_phi_incrs_profitable, set visited field.
12585 (all_phi_incrs_profitable): New wrapper function.
12587 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12588 Alan Hayward <alan.hayward@arm.com>
12589 David Sherwood <david.sherwood@arm.com>
12591 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12592 that takes the outer and inner modes.
12593 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12594 comparison as the canonical test for a paradoxical subreg.
12595 * combine.c (simplify_set): Use paradoxical_subreg_p.
12596 (make_extraction): Likewise.
12597 (force_to_mode): Likewise.
12598 (rtx_equal_for_field_assignment_p): Likewise.
12599 (gen_lowpart_for_combine): Likewise.
12600 (simplify_comparison): Likewise.
12601 * cse.c (equiv_constant): Likewise.
12602 * expmed.c (store_bit_field_1): Likewise.
12603 * final.c (alter_subreg): Likewise.
12604 * fwprop.c (propagate_rtx): Likewise.
12605 (forward_propagate_subreg): Likewise.
12606 * ira-conflicts.c (ira_build_conflicts): Likewise.
12607 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12608 * lra-constraints.c (curr_insn_transform): Likewise.
12609 (split_reg): Likewise.
12610 * lra-eliminations.c (move_plus_up): Likewise.
12611 (lra_eliminate_regs_1): Likewise.
12612 * recog.c (general_operand): Likewise.
12613 * ree.c (combine_reaching_defs): Likewise.
12614 * reload.c (push_reload): Likewise.
12615 (find_reloads): Likewise.
12616 * reload1.c (elimination_effects): Likewise.
12617 (compute_reload_subreg_offset): Likewise.
12618 (choose_reload_regs): Likewise.
12619 * rtlanal.c (subreg_lsb_1): Likewise.
12620 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12621 (simplify_subreg): Likewise.
12622 * var-tracking.c (track_loc_p): Likewise.
12623 * emit-rtl.c (byte_lowpart_offset): Likewise.
12624 (paradoxical_subreg_p): Delete out-of-line definition.
12626 2017-08-22 Jeff Law <law@redhat.com>
12628 PR tree-optimization/81741
12629 PR tree-optimization/71947
12630 * tree-ssa-dom.c: Include tree-inline.h.
12631 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12632 equivalences if one is more expensive to compute than the other.
12633 * tree-ssa-scopedtables.h (class const_or_copies): Make
12634 record_const_or_copy_raw method private.
12635 (class avail_exprs_stack): New method simplify_binary_operation.
12636 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12637 avail_exprs_stack::simplify_binary_operation as needed.
12638 (avail_exprs_stack::simplify_binary_operation): New function.
12640 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12642 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12643 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12644 (DOT_SYMBOLS): Likewise.
12645 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12646 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12647 (RS6000_ABI_NAME): Likewise.
12648 (TARGET_CMODEL): Likewise.
12649 (TOC_SECTION_ASM_OP): Likewise.
12650 (SET_CMODEL): New macro.
12651 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12653 2017-08-22 Richard Biener <rguenther@suse.de>
12655 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12656 to allow for free-lang-data replacements similar to verify_type_variant.
12658 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12659 Alan Hayward <alan.hayward@arm.com>
12660 David Sherwood <david.sherwood@arm.com>
12662 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12663 VOIDmode for the LABEL_REF.
12665 2017-08-22 Richard Biener <rguenther@suse.de>
12667 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12669 2017-08-22 Richard Biener <rguenther@suse.de>
12671 * common.opt (feliminate-dwarf2-dups): Ignore.
12672 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12673 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12674 same_die_p_wrap, compute_section_prefix,
12675 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12676 (comdat_symbol_id, comdat_symbol_number): Likewise.
12677 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12679 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12680 (output_die): Mark unreachable path unreachable.
12681 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12682 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12683 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12684 (dwarf2out_early_finish): Likewise.
12686 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12688 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12690 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12693 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12694 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12695 (avr_attribute_table) <io, io_low, address>: Initialize
12696 .decl_required with true.
12698 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12700 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12701 undocumented debugging options.
12702 (-mvsx-scalar-double): Likewise.
12703 (-mallow-df-permute): Likewise.
12704 (-mvectorize-builtins): Likewise.
12705 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12706 (rs6000_builtin_vectorized_function): Likewise.
12707 (rs6000_builtin_md_vectorized_function): Likewise.
12708 (rs6000_opt_vars): Likewise.
12710 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12713 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12714 (*btrq_imm): Rename from *btrq.
12715 (*btcq_imm): Rename from *btcq.
12716 (btsc): New code attribute.
12717 (*<btsc><mode>): New insn pattern.
12718 (*btr<mode>): Ditto.
12719 (*<btsc><mode>_mask): New insn_and_split pattern.
12720 (*btr<mode>_mask): Ditto.
12722 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12723 Alan Hayward <alan.hayward@arm.com>
12724 David Sherwood <david.sherwood@arm.com>
12726 * function.c (pad_below): Simplify padding calculation.
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 * target.def (function_prologue): Remove frame size argument.
12733 (function_epilogue): Likewise.
12734 * doc/tm.texi: Regenerate.
12735 * final.c (final_start_function): Update call to function_prologue.
12736 (final_end_function): Update call to function_epilogue.
12737 (default_function_pro_epilogue): Remove frame size argument.
12738 * output.h (default_function_pro_epilogue): Likewise.
12739 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12740 (arm_output_function_prologue): Likewise.
12741 * config/frv/frv.c (frv_function_prologue): Likewise.
12742 (frv_function_epilogue): Likewise.
12743 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12744 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12745 (ia64_output_function_epilogue): Likewise.
12746 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12747 (m32r_output_function_epilogue): Likewise.
12748 * config/microblaze/microblaze.c (microblaze_function_prologue)
12749 (microblaze_function_epilogue): Likewise.
12750 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12751 (mips_output_function_epilogue): Likewise.
12752 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12753 (mmix_target_asm_function_epilogue): Likewise.
12754 * config/msp430/msp430.c (msp430_start_function): Likewise.
12755 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12756 (nds32_asm_function_epilogue): Likewise.
12757 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12758 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12759 (pa_output_function_epilogue): Likewise.
12760 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12761 (rs6000_output_function_epilogue): Likewise.
12762 * config/rl78/rl78.c (rl78_start_function): Likewise.
12763 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12764 (rs6000_output_function_epilogue): Likewise.
12765 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12766 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12767 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12768 (sparc_asm_function_epilogue): Likewise.
12770 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12772 * tree.h (type_has_mode_precision_p): New function.
12773 * convert.c (convert_to_integer_1): Use it.
12774 * expr.c (expand_expr_real_2): Likewise.
12775 (expand_expr_real_1): Likewise.
12776 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12777 * match.pd: Likewise.
12778 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12779 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12780 * tree-tailcall.c (process_assignment): Likewise.
12781 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12782 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12783 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12784 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12785 (vectorizable_assignment): Likewise.
12786 (vectorizable_shift): Likewise.
12787 (vectorizable_operation): Likewise.
12788 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12790 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12792 * match.pd: Add pow (C, x) simplification.
12794 2017-08-21 Richard Biener <rguenther@suse.de>
12796 PR tree-optimization/81900
12797 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12798 for blocks with abnormal predecessors.
12799 (compute_antic): Do not set visited flag prematurely.
12801 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12804 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12806 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12808 * stor-layout.h (vector_type_mode): Move to...
12809 * tree.h (vector_type_mode): ...here.
12810 * stor-layout.c (vector_type_mode): Move to...
12811 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12813 2017-08-21 Richard Biener <rguenther@suse.de>
12815 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12816 register_external_die hooks.
12817 (debug_false_tree_charstarstar_uhwistar): Declare.
12818 (debug_nothing_tree_charstar_uhwi): Likewise.
12819 * debug.c (do_nothing_debug_hooks): Adjust.
12820 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12821 (debug_nothing_tree_charstar_uhwi): Likewise.
12822 * dbxout.c (dbx_debug_hooks): Adjust.
12823 (xcoff_debug_hooks): Likewise.
12824 * sdbout.c (sdb_debug_hooks): Likewise.
12825 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12826 * dwarf2out.c (macinfo_label_base): New global.
12827 (dwarf2out_register_external_die): New function for the
12828 register_external_die hook.
12829 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12830 (dwarf2_debug_hooks): Use them.
12831 (dwarf2_lineno_debug_hooks): Adjust.
12832 (struct die_struct): Add with_offset flag.
12833 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12834 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12835 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12836 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12837 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12838 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12839 defining section names for the early LTO debug variants.
12840 (reset_indirect_string): New helper.
12841 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12842 (print_dw_val): Add support for offsetted symbol references.
12843 (get_ultimate_context): Split out from is_cxx.
12844 (is_cxx): Use get_ultimate_context.
12845 (is_fortran): Add decl overload.
12846 (compute_comp_unit_symbol): Split out worker from
12847 compute_section_prefix.
12848 (compute_section_prefix): Call compute_comp_unit_symbol and
12849 set comdat_type_p here.
12850 (output_die): Skip DIE symbol output for the LTO added one.
12851 Handle DIE symbol references with offset.
12852 (output_comp_unit): Guard section name mangling properly.
12853 For LTO debug sections emit a symbol at the section beginning
12854 which we use to refer to its DIEs.
12855 (add_abstract_origin_attribute): For DIEs registered via
12856 dwarf2out_register_external_die directly refer to the early
12857 DIE rather than indirectly through the shadow one we created.
12858 Remove obsolete call to dwarf2out_abstract_function for
12859 non-function/block origins.
12860 (gen_array_type_die): When generating early LTO debug do
12861 not emit DW_AT_string_length.
12862 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12863 late when in LTO. As suggested place a gcc_unreachable for
12864 the DECL_ABSTRACT_P case.
12865 (gen_subprogram_die): Avoid another specification DIE
12866 for early built declarations/definitions for the late LTO case.
12867 (gen_variable_die): Add type references for late duplicated VLA dies
12869 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12870 we have the abstract instance already.
12871 (process_scope_var): Adjust decl DIE contexts in LTO which
12872 first puts them in limbo.
12873 (gen_decl_die): Do not generate type DIEs late apart from
12874 types for VLAs or for decls we do not yet have a DIE. Do not
12875 call dwarf2out_abstract_function late.
12876 (dwarf2out_early_global_decl): Make sure to create DIEs
12877 for abstract instances of a decl first.
12878 (dwarf2out_late_global_decl): Adjust comment.
12879 (output_macinfo_op): With multiple macro sections use
12880 macinfo_label_base to distinguish labels.
12881 (output_macinfo): Likewise. Update macinfo_label_base.
12882 Pass in the line info label.
12883 (note_variable_value_in_expr): When generating LTO resolve
12884 all variable values here by generating DIEs as needed.
12885 (init_sections_and_labels): Add early LTO debug flag parameter
12886 and generate different sections and names if set. Add generation
12887 counter for the labels so we can have multiple of them.
12888 (reset_dies): Helper to allow DIEs to be output multiple times.
12889 (dwarf2out_finish): When outputting DIEs to the fat part of an
12890 LTO object first reset DIEs.
12891 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12892 (modified_type_die): Check for decl_ultimate_origin being self
12894 (gen_type_die_with_usage): Likewise.
12895 (gen_typedef_die): Allow decl_ultimate_origin being self.
12896 (set_decl_abstract_flags): Remove.
12897 (set_block_abstract_flags): Likewise.
12898 (dwarf2out_abstract_function): Treat the early generated DIEs
12899 as the abstract copy and only add DW_AT_inline and
12900 DW_AT_artificial here and call set_decl_origin_self.
12901 If the DIE has an abstract origin don't do anything.
12902 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12903 if we have none yet (Go fails to build one, PR78628).
12904 (variably_modified_type_p): Prevent endless recursion for Ada
12905 cyclic pointer types.
12906 * lto-streamer-in.c: Include debug.h.
12907 (dref_queue): New global.
12908 (lto_read_tree_1): Stream in DIE references.
12909 (lto_input_tree): Register DIE references.
12910 (input_function): Stream DECL_DEBUG_ARGS.
12911 * lto-streamer-out.c: Include debug.h.
12912 (lto_write_tree_1): Output DIE references.
12913 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12914 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12915 (output_function): Stream DECL_DEBUG_ARGS.
12916 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12917 Stream DECL_ABSTRACT_ORIGIN.
12918 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12919 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12920 DECL_CONTEXT for file-scope decls.
12921 * lto-streamer.h (struct dref_entry): Declare.
12922 (dref_queue): Likewise.
12923 * cfgexpand.c (pass_expand::execute): Do not call the
12924 outlining_inline_function hook here.
12925 * lto-wrapper.c (debug_obj): New global.
12926 (tool_cleanup): Unlink it if required.
12927 (debug_objcopy): New function.
12928 (run_gcc): Handle early debug sections in the IL files by
12929 extracting them to separate files, partially linkin them and
12930 feeding the result back as result to the linker.
12931 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12932 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12933 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12934 sections into a separate segment.
12935 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12937 (darwin_asm_dwarf_section): Likewise.
12938 (darwin_asm_output_dwarf_offset): Likewise.
12939 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12941 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12942 Alan Hayward <alan.hayward@arm.com>
12943 David Sherwood <david.sherwood@arm.com>
12945 * read-md.h (md_reader::record_potential_iterator_use): Replace
12946 pointer argument with an rtx and an index.
12947 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12948 (apply_mode_iterator): Likewise.
12949 (apply_code_iterator): Likewise.
12950 (apply_int_iterator): Likewise.
12951 (apply_subst_iterator): Likewise.
12952 (record_iterator_use): Likewise.
12953 (record_attribute_use): Likewise.
12954 (md_reader::record_potential_iterator_use): Likewise. Update calls
12955 to record_iterator_use and apply_iterator.
12956 (iterator_use): Replace ptr with x and index.
12957 (attribute_use): Likewise.
12958 (apply_attribute_uses): Update calls to apply_iterator.
12959 (apply_iterators): Likewise. Update initialization of iterator_use.
12960 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12961 and record_potential_iterator_use.
12962 (rtx_reader::read_rtx_operand): Likewise.
12964 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12965 Alan Hayward <alan.hayward@arm.com>
12966 David Sherwood <david.sherwood@arm.com>
12968 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12971 2017-08-21 Richard Biener <rguenther@suse.de>
12973 PR middle-end/81884
12974 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12975 at struct end conservatively when comparing common bases.
12977 2017-08-21 Richard Biener <rguenther@suse.de>
12979 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12980 (mem_ref_in_stmt): Remove.
12981 (determine_max_movement): Use ref index to get at the reference.
12982 (invariantness_dom_walker::before_dom_children): Deal with
12983 lim data already initialized.
12984 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12986 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12988 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12989 (TARGET_ISA_ROUND): Ditto.
12990 (TARGET_ROUND): Ditto.
12991 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12992 * config/i386/i386.md: Ditto.
12993 * config/i386/sse.md: Ditto.
12994 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12995 with OPTION_MASK_ISA_SSE4_1.
12997 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13000 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13001 __builtin_ia32_lzcnt_u16.
13003 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13006 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13007 (rs6000_set_current_function): Rewrite function to use it.
13009 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13012 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13013 and TYPE_WARN_IF_NOT_ALIGN.
13014 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13015 (handle_warn_if_not_align): New.
13016 (place_union_field): Call handle_warn_if_not_align.
13017 (place_field): Call handle_warn_if_not_align.
13018 Copy TYPE_WARN_IF_NOT_ALIGN.
13019 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13020 (layout_type): Likewise.
13021 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13023 (tree_decl_common): Add warn_if_not_align.
13024 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13025 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13026 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13027 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13028 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13029 * doc/extend.texi: Document warn_if_not_aligned attribute.
13030 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13032 2017-08-17 Martin Liska <mliska@suse.cz>
13035 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13036 (get_data_dependence): Use it as pointer type.
13037 (distribute_loop): Likewise.
13039 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13041 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13042 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13043 (p8_vmrgow_<mode>_direct): New define_insn.
13044 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13045 handle endianness for vmrgew and vmrgow permute patterns.
13047 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13049 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13050 * config/rs6000/rs6000-cpus.def: Remove comment.
13051 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13052 (POWERPC_MASKS): Likewise.
13053 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13054 use of TARGET_VSX_TIMODE.
13055 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13056 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13057 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13058 (rs6000_option_override_internal): Remove dead code.
13059 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13060 (rs6000_legitimize_reload_address): Likewise.
13061 (rs6000_legitimate_address_p): Likewise.
13062 (rs6000_opt_masks): Delete "vsx-timode".
13063 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13064 from function comment.
13065 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13066 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13067 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13069 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13070 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13071 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13072 (VSX_TI): Likewise.
13074 (define_peephole2): Likewise.
13076 2017-08-17 Martin Sebor <msebor@redhat.com>
13079 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13080 past the end of an array.
13081 (test_pp_format): Add test cases.
13083 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13085 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13086 missing ECF_NOTHROW flags.
13088 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13091 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13092 operands residing in integer registers.
13093 (*vsx_le_perm_load_<mode>): Likewise.
13094 (*vsx_le_perm_store_<mode>): Likewise.
13095 (define_peephole2): Add peepholes to optimize the above.
13097 2017-08-17 Marek Polacek <polacek@redhat.com>
13099 PR middle-end/81814
13100 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13101 to mimic what shorten_compare did. Change the return type to bool.
13102 (fold_cond_expr_with_comparison): Update call to
13103 operand_equal_for_comparison_p.
13104 (fold_ternary_loc): Likewise.
13106 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13108 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13110 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13112 2017-08-17 Richard Biener <rguenther@suse.de>
13114 * tree-ssa-structalias.c (solve_graph): When propagating
13115 to successors update the graphs succ edges and avoid duplicate work.
13117 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13120 * config/i386/i386.c (ix86_option_override_internal): Save target
13121 specific options after ix86_stack_protector_guard_reg was changed.
13123 2017-08-17 Richard Biener <rguenther@suse.de>
13125 PR tree-optimization/81827
13126 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13127 (new_var_info): Initialize it conservatively.
13128 (get_call_vi): Mark register vars.
13129 (new_scalar_tmp_constraint_exp): Likewise.
13130 (handle_rhs_call): Likewise.
13131 (handle_const_call): Likewise.
13132 (create_function_info_for): Likewise.
13133 (solve_constraints): Sort varinfos to separate register from
13134 non-register vars to pack points-to solution bitmaps during
13137 2017-08-17 Marek Polacek <polacek@redhat.com>
13139 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13141 2017-08-17 Richard Biener <rguenther@suse.de>
13143 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13144 to true when overflow is undefined and we saturated the result.
13146 2017-08-17 Alan Modra <amodra@gmail.com>
13149 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13150 Don't use store multiple if only one reg needs saving.
13151 (interesting_frame_related_regno): New function.
13152 (rs6000_frame_related): Don't emit frame info for regs that
13154 (rs6000_emit_epilogue): Likewise.
13156 2017-08-16 Nathan Sidwell <nathan@acm.org>
13158 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13159 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13160 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13161 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13162 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13163 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13164 (verify_type): Adjust for TYPE_BINFO move.
13165 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13166 process TYPE_BINFO directly.
13167 (hash_tree): Likewise.
13168 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13170 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13173 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13175 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13177 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13180 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13181 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13182 Add ix86_binary_operator_ok to insn constraint.
13183 (*iordi_1_bts): Ditto.
13184 (*xordi_1_btc): Ditto.
13185 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13186 Update corresponding peephole2 pattern.
13190 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13192 PR tree-optimization/81832
13193 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13194 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13196 2017-08-16 Marek Polacek <polacek@redhat.com>
13199 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13200 regarding size_zero_node.
13202 2017-08-16 Martin Liska <mliska@suse.cz>
13205 * config.gcc: Respect previously set extra_objs in case
13208 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13210 PR tree-optimization/81835
13211 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13212 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13213 not depend on the phi.
13215 2017-08-16 Alan Modra <amodra@gmail.com>
13217 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13220 2017-08-16 Alan Modra <amodra@gmail.com>
13222 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13223 (save_reg_p): ..into this. Update all callers.
13224 (first_reg_to_save): Simplify.
13226 2017-08-16 Alan Modra <amodra@gmail.com>
13228 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13231 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13235 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13236 constant count if that count is less than 32.
13238 2017-08-15 Nathan Sidwell <nathan@acm.org>
13240 * gcc.c (execute): Emit friendlier message if inferior is killed
13241 by an external cause.
13243 2017-08-15 Richard Biener <rguenther@suse.de>
13245 PR tree-optimization/81790
13246 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13247 CONSTRUCTORs from simplifying and VN.
13249 2017-08-14 Martin Sebor <msebor@redhat.com>
13251 * builtin-attrs.def: Add comments.
13253 2017-08-14 Martin Sebor <msebor@redhat.com>
13256 * doc/extend.texi (attribute nonstring): Document new attribute.
13258 2017-08-14 Martin Sebor <msebor@redhat.com>
13261 * tree-diagnostic.c (default_tree_printer): Handle %G.
13262 * gimple-pretty-print.h (percent_G_format): Declare new function.
13263 * gimple-pretty-print.c (percent_G_format): Define.
13264 * tree-pretty-print.c (percent_K_format): Add argument.
13266 2017-08-14 Martin Sebor <msebor@redhat.com>
13268 PR translation/79998
13269 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13270 Remove a stray space.
13272 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13275 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13276 (*iordi_1_bts): Ditto.
13277 (*xordi_1_btc): Ditto.
13279 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13282 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13283 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13285 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13286 quoted strings, and make more translator-friendly.
13287 (darwin_rs6000_override_options): Likewise.
13288 (rs6000_option_override_internal): Likewise.
13289 (rs6000_return_in_memory): Fix overlong line.
13290 (init_cmulative_args): Use quoted strings, and make more
13291 translator-friendly.
13292 (rs6000_pass_by_reference): Fix overlong line.
13293 (def_builtin): Use quoted strings.
13294 (altivec_expand_predicate_builtin): Use quoted strings, and make
13295 more translator-friendly.
13296 (htm_expand_builtin): Use quoted strings.
13297 (cpu_expand_builtin): Use quoted strings, and make more
13298 translator-friendly.
13299 (altivec_expand_builtin): Likewise.
13300 (paired_expand_predicate_builtin): Likewise.
13301 (rs6000_invalid_builtin): Likewise.
13302 (builtin_function_type): Use quoted strings.
13303 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13304 more translator-friendly.
13305 (rs6000_trampoline_init): Likewise.
13306 (rs6000_handle_altivec_attribute): Likewise.
13307 (rs6000_inner_target_options): Use quoted strings.
13308 (rs6000_disable_incompatible_switches): Likewise.
13309 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13310 strings, and make more translator-friendly.
13311 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13313 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13315 PR tree-optimization/81799
13316 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13317 cond_expr to simple gimple operand.
13319 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13321 PR middle-end/46932
13322 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13324 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13328 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13329 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13330 TARGET_GASISR_PROLOGUES.
13331 * config/avr/avr.c (avr_option_override): Same.
13332 (avr_pass_pre_proep::execute): Same.
13334 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13337 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13338 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13340 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13342 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13343 define_insn_and_split. Split to a memory load from 0 in
13344 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13345 using PTR mode iterator.
13346 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13347 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13348 (*add_tp_<mode>): Redefine as define_insn_and_split.
13349 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13350 address space. Merge with *add_tp_x32 using PTR mode iterator.
13351 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13352 Split to an add with a memory load from 0 in
13353 DEFAULT_TLS_SEG_REG address space.
13355 2017-08-12 Andrew Pinski <apinski@cavium.com>
13357 * config/aarch64/aarch64-option-extensions.def (rdma):
13358 Fix feature string to what Linux prints out in /proc/cpuinfo.
13360 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13363 * dwarf2out.c (modified_type_die): For C typedef types that have
13364 an ultimate origin, process the ultimate origin instead of the
13366 (gen_typedef_die): Assert that input DECLs have no ultimate
13368 (gen_type_die_with_usage): For typedef variants that have an
13369 ultimate origin, just call gen_decl_die on the original DECL.
13370 (process_scope_var): Avoid creating DIEs for local typedefs and
13371 concrete static variables.
13373 2017-08-12 Alan Modra <amodra@gmail.com>
13377 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13378 match gnu-user.h startfile.
13379 (ENDFILE_LINUX_SPEC): Similarly.
13381 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13384 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13386 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13388 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13389 * config/aarch64/aarch64.md (mov<mode>): Change.
13390 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13391 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13392 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13394 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13396 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13397 for storage order barriers.
13399 2017-08-11 Martin Liska <mliska@suse.cz>
13402 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13403 variables of void type.
13405 2017-08-11 Martin Liska <mliska@suse.cz>
13407 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13408 TARGET_SUPPORTS_ALIASES.
13409 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13410 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13411 (optimize_weakref): Likewise.
13412 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13413 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13414 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13416 2017-08-11 Martin Liska <mliska@suse.cz>
13419 * config/i386/i386.c (make_resolver_func): Do complete
13420 refactoring of the function.
13422 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13425 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13426 * config/i386/i386.c (ix86_stack_protect_guard): Use
13427 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13428 * doc/invoke.texi (x86 Options): Document
13429 -mstack-protector-guard-symbol= option.
13431 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13433 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13434 * config/i386/i386.c (ix86_split_stack_guard): New function.
13435 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13436 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13437 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13438 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13439 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13440 (split_stack_space_check): Call ix86_split_stack_guard.
13442 2017-08-10 Martin Sebor <msebor@redhat.com>
13444 * print-tree.c (print_node): Print location using the established
13446 Replace spaces with colons.
13447 (debug_raw, debug): Ditto.
13449 2017-08-10 Martin Sebor <msebor@redhat.com>
13452 * pretty-print.c (pp_format): Correct the handling of %s precision.
13454 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13457 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13459 (ix86_finalize_stack_frame_flags): This. Also clear
13460 frame_pointer_needed if -fno-omit-frame-pointer is used without
13462 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13463 with ix86_finalize_stack_frame_flags.
13464 (ix86_expand_epilogue): Likewise.
13465 (ix86_expand_split_stack_prologue): Likewise.
13466 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13468 2017-08-10 Martin Liska <mliska@suse.cz>
13471 * c-attribs.c (handle_target_attribute):
13472 Report warning for an empty string argument of target attribute.
13474 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13477 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13479 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13480 or DECL_NONLOCAL labels.
13481 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13482 or DECL_NONLOCAL labels here.
13484 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13486 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13487 to indicate when early gimple folding has been disabled.
13488 (rs6000_gimple_fold_builtin): Add debug content.
13489 (rs6000_invalid_builtin): Fix whitespace.
13490 (rs6000_expand_builtin): Fix whitespace.
13491 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13493 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13496 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13497 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13499 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13501 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13502 (qdf24xx): Likewise.
13503 * config/aarch64/aarch64.md: Include falkor.md.
13504 * config/aarch64/falkor.md: New.
13506 2017-08-09 Marek Polacek <polacek@redhat.com>
13509 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13510 * diagnostic.c (emit_diagnostic): Add a comment.
13511 (emit_diagnostic_valist): New function.
13513 2017-08-09 Marek Polacek <polacek@redhat.com>
13516 * input.c (make_location): New overload.
13517 * input.h (make_location): Declare.
13519 2017-08-08 Alan Modra <amodra@gmail.com>
13520 H.J. Lu <hongjiu.lu@intel.com>
13523 * gcc.c (NO_PIE_SPEC): Delete.
13524 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13526 (LINK_PIE_SPEC): ..to here.
13527 (LINK_COMMAND_SPEC): Support -no-pie.
13528 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13529 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13531 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13532 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13533 (ENDFILE_CRTEND_SPEC): Similarly.
13535 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13538 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13539 (mstack-protector-guard-offset=): Ditto.
13540 * config/i386/i386.c (ix86_option_override): Handle
13541 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13543 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13544 ix86_stack_protect_guard_offset variables.
13545 (TARGET_STACK_PROTECT_GUARD): Always define.
13546 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13547 and -mstack-protector-guard-offset= options.
13549 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13551 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13552 boundary case for the last candidate.
13554 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13556 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13557 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13559 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13561 PR middle-end/19706
13562 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13563 * config/aarch64/aarch64-builtins.c
13564 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13565 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13566 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13568 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13569 Andrew Pinski <pinskia@gmail.com>
13571 PR middle-end/19706
13572 * internal-fn.def (XORSIGN): New.
13573 * optabs.def (xorsign_optab): New.
13574 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13575 (convert_expand_mult_copysign): New.
13576 (pass_optimize_widening_mul::execute): Call
13577 convert_expand_mult_copysign.
13579 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13581 PR tree-optimization/81354
13582 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13583 Insert on edges rather than explicitly creating landing pads.
13584 (analyze_candidates_and_replace): Commit edge inserts.
13586 2017-08-08 Richard Biener <rguenther@suse.de>
13588 PR middle-end/81719
13589 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13590 (expand_simple_operations): Also look through ADDR_EXPRs with
13591 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13593 2017-08-08 Richard Biener <rguenther@suse.de>
13595 PR tree-optimization/81723
13596 * tree-vect-slp.c (struct bst_traits): New hash traits.
13597 (bst_fail): New global.
13598 (vect_build_slp_tree_2): New worker, split out from ...
13599 (vect_build_slp_tree): ... this now wrapping it with using
13600 bst_fail set to cache SLP tree build fails. Properly handle
13602 (vect_analyze_slp_instance): Allocate and free bst_fail.
13604 2017-08-08 Martin Liska <mliska@suse.cz>
13607 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13608 LABEL_DECLs that can be from a different function.
13610 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13612 PR tree-optimization/81744
13613 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13614 loop's number of iterations.
13616 2017-08-08 Martin Liska <mliska@suse.cz>
13618 * asan.c: Include header files.
13619 * attribs.c (build_decl_attribute_variant): New function moved
13621 (build_type_attribute_qual_variant): Likewise.
13622 (cmp_attrib_identifiers): Likewise.
13623 (simple_cst_list_equal): Likewise.
13624 (omp_declare_simd_clauses_equal): Likewise.
13625 (attribute_value_equal): Likewise.
13626 (comp_type_attributes): Likewise.
13627 (build_type_attribute_variant): Likewise.
13628 (lookup_ident_attribute): Likewise.
13629 (remove_attribute): Likewise.
13630 (merge_attributes): Likewise.
13631 (merge_type_attributes): Likewise.
13632 (merge_decl_attributes): Likewise.
13633 (merge_dllimport_decl_attributes): Likewise.
13634 (handle_dll_attribute): Likewise.
13635 (attribute_list_equal): Likewise.
13636 (attribute_list_contained): Likewise.
13637 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13638 (lookup_attribute_by_prefix): Likewise.
13639 * bb-reorder.c: Include header files.
13640 * builtins.c: Likewise.
13641 * calls.c: Likewise.
13642 * cfgexpand.c: Likewise.
13643 * cgraph.c: Likewise.
13644 * cgraphunit.c: Likewise.
13645 * convert.c: Likewise.
13646 * dwarf2out.c: Likewise.
13647 * final.c: Likewise.
13648 * fold-const.c: Likewise.
13649 * function.c: Likewise.
13650 * gimple-expr.c: Likewise.
13651 * gimple-fold.c: Likewise.
13652 * gimple-pretty-print.c: Likewise.
13653 * gimple.c: Likewise.
13654 * gimplify.c: Likewise.
13655 * hsa-common.c: Likewise.
13656 * hsa-gen.c: Likewise.
13657 * internal-fn.c: Likewise.
13658 * ipa-chkp.c: Likewise.
13659 * ipa-cp.c: Likewise.
13660 * ipa-devirt.c: Likewise.
13661 * ipa-fnsummary.c: Likewise.
13662 * ipa-inline.c: Likewise.
13663 * ipa-visibility.c: Likewise.
13665 * lto-cgraph.c: Likewise.
13666 * omp-expand.c: Likewise.
13667 * omp-general.c: Likewise.
13668 * omp-low.c: Likewise.
13669 * omp-offload.c: Likewise.
13670 * omp-simd-clone.c: Likewise.
13671 * opts-global.c: Likewise.
13672 * passes.c: Likewise.
13673 * predict.c: Likewise.
13674 * sancov.c: Likewise.
13675 * sanopt.c: Likewise.
13676 * symtab.c: Likewise.
13677 * toplev.c: Likewise.
13678 * trans-mem.c: Likewise.
13679 * tree-chkp.c: Likewise.
13680 * tree-eh.c: Likewise.
13681 * tree-into-ssa.c: Likewise.
13682 * tree-object-size.c: Likewise.
13683 * tree-parloops.c: Likewise.
13684 * tree-profile.c: Likewise.
13685 * tree-ssa-ccp.c: Likewise.
13686 * tree-ssa-live.c: Likewise.
13687 * tree-ssa-loop.c: Likewise.
13688 * tree-ssa-sccvn.c: Likewise.
13689 * tree-ssa-structalias.c: Likewise.
13690 * tree-ssa.c: Likewise.
13691 * tree-streamer-in.c: Likewise.
13692 * tree-vectorizer.c: Likewise.
13693 * tree-vrp.c: Likewise.
13694 * tsan.c: Likewise.
13695 * ubsan.c: Likewise.
13696 * varasm.c: Likewise.
13697 * varpool.c: Likewise.
13698 * tree.c: Remove functions moved to attribs.[ch].
13699 * tree.h: Likewise.
13700 * config/aarch64/aarch64.c: Add attrs.h header file.
13701 * config/alpha/alpha.c: Likewise.
13702 * config/arc/arc.c: Likewise.
13703 * config/arm/arm.c: Likewise.
13704 * config/avr/avr.c: Likewise.
13705 * config/bfin/bfin.c: Likewise.
13706 * config/c6x/c6x.c: Likewise.
13707 * config/cr16/cr16.c: Likewise.
13708 * config/cris/cris.c: Likewise.
13709 * config/darwin.c: Likewise.
13710 * config/epiphany/epiphany.c: Likewise.
13711 * config/fr30/fr30.c: Likewise.
13712 * config/frv/frv.c: Likewise.
13713 * config/ft32/ft32.c: Likewise.
13714 * config/h8300/h8300.c: Likewise.
13715 * config/i386/winnt.c: Likewise.
13716 * config/ia64/ia64.c: Likewise.
13717 * config/iq2000/iq2000.c: Likewise.
13718 * config/lm32/lm32.c: Likewise.
13719 * config/m32c/m32c.c: Likewise.
13720 * config/m32r/m32r.c: Likewise.
13721 * config/m68k/m68k.c: Likewise.
13722 * config/mcore/mcore.c: Likewise.
13723 * config/microblaze/microblaze.c: Likewise.
13724 * config/mips/mips.c: Likewise.
13725 * config/mmix/mmix.c: Likewise.
13726 * config/mn10300/mn10300.c: Likewise.
13727 * config/moxie/moxie.c: Likewise.
13728 * config/msp430/msp430.c: Likewise.
13729 * config/nds32/nds32-isr.c: Likewise.
13730 * config/nds32/nds32.c: Likewise.
13731 * config/nios2/nios2.c: Likewise.
13732 * config/nvptx/nvptx.c: Likewise.
13733 * config/pa/pa.c: Likewise.
13734 * config/pdp11/pdp11.c: Likewise.
13735 * config/powerpcspe/powerpcspe.c: Likewise.
13736 * config/riscv/riscv.c: Likewise.
13737 * config/rl78/rl78.c: Likewise.
13738 * config/rx/rx.c: Likewise.
13739 * config/s390/s390.c: Likewise.
13740 * config/sh/sh.c: Likewise.
13741 * config/sol2.c: Likewise.
13742 * config/sparc/sparc.c: Likewise.
13743 * config/spu/spu.c: Likewise.
13744 * config/stormy16/stormy16.c: Likewise.
13745 * config/tilegx/tilegx.c: Likewise.
13746 * config/tilepro/tilepro.c: Likewise.
13747 * config/v850/v850.c: Likewise.
13748 * config/vax/vax.c: Likewise.
13749 * config/visium/visium.c: Likewise.
13750 * config/xtensa/xtensa.c: Likewise.
13752 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13755 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13756 constraints since the -mupper-regs-* switches have been
13758 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13759 into a vector from a double word element that was extracted from
13760 another vector, and eliminate extra XXPERMDI instructions.
13761 (vsx_concat_<mode>_2): Likewise.
13762 (vsx_concat_<mode>_3): Likewise.
13763 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13764 concat to allow optimizing inserts from previous extracts.
13766 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13768 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13769 memory reference to a SSP offset in TLS address space.
13770 (ix86_print_operand) <case '@'>: Remove.
13771 (ix86_print_operand_punct_valid_p): Remove '@' code.
13772 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13773 UNSPEC_SP_TLS_TEST.
13774 (stack_tls_protect_set_<mode>): Remove.
13775 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13776 (stack_tls_protect_test_<mode>): Remove.
13777 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13779 2017-08-07 Olivier Hainque <hainque@adacore.com>
13782 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13784 2017-08-07 Douglas Rupp <rupp@adacore.com>
13786 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13787 variable was referenced as multidir in command.
13789 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13792 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13795 2017-08-07 Martin Liska <mliska@suse.cz>
13797 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13798 * config/rl78/rl78.c: Add include of attribs.h.
13799 * config/sh/sh.c: Likewise.
13800 * config/v850/v850.c: Likewise.
13802 2017-08-07 Tom de Vries <tom@codesourcery.com>
13804 PR middle-end/78266
13805 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13807 2017-08-07 Martin Liska <mliska@suse.cz>
13809 * config/mips/mips.c: Include attribs.h.
13811 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13814 * doc/invoke.texi: Document change in behvaior for -Ofast for
13817 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13819 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13821 (aarch64_pop_regs): Likewise.
13822 (aarch64_gen_load_pair): Likewise.
13823 (aarch64_save_callee_saves): Likewise.
13824 (aarch64_restore_callee_saves): Likewise.
13826 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13828 * config/i386/i386.c: Revert the last change.
13830 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13833 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13835 (ix86_finalize_stack_frame_flags): This. Also clear
13836 frame_pointer_needed if -fno-omit-frame-pointer is used without
13838 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13839 with ix86_finalize_stack_frame_flags.
13840 (ix86_expand_epilogue): Likewise.
13841 (ix86_expand_split_stack_prologue): Likewise.
13843 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13846 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13847 to P_AES for Westmere.
13849 2017-08-07 Jonathan Yong <10walls@gmail.com>
13851 * config/i386/mingw.opt (fset-stack-executable): Removed.
13852 * config/i386/cygming.opt (fset-stack-executable): Moved
13854 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13856 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13858 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13860 2017-08-07 Marek Polacek <polacek@redhat.com>
13862 PR middle-end/81737
13863 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13865 2017-08-07 Martin Liska <mliska@suse.cz>
13867 * attribs.h (canonicalize_attr_name): New function.
13868 (cmp_attribs): Move from c-format.c and adjusted.
13869 (is_attribute_p): Moved from tree.h.
13870 * tree-inline.c: Add new includes.
13871 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13872 (private_is_attribute_p): Remove.
13873 (private_lookup_attribute): Likewise.
13874 (private_lookup_attribute_by_prefix): Simplify.
13875 (remove_attribute): Use is_attribute_p.
13876 * tree.h: Remove removed declarations.
13878 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13880 PR middle-end/81698
13881 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13882 instead of computing it in the function. Formatting fix.
13883 (expand_case): Don't rely on default_edge being the first edge,
13884 clear it if removing it, pass default_edge to
13885 emit_case_dispatch_table.
13886 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13889 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13891 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13892 insn in the function, emit NOP after the insn.
13894 2017-08-06 Tom de Vries <tom@codesourcery.com>
13896 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13899 2017-08-06 Tom de Vries <tom@codesourcery.com>
13901 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13904 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13906 PR tree-optimization/57371
13907 * match.pd: New pattern.
13909 2017-08-04 Marek Polacek <polacek@redhat.com>
13911 PR middle-end/81695
13912 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13913 perform the computation in offset_int.
13915 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13917 PR tree-optimization/81136
13918 * tree-vectorizer.h: Include tree-hash-traits.h.
13919 (vec_base_alignments): New typedef.
13920 (vec_info): Add a base_alignments field.
13921 (vect_record_base_alignments): Declare.
13922 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13924 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13925 (create_data_ref): Add an is_conditional_in_stmt argument.
13926 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13927 the is_conditional_in_stmt field.
13928 (data_ref_loc): Add an is_conditional_in_stmt field.
13929 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13930 (find_data_references_in_stmt): Update call to create_data_ref.
13931 (graphite_find_data_references_in_stmt): Likewise.
13932 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13933 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13934 (vect_record_base_alignment): New function.
13935 (vect_record_base_alignments): Likewise.
13936 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13937 for nested statements even if we fail to compute a misalignment.
13938 Use pooled base alignments for unconditional references.
13939 (vect_find_same_alignment_drs): Compare base addresses instead
13941 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13942 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13944 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13946 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13947 Add an explicit name for the enum. Use auto_vec for slp_instances
13948 and grouped_stores.
13949 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13951 (_bb_vec_info): Add a constructor and destructor.
13952 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13953 (destroy_loop_vec_info): Delete.
13954 (vect_destroy_datarefs): Likewise.
13955 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13956 (vec_info::vec_info): New function.
13957 (vec_info::~vec_info): Likewise.
13958 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13959 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13960 destroy_loop_vec_info.
13961 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13962 (_loop_vec_info::_loop_vec_info): ...this.
13963 (destroy_loop_vec_info): Replace with...
13964 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13965 the stmt_vec_infos. Leave handling of vec_info information to its
13966 destructor. Remove explicit vector releases.
13967 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13968 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13969 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13970 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13971 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13972 (destroy_bb_vec_info): Replace with...
13973 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13974 information to its destructor.
13975 (vect_slp_analyze_bb_1): Use new and delete instead of
13976 new_bb_vec_info and destroy_bb_vec_info.
13977 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13980 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13982 * tree-data-ref.h (subscript): Add access_fn field.
13983 (data_dependence_relation): Add could_be_independent_p.
13984 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13985 (same_access_functions): Move to tree-data-ref.c.
13986 * tree-data-ref.c (ref_contains_union_access_p): New function.
13987 (access_fn_component_p): Likewise.
13988 (access_fn_components_comparable_p): Likewise.
13989 (dr_analyze_indices): Add a reference to access_fn_component_p.
13990 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13992 (constant_access_functions): Likewise.
13993 (add_other_self_distances): Likewise.
13994 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13995 (initialize_data_dependence_relation): Use XCNEW and remove
13996 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13997 of access functions that have the same type. Allow the
13998 subsequence to end with different bases in some circumstances.
13999 Record the chosen access functions in SUB_ACCESS_FN.
14000 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14001 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14002 (subscript_dependence_tester_1): Likewise dra and drb.
14003 (build_classic_dist_vector): Update calls accordingly.
14004 (subscript_dependence_tester): Likewise.
14005 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14006 DDR_COULD_BE_INDEPENDENT_P.
14007 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14008 comp_alias_ddrs instead of may_alias_ddrs.
14009 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14011 (vect_analyze_data_ref_dependence): Use it if
14012 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14013 distance vectors if that fails.
14014 (dependence_distance_ge_vf): New function.
14015 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14016 LOOP_VINFO_MAY_ALIAS_DDRS.
14018 2017-08-04 Richard Biener <rguenther@suse.de>
14020 PR middle-end/81705
14021 * fold-const.c (fold_binary_loc): Properly restrict
14022 minus_var0 && minus_var1 case when associating undefined overflow
14025 2017-08-04 Tom de Vries <tom@codesourcery.com>
14027 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14029 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14031 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14032 Don't start diagnostic messages with a capital letter.
14033 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14035 (rs6000_invalid_builtin): Likewise.
14036 (rs6000_trampoline_init): Likewise.
14038 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14041 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14042 after setting changeable df flags.
14044 2017-08-03 Richard Biener <rguenther@suse.de>
14046 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14047 up if the use is in USE - X.
14049 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14051 * toplev.c (dumpfile.h): New include.
14052 (internal_error_reentered): New static function. Use it...
14053 (internal_error_function): ...here to handle reentered internal_error.
14055 2017-08-03 Richard Biener <rguenther@suse.de>
14057 PR middle-end/81148
14058 * fold-const.c (split_tree): Add minus_var and minus_con
14059 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14060 here but always use minus_*.
14061 (associate_trees): Assert we never associate with MINUS_EXPR
14062 and NULL first operand. Do not recurse for PLUS_EXPR operands
14063 when associating as MINUS_EXPR either.
14064 (fold_binary_loc): Track minus_var and minus_con.
14066 2017-08-03 Tom de Vries <tom@codesourcery.com>
14069 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14070 ACCEL_COMPILER, apply finish_options on
14071 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14073 2017-08-03 Tom de Vries <tom@codesourcery.com>
14076 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14077 function_entry_patch_area_size > 0.
14079 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14082 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14083 instead of 10??LU, perform unit multiplication in wide_int,
14084 don't change alloc_object_size_limit if the limit is larger
14087 PR tree-optimization/81655
14088 PR tree-optimization/81588
14089 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14090 the case when ranges[i].low and high are 1 for unsigned type with
14093 PR middle-end/81052
14094 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14095 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14097 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14099 * tree-vrp.h: Add include guard.
14101 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14104 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14105 (ud2): New insn pattern.
14106 * config/i386/i386.c (ix86_expand_epilogue):
14107 For naked functions, generate ud2 instead of trap insn.
14109 2017-08-02 Marek Polacek <polacek@redhat.com>
14112 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14114 2017-08-02 Tom de Vries <tom@codesourcery.com>
14115 Cesar Philippidis <cesar@codesourcery.com>
14117 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14118 Add missing edge probabilities.
14120 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14122 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14123 Correct endianness.
14125 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14127 PR middle-end/79499
14128 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14129 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14130 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14131 prologue_seq sequences - if any.
14133 2017-08-02 Richard Biener <rguenther@suse.de>
14135 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14136 via vectors if supported, integer extracts via punning if supported
14137 or otherwise vector extracts.
14139 2017-08-02 Richard Biener <rguenther@suse.de>
14141 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14143 (bitmap_insert_into_set): ... this.
14145 2017-08-02 Richard Biener <rguenther@suse.de>
14147 PR tree-optimization/81633
14149 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14151 PR tree-optimization/71752
14152 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14154 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14156 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14157 (machine_function::call_ms2sysv_pad_out): Remove field.
14158 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14159 (ix86_compute_frame_layout): Likewise.
14161 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14164 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14165 attribute with interrupt attribute.
14167 2017-08-01 Andrew Pinski <apinski@cavium.com>
14169 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14170 BIT_INSERT_EXPR's operand 1
14171 to see if the types precision matches.
14173 2017-08-01 Martin Liska <mliska@suse.cz>
14175 PR middle-end/70140
14176 * builtins.c (expand_builtin_memcpy_args): Remove.
14177 (expand_builtin_memcpy): Call newly added function
14178 expand_builtin_memory_copy_args.
14179 (expand_builtin_memcpy_with_bounds): Likewise.
14180 (expand_builtin_mempcpy): Remove last argument.
14181 (expand_builtin_mempcpy_with_bounds): Likewise.
14182 (expand_builtin_memory_copy_args): New function created from
14183 expand_builtin_mempcpy_args with small modifications.
14184 (expand_builtin_mempcpy_args): Remove.
14185 (expand_builtin_stpcpy): Remove unused argument.
14186 (expand_builtin): Likewise.
14187 (expand_builtin_with_bounds): Likewise.
14189 2017-08-01 Martin Liska <mliska@suse.cz>
14192 Make mempcpy more optimal (PR middle-end/70140).
14194 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14197 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14198 __builtin_vec_cmpne verify both arguments are compatible vectors
14199 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14200 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14201 move computation of aligned to after checking the argument types.
14205 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14206 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14209 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14210 Doug Rupp <rupp@adacore.com>
14211 Olivier Hainque <hainque@adacore.com>
14213 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14214 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14216 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14217 for TARGET_OS_CPP_BUILTIN.
14218 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14219 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14221 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14222 passing required abi options to the assembler for EABI configurations.
14223 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14224 of .text.hot and .text.unlikely sections for kernel modules when
14225 using ARM style exceptions.
14226 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14227 options. Add EXTRA_CC1_SPEC.
14228 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14230 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14232 (ARM_TARGET2_DWARF_FORMAT): Define.
14233 * config/arm/t-vxworks: Adjust multilib control to removal of the
14234 Diab command line options.
14236 2017-08-01 Martin Liska <mliska@suse.cz>
14238 PR gcov-profile/81561
14239 * gcov.c (unblock): Make unblocking safe as we need to preserve
14240 index correspondence of blocks and block_lists.
14242 2017-08-01 Richard Biener <rguenther@suse.de>
14244 PR tree-optimization/81181
14245 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14246 (compute_antic): ... end of iteration here.
14248 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14250 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14251 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14252 (ftree-slp-vectorize): Likewise.
14253 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14254 can no longer be set independent of flag_tree_loop_vectorize.
14255 * omp-general.c (emp_max_vf): Likewise.
14256 * opts.c (enable_fdo_optimizations): Remove references to
14257 flag_tree_vectorize, these are now implicit.
14258 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14259 and leave it for the options machinery.
14261 2017-08-01 Martin Liska <mliska@suse.cz>
14263 PR middle-end/70140
14264 * builtins.c (expand_builtin_memcpy_args): Remove.
14265 (expand_builtin_memcpy): Call newly added function
14266 expand_builtin_memory_copy_args.
14267 (expand_builtin_memcpy_with_bounds): Likewise.
14268 (expand_builtin_mempcpy): Remove last argument.
14269 (expand_builtin_mempcpy_with_bounds): Likewise.
14270 (expand_builtin_memory_copy_args): New function created from
14271 expand_builtin_mempcpy_args with small modifications.
14272 (expand_builtin_mempcpy_args): Remove.
14273 (expand_builtin_stpcpy): Remove unused argument.
14274 (expand_builtin): Likewise.
14275 (expand_builtin_with_bounds): Likewise.
14277 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14280 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14281 print "ds:" only for immediates in generic address space.
14283 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14286 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14287 (ix86_function_ok_for_sibcall): Return false for naked functions.
14289 2017-08-01 Richard Biener <rguenther@suse.de>
14291 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14292 (compute_antic): Seed worklist with exit block predecessors.
14293 * cfganal.c (dfs_find_deadend): For a cycle return the source
14294 of the edge closing it.
14296 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14298 * config/aarch64/aarch64.c
14299 (aarch64_can_const_movi_rtx_p): Move 0 check.
14301 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14303 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14304 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14306 * match.pd: Ditto in address comparison pattern.
14308 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14310 PR tree-optimization/81627
14311 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14312 closed ssa form for store-store chain.
14314 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14316 PR tree-optimization/81620
14317 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14318 for store-store chain.
14320 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14322 PR tree-optimization/81588
14323 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14324 ranges[i].in_p, invert comparison code ccode. For >/>=,
14325 swap rhs1 and rhs2 and comparison code unconditionally,
14326 for </<= don't do that. Don't swap rhs1/rhs2 again if
14327 ranges[i].in_p, instead invert comparison code ccode if
14328 opcode or oe->rank is BIT_IOR_EXPR.
14331 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14332 a direct optab to conversion optab.
14333 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14334 with GET_MODE_INNER as last argument instead of optab_handler.
14335 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14336 vector extraction if possible and optab is available.
14337 * expr.c (store_constructor): Use convert_optab_handler instead
14338 of optab_handler. Use vector initialization from smaller
14339 vectors if possible and optab is available.
14340 * tree-vect-stmts.c (vectorizable_load): Likewise.
14341 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14342 now have two modes.
14343 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14344 of vec_init from half-sized vectors with the same element mode.
14345 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14346 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14347 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14348 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14349 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14350 after mode in gen_vec_extract* calls.
14351 (vec_extract<mode>): Renamed to ...
14352 (vec_extract<mode><ssescalarmodelower>): ... this.
14353 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14354 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14355 element mode after mode in gen_vec_init* calls.
14356 (VEC_INIT_HALF_MODE): New mode iterator.
14357 (vec_init<mode>): Renamed to ...
14358 (vec_init<mode><ssescalarmodelower>): ... this.
14359 (vec_init<mode><ssehalfvecmodelower>): New expander.
14360 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14361 (vec_extractv2sfsf): ... this.
14362 (vec_initv2sf): Renamed to ...
14363 (vec_initv2sfsf): ... this.
14364 (vec_extractv2si): Renamed to ...
14365 (vec_extractv2sisi): ... this.
14366 (vec_initv2si): Renamed to ...
14367 (vec_initv2sisi): ... this.
14368 (vec_extractv4hi): Renamed to ...
14369 (vec_extractv4hihi): ... this.
14370 (vec_initv4hi): Renamed to ...
14371 (vec_initv4hihi): ... this.
14372 (vec_extractv8qi): Renamed to ...
14373 (vec_extractv8qiqi): ... this.
14374 (vec_initv8qi): Renamed to ...
14375 (vec_initv8qiqi): ... this.
14376 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14377 (vec_init<mode>): Renamed to ...
14378 (vec_init<mode><VEC_base_l>): ... this.
14379 (vec_extract<mode>): Renamed to ...
14380 (vec_extract<mode><VEC_base_l>): ... this.
14381 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14382 (vec_initv2sfsf): ... this.
14383 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14384 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14385 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14386 element mode after mode in gen_vec_init* calls.
14387 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14388 (vec_init<mode><Vel>): ... this.
14389 (vec_extract<mode>): Renamed to ...
14390 (vec_extract<mode><Vel>): ... this.
14391 * config/aarch64/iterators.md (Vel): New mode attribute.
14392 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14393 Add element mode after mode in gen_vec_extract* calls.
14394 * config/s390/vector.md (non_vec_l): New mode attribute.
14395 (vec_extract<mode>): Renamed to ...
14396 (vec_extract<mode><non_vec_l>): ... this.
14397 (vec_init<mode>): Renamed to ...
14398 (vec_init<mode><non_vec_l>): ... this.
14399 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14400 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14402 * config/arm/iterators.md (V_elem_l): New mode attribute.
14403 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14404 (vec_extract<mode><V_elem_l>): ... this.
14405 (vec_extractv2di): Renamed to ...
14406 (vec_extractv2didi): ... this.
14407 (vec_init<mode>): Renamed to ...
14408 (vec_init<mode><V_elem_l>): ... this.
14409 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14410 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14411 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14412 Add element mode after gen_vec_extract* calls.
14413 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14414 (vec_init<mode><unitmode>): ... this.
14415 (vec_extract<mode>): Renamed to ...
14416 (vec_extract<mode><unitmode>): ... this.
14417 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14418 (vec_init<mode><unitmode>): ... this.
14419 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14420 (vec_initv2sfsf): ... this.
14421 (vec_extractv2sf): Renamed to ...
14422 (vec_extractv2sfsf): ... this.
14423 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14424 Add element mode after gen_vec_extract* calls.
14425 * config/mips/mips.md (unitmode): New mode iterator.
14426 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14427 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14428 * config/spu/spu.md (inner_l): New mode attribute.
14429 (vec_init<mode>): Renamed to ...
14430 (vec_init<mode><inner_l>): ... this.
14431 (vec_extract<mode>): Renamed to ...
14432 (vec_extract<mode><inner_l>): ... this.
14433 * config/sparc/sparc.md (veltmode): New mode iterator.
14434 (vec_init<VMALL:mode>): Renamed to ...
14435 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14436 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14437 (vec_initv2sisi): ... this.
14438 (vec_initv2sf): Renamed to ...
14439 (vec_initv2sfsf): ... this.
14440 (vec_extractv2sf): Renamed to ...
14441 (vec_extractv2sfsf): ... this.
14442 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14443 (vec_init<mode>): Renamed to ...
14444 (vec_init<mode><VEC_base_l>): ... this.
14445 (vec_extract<mode>): Renamed to ...
14446 (vec_extract<mode><VEC_base_l>): ... this.
14447 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14448 (vec_initv2sfsf): ... this.
14449 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14450 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14451 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14452 gen_vec_init* calls.
14454 2017-08-01 Richard Biener <rguenther@suse.de>
14456 PR tree-optimization/81297
14457 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14458 TREE_OVERFLOW from INTEGER_CSTs.
14460 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14462 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14464 2017-07-31 Carl Love <cel@us.ibm.com>
14466 * config/rs6000/rs6000-c: Add support for built-in functions
14467 vector signed char vec_xl_be (signed long long, signed char *);
14468 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14469 vector signed int vec_xl_be (signed long long, signed int *);
14470 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14471 vector signed long long vec_xl_be (signed long long, signed long long *);
14472 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14473 vector signed short vec_xl_be (signed long long, signed short *);
14474 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14475 vector double vec_xl_be (signed long long, double *);
14476 vector float vec_xl_be (signed long long, float *);
14477 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14478 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14479 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14481 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14482 (altivec_expand_builtin): Add switch statement to call
14483 altivec_expand_xl_be for each builtin.
14484 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14485 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14486 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14487 __builtin_vsx_le_be_v16qi.
14488 * doc/extend.texi: Update the built-in documentation file for the
14489 new built-in functions.
14491 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14494 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14496 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14498 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14500 * config.gcc: Add z14.
14501 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14502 CPU model numbers for z13s and z14.
14503 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14505 * config/s390/s390-opts.h (enum processor_type): Rename
14506 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14507 * config/s390/s390.c (processor_table): Add field for CPU name to
14508 be passed to Binutils.
14509 (s390_asm_output_machine_for_arch): Use the new field in
14510 processor_table for Binutils.
14511 (s390_expand_builtin): Replace arch12 with z14.
14512 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14513 (s390_get_sched_attrmask): Likewise.
14514 (s390_get_unit_mask): Likewise.
14515 * config/s390/s390.opt: Add z14 to processor_type enum.
14517 2017-07-31 Martin Jambor <mjambor@suse.cz>
14520 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14521 regardless of optimization level.
14523 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14524 Martin Liska <mliska@suse.cz>
14526 * predict.def: Remove old comment and adjust probability.
14527 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14528 PREDICT statements.
14530 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14533 * config/i386/i386.c (ix86_function_naked): New function.
14534 (ix86_can_use_return_insn_p): Return false for naked functions.
14535 (ix86_expand_prologue): Skip prologue for naked functions.
14536 (ix86_expand_epilogue): Skip epilogue for naked functions
14537 and emit trap instruction.
14538 (ix86_warn_func_return): New function.
14539 (ix86_attribute_table): Add "naked" attribute specification.
14540 (TARGET_WARN_FUNC_RETURN): Define.
14541 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14543 2017-07-31 Martin Liska <mliska@suse.cz>
14545 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14546 (dump_gimple_bb_header): Always dump BB info.
14547 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14549 2017-07-31 Martin Liska <mliska@suse.cz>
14552 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14553 also with current_function_decl non-null equality.
14555 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14558 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14559 change type to the element type, instead add eltype variable and
14560 use it where we are interested in the element type.
14562 PR tree-optimization/81603
14563 * ipa-polymorphic-call.c
14564 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14565 offset arithmetic in offset_int, bail out if the resulting bit offset
14566 doesn't fit into shwi.
14568 2017-07-31 Martin Liska <mliska@suse.cz>
14570 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14571 (gimplify_save_expr): Fix comment.
14573 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14576 * config/i386/i386.c (ix86_function_arg): Update arguments for
14578 (ix86_compute_frame_layout): Set the initial stack offset to
14579 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14580 INCOMING_FRAME_SP_OFFSET.
14581 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14582 stack before exception handler returns.
14583 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14584 the 'ERROR_CODE' for exception handler.
14586 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14588 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14589 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14590 (ASM_OUTPUT_REG_POP): Ditto.
14591 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14592 instead of asm_fprintf to output pure string.
14594 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14596 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14597 to imported_module_or_decl hook.
14598 (debug_nothing_tree_tree_tree_bool): Remove.
14599 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14600 * debug.c (do_nothing_debug_hooks): Use
14601 debug_nothing_tree_tree_tree_bool_bool instead of
14602 debug_nothing_tree_tree_tree_bool.
14603 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14604 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14605 * sdbout.c (sdb_debug_hooks): Likewise.
14606 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14607 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14609 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14610 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14611 attribute, don't add anything.
14613 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14615 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14616 (fold_build2_stat_loc): Likewise.
14617 (fold_build3_stat_loc): Likewise.
14618 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14619 (fold_build1_loc): Remove macro.
14620 (fold_build2_loc): Likewise.
14621 (fold_build3_loc): Likewise.
14623 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14625 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14626 (gimple_build_debug_bind_source_stat): Likewise.
14627 * gimple.h (gimple_build_debug_bind): Remove macro.
14628 (gimple_build_debug_bind_source): Likewise.
14630 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14632 * bitmap.c (bitmap_alloc): Adjust.
14633 (bitmap_gc_alloc): Likewise.
14634 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14636 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14638 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14639 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14640 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14641 (bitmap_gc_alloc_stat): Likewise.
14642 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14644 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14646 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14647 * rtl.h (shallow_copy_rtx): Remove macro.
14649 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14651 * emit-rtl.c (gen_raw_REG): Adjust.
14652 * gengenrtl.c (gendef): Likewise.
14653 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14654 * rtl.h (rtx_alloc): Remove macro.
14656 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14658 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14659 (build_tree_list_stat): Likewise.
14660 * tree.h (build_tree_list): Remove macro.
14661 (build_tree_list_vec): Likewise.
14663 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14665 * tree.c (make_vector_stat): Remove _stat from name.
14666 (build_vector_stat): Likewise.
14667 * tree.h (make_vector_stat): Remove macro.
14668 (build_vector_stat): Likewise.
14670 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14672 * tree.h (build_var_debug_value): Remove prototype.
14674 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14676 * tree.c (tree_cons_stat): Remove _stat from name.
14677 * tree.h (tree_cons): Remove macro.
14679 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14681 * tree.c (build_vl_exp_stat): Remove _stat from name.
14682 * tree.h (build_vl_exp): Remove macro.
14684 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14686 * tree.c (build_decl_stat): Remove _stat from name.
14687 * tree.h (build_decl): Remove macro.
14689 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14691 * gimple.c (gimple_build_with_ops_stat): Adjust.
14692 (gimple_alloc_stat): Remove _stat from name.
14693 * gimple.h (gimple_alloc): Remove macro.
14695 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14697 * tree.c (make_tree_vec_stat): Remove _stat from name.
14698 (grow_tree_vec_stat): Likewise.
14699 * tree.h (make_tree_vec_stat): Adjust prototype.
14700 (grow_tree_vec_stat): Likewise.
14701 (make_tree_vec): Remove macro.
14702 (grow_tree_vec): Likewise.
14704 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14706 * fold-const.c (fold_build1_stat_loc): Adjust.
14707 (fold_build2_stat_loc): Likewise.
14708 (fold_build3_stat_loc): Likewise.
14709 * tree.c (build0_stat): Remove _stat from name.
14710 (build1_stat): Likewise.
14711 (build2_stat): Likewise.
14712 (build3_stat): Likewise.
14713 (build4_stat): Likewise.
14714 (build5_stat): Likewise.
14715 * tree.h (build1_loc): Remove macro, and rename _stat function
14717 (build2_loc): Likewise.
14718 (build3_loc): Likewise.
14719 (build4_loc): Likewise.
14720 (build5_loc): Likewise.
14722 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14724 * tree.c (make_int_cst_stat): Remove _stat from name.
14725 * tree.h (make_int_cst_stat): Adjust prototype.
14726 (make_int_cst): Remove macro.
14728 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14730 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14731 * tree.h (make_tree_binfo_stat): Adjust prototype.
14732 (make_tree_binfo): Remove.
14734 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14736 * tree.c (copy_node_stat): Rename to copy_node.
14737 (build_distinct_type_copy): Adjust.
14738 * tree.h (copy_node_stat): Adjust prototype.
14739 (copy_node): Remove macro.
14741 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14743 * tree.c (make_node_stat): rename to make_node.
14744 (build_tree_list_stat): Adjust.
14745 (build0_stat): Likewise.
14746 (build2_stat): Likewise.
14747 (build3_stat): Likewise.
14748 (build4_stat): Likewise.
14749 (build5_stat): Likewise.
14750 (build_decl_stat): Likewise.
14751 * tree.h (make_node_stat): Adjust prototype.
14752 (make_node): remove macro.
14754 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14756 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14757 (PPC_FEATURE2_SCV): Likewise.
14758 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14760 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14762 * config/aarch64/aarch64.c
14763 (aarch64_internal_mov_immediate): Add new special pattern.
14764 * config/aarch64/aarch64.md (*movdi_aarch64):
14765 Add reg/32bit const mov case.
14767 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14768 Richard Sandiford <richard.sandiford@linaro.org>
14770 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14771 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14772 Add integer and movi cases.
14773 (movi-split-hf-df-sf split, fp16): New.
14774 (enabled): Added TARGET_FP_F16INST.
14775 * config/aarch64/iterators.md (GPF_HF): New.
14776 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14778 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14780 * config/aarch64/aarch64.c
14781 (aarch64_simd_container_mode): Add prototype.
14782 (aarch64_expand_mov_immediate): Add HI support.
14783 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14784 (aarch64_can_const_movi_rtx_p): New.
14785 (aarch64_preferred_reload_class):
14786 Remove restrictions of using FP registers for certain SIMD operations.
14787 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14788 (aarch64_valid_floating_const): Add integer move validation.
14789 (aarch64_simd_imm_scalar_p): Remove.
14790 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14791 (aarch64_legitimate_constant_p): Expand list of supported cases.
14792 * config/aarch64/aarch64-protos.h
14793 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14794 (aarch64_reinterpret_float_as_int): New.
14795 (aarch64_simd_imm_scalar_p): Remove.
14796 * config/aarch64/constraints.md (Uvi): New.
14797 (Dd): Split into Ds and new Dd.
14798 * config/aarch64/aarch64.md (*movsi_aarch64):
14800 (*movdi_aarch64): Add SIMD mov case.
14802 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14804 * tree-predcom.c: (struct chain): Handle store-store chain in which
14805 stores for elimination only store loop invariant values.
14806 (execute_pred_commoning_chain): Ditto.
14807 (prepare_initializers_chain_store_elim): Ditto.
14808 (prepare_finalizers): Ditto.
14809 (is_inv_store_elimination_chain): New function.
14810 (initialize_root_vars_store_elim_1): New function.
14812 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14814 * tree-predcom.c: Revise general description of the pass.
14815 (enum chain_type): New enum type for store elimination.
14816 (struct chain): New field supporting store elimination.
14817 (struct component): Ditto.
14818 (dump_chain): Dump store-stores chain.
14819 (release_chain): Release resources.
14820 (split_data_refs_to_components): Compute and create component
14821 contains only stores for elimination.
14822 (get_chain_last_ref_at): New function.
14823 (make_invariant_chain): Initialization.
14824 (make_rooted_chain): Specify chain type in parameter and record it.
14825 (add_looparound_copies): Skip for store-stores chain.
14826 (determine_roots_comp): Compute type of chain and pass it to
14828 (initialize_root_vars_store_elim_2): New function.
14829 (finalize_eliminated_stores): New function.
14830 (remove_stmt): Handle store for elimination.
14831 (execute_pred_commoning_chain): Execute predictive commoning on
14832 store-store chains.
14833 (determine_unroll_factor): Skip unroll for store-stores chain.
14834 (prepare_initializers_chain_store_elim): New function.
14835 (prepare_initializers_chain): Hanlde store-store chain.
14836 (prepare_finalizers_chain, prepare_finalizers): New function.
14837 (tree_predictive_commoning_loop): Return integer value indicating
14838 if loop is unrolled or lcssa form is corrupted.
14839 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14841 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14843 * tree-predcom.c (initialize_root): Delete.
14844 (execute_pred_commoning_chain): Initialize root vars and replace
14845 reference of non-combined chain directly, rather than call above
14848 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14850 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14851 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14853 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14855 * tree-predcom.c (struct chain): New field init_seq.
14856 (release_chain): Release init_seq.
14857 (prepare_initializers_chain): Record intialization stmts in above
14859 (insert_init_seqs): New function.
14860 (tree_predictive_commoning_loop): Call insert_init_seqs.
14862 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14864 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14866 2017-07-28 Richard Biener <rguenther@suse.de>
14868 * match.pd: Remove superfluous :c.
14869 * genmatch.c (simplify::id): Add member.
14870 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14872 (current_id): New global.
14873 (dt_node::parent): Move from ...
14874 (dt_operand::parent): ... here. Add for_id member.
14875 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14876 (decision_tree::find_node): Relax order requirement when
14877 merging DT_TRUE nodes to ones inbetween the current simplify
14878 and the one we try to merge with. Add diagnostic whenever
14879 we need to enforce pattern order by not merging.
14880 (decision_tree::insert): Set current_id.
14881 (decision_tree::print_node): Dump parent node and for_id.
14882 (parser::last_id): Add member.
14883 (parser::push_simplify): Assign unique id.
14884 (parser::parser): Initialize last_id.
14886 2017-07-28 Martin Liska <mliska@suse.cz>
14889 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14890 gimple_build_debug_bind.
14892 2017-07-28 Richard Biener <rguenther@suse.de>
14894 PR tree-optimization/81502
14895 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14897 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14899 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14900 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14901 for BIT_FIELD_REF args.
14902 * fold-const.c (make_bit_field_ref): Likewise.
14903 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14905 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14908 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14909 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14910 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14911 Or it into SANITIZER_UNDEFINED.
14912 * ubsan.c: Include gimple-fold.h and varasm.h.
14913 (ubsan_expand_ptr_ifn): New function.
14914 (instrument_pointer_overflow): New function.
14915 (maybe_instrument_pointer_overflow): New function.
14916 (instrument_object_size): Formatting fix.
14917 (pass_ubsan::execute): Call instrument_pointer_overflow
14918 and maybe_instrument_pointer_overflow.
14919 * internal-fn.c (expand_UBSAN_PTR): New function.
14920 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14921 * sanitizer.def (__ubsan_handle_pointer_overflow,
14922 __ubsan_handle_pointer_overflow_abort): New builtins.
14923 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14924 * internal-fn.def (UBSAN_PTR): New internal function.
14925 * opts.c (sanitizer_opts): Add pointer-overflow.
14926 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14927 * fold-const.c (build_range_check): Compute pointer range check in
14928 integral type if pointer arithmetics would be needed. Formatting
14931 2017-07-28 Martin Liska <mliska@suse.cz>
14934 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14935 parameters that are of a variable-length.
14937 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14939 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14941 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14942 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14943 (CRT_CALL_STATIC_FUNCTION): Likewise.
14944 (ASM_DEFAULT_SPEC): New define.
14945 (ASM_SPEC32): Likewise.
14946 (ASM_SPEC64): Likewise.
14947 (ASM_SPEC_COMMON): Likewise.
14948 (ASM_SPEC): Likewise.
14949 (INVALID_64BIT): Likewise.
14950 (LINK_OS_DEFAULT_SPEC): Likewise.
14951 (LINK_OS_SPEC32): Likewise.
14952 (LINK_OS_SPEC64): Likewise.
14953 (POWERPC_LINUX): Likewise.
14954 (PTRDIFF_TYPE): Likewise.
14955 (RESTORE_FP_PREFIX): Likewise.
14956 (RESTORE_FP_SUFFIX): Likewise.
14957 (SAVE_FP_PREFIX): Likewise.
14958 (SAVE_FP_SUFFIX): Likewise.
14959 (SIZE_TYPE): Likewise.
14960 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14961 (TARGET_64BIT): Likewise.
14962 (TARGET_64BIT): Likewise.
14963 (TARGET_AIX): Likewise.
14964 (WCHAR_TYPE_SIZE): Likewise.
14965 (WCHAR_TYPE): Undefine.
14966 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14967 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14968 (CPP_OS_RTEMS_SPEC): Delete.
14969 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14970 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14972 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14974 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14976 PR tree-optimization/81578
14977 * tree-parloops.c (build_new_reduction): Bail out if
14978 reduction_code isn't one of the standard OpenMP reductions.
14979 Move the details printing after that decision.
14981 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14983 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14984 related to reload_in_progress.
14985 (splat_input_operand): Likewise.
14986 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14988 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14990 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14991 (TARGET_INSTANTIATE_DECLS): Likewise.
14992 (legitimate_indexed_address_p): Delete reload_in_progress code.
14993 (rs6000_debug_legitimate_address_p): Likewise.
14994 (rs6000_eliminate_indexed_memrefs): Likewise.
14995 (rs6000_emit_le_vsx_store): Likewise.
14996 (rs6000_emit_move_si_sf_subreg): Likewise.
14997 (rs6000_emit_move): Likewise.
14998 (register_to_reg_type): Likewise.
14999 (rs6000_pre_atomic_barrier): Likewise.
15000 (rs6000_machopic_legitimize_pic_address): Likewise.
15001 (rs6000_allocate_stack_temp): Likewise.
15002 (rs6000_address_for_fpconvert): Likewise.
15003 (rs6000_address_for_altivec): Likewise.
15004 (rs6000_secondary_memory_needed_rtx): Delete function.
15005 (rs6000_check_sdmode): Likewise.
15006 (rs6000_alloc_sdmode_stack_slot): Likewise.
15007 (rs6000_instantiate_decls): Likewise.
15008 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15009 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15010 Delete reload_in_progress.
15011 (*vec_reload_and_plus_<mptrsize>): Likewise.
15012 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15013 (vsx_div_v2di): Likewise.
15014 (vsx_udiv_v2di): Likewise.
15016 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15018 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15019 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15020 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15021 (rs6000_debug_reg_global): Delete print of LRA status.
15022 (rs6000_option_override_internal): Delete dead LRA related code.
15023 (rs6000_lra_p): Delete function.
15024 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15026 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15028 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15029 * config/riscv/rtems.h: New file.
15031 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15032 Sudakshina Das <sudi.das@arm.com>
15034 * config/aarch64/aarch64.md
15035 (define_split for and<mode>3nr_compare): Move
15036 non aarch64_logical_operand to a register.
15037 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15038 register immediate operand to a register.
15039 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15041 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15043 PR middle-end/81564
15044 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15046 2017-07-27 Richard Biener <rguenther@suse.de>
15048 PR tree-optimization/81573
15049 PR tree-optimization/81494
15050 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15051 multi defuse cycle case.
15053 2017-07-27 Richard Biener <rguenther@suse.de>
15055 PR tree-optimization/81571
15056 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15059 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15061 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15062 earlier and only if MASK_FPU is set. Adjust formatting.
15064 2017-07-27 Martin Liska <mliska@suse.cz>
15066 * opt-functions.awk: Add validation of value of Init.
15067 * optc-gen.awk: Pass new argument.
15069 2017-07-27 Martin Liska <mliska@suse.cz>
15071 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15072 Fix wrong condition.
15074 2017-07-27 Martin Liska <mliska@suse.cz>
15076 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15077 BBs and edges seen by autoFDO.
15079 2017-07-27 Richard Biener <rguenther@suse.de>
15081 PR tree-optimization/81502
15082 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15083 with incompatible but same sized type.
15084 (execute_update_addresses_taken): Likewise.
15086 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15088 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15089 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15091 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15094 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15095 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15096 Change s_operand to memory_operand.
15098 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15100 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15101 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15102 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15103 Emit instructions rather than returning an expression. Handle TFmode
15104 and KFmode by casting to TImode.
15105 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15106 (rs6000_emit_le_vsx_store): Likewise.
15107 * config/rs6000/vsx.md (VSX_TI): New iterator.
15108 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15109 (*vsx_le_undo_permute_<mode>): Likewise.
15110 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15111 emit the split sequence.
15112 (*vsx_le_perm_store_<mode>): Likewise.
15114 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15116 PR tree-optimization/81555
15117 PR tree-optimization/81556
15118 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15119 if true, force CHANGED for the recursive invocation.
15120 (reassociate_bb): Remember original length of ops array, pass
15121 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15123 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15124 attributes for noipa attribute. For naked attribute use
15125 lookup_attribute first before lookup_attribute_spec.
15126 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15128 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15129 for functions with noipa attribute.
15130 (cgraph_externally_visible_p): Return true for functions with noipa
15132 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15133 for functions with noipa attribute.
15134 * doc/extend.texi: Document noipa function attribute.
15135 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15136 also for functions with noipa attribute.
15137 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15139 2017-07-26 Andrew Pinski <apinski@cavium.com>
15141 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15142 vec_unalign_load_cost and vec_unalign_store_cost.
15144 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15146 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15147 -mvsx-small-integer option.
15148 (ISA_3_0_MASKS_IEEE): Likewise.
15149 (OTHER_VSX_VECTOR_MASKS): Likewise.
15150 (POWERPC_MASKS): Likewise.
15151 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15152 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15153 code, only testing for DImode being allowed in non-VSX floating
15155 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15156 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15158 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15159 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15160 TARGET_P8_VECTOR test.
15161 (rs6000_secondary_reload_simple_move): Likewise.
15162 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15163 since TARGET_P9_VECTOR was already tested.
15164 (rs6000_opt_masks): Remove -mvsx-small-integer.
15165 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15166 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15168 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15169 test for TARGET_VEXTRACTUB was used, and that uses
15171 (p9 extract splitter): Likewise.
15172 (vsx_extract_<mode>_di_p9): Likewise.
15173 (vsx_extract_<mode>_store_p9): Likewise.
15174 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15175 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15176 the elimination of TARGET_VSX_SMALL_INTEGER.
15177 (vsx_extract_<mode>_p8): Likewise.
15178 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15179 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15180 (vsx_set_<mode>_p9): Likewise.
15181 (vsx_set_v4sf_p9): Likewise.
15182 (vsx_set_v4sf_p9_zero): Likewise.
15183 (vsx_insert_extract_v4sf_p9): Likewise.
15184 (vsx_insert_extract_v4sf_p9_2): Likewise.
15185 * config/rs6000/rs6000.md (sign extend splitter): Change
15186 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15187 (floatsi<mode>2_lfiwax_mem): Likewise.
15188 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15189 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15190 since a test for TARGET_P9_VECTOR was used.
15191 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15192 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15193 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15194 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15195 TARGET_P8_VECTOR test.
15196 (fix_trunc<mode>si2_stfiwx): Likewise.
15197 (fix_trunc<mode>si2_internal): Likewise.
15198 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15199 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15201 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15202 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15203 TARGET_P8_VECTOR test.
15204 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15205 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15206 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15208 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15209 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15210 since a test for TARGET_P9_VECTOR was used.
15211 (splitter for loading small constants): Likewise.
15213 2017-07-26 Andrew Pinski <apinski@cavium.com>
15215 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15218 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15221 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15222 (fp_valid_at): Likewise.
15224 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15226 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15227 (qdf24xx_addrcost_table): Likewise.
15228 (cortexa57_tunings): Update to use generic_branch_cost.
15229 (cortexa72_tunings): Likewise.
15230 (cortexa73_tunings): Likewise.
15231 (qdf24xx_tunings): Likewise.
15233 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15235 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15236 (thunderx2t99_branch_cost): Likewise.
15237 (cortexa35_tunings): Update to use generic_branch_cost.
15238 (cortexa53_tunings): Likewise.
15239 (cortexa57_tunings): Likewise.
15240 (cortexa72_tunings): Likewise.
15241 (cortexa73_tunings): Likewise.
15242 (thunderx2t99_tunings): Likewise.
15244 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15246 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15247 (sparc_option_override): Honour MASK_FSMULD.
15248 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15249 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15250 * config/sparc/sparc.opt (mfsmuld): New option.
15251 * doc/invoke.texi (mfsmuld): Document option.
15253 2017-07-26 Marek Polacek <polacek@redhat.com>
15255 PR middle-end/70992
15256 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15258 2017-07-26 Richard Biener <rguenther@suse.de>
15260 * gimple-match-head.c (do_valueize): Return OP if valueize
15262 (get_def): New helper to get at the def stmt of a SSA name
15263 if valueize allows.
15264 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15265 do_valueize to get at the def stmt.
15266 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15268 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15270 PR middle-end/46932
15271 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15273 2017-07-26 Martin Liska <mliska@suse.cz>
15276 * function.c (expand_function_start): Make expansion of
15277 nonlocal_goto_save_area after parm_birth_insn.
15279 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15281 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15282 from all CPU target flags enable members.
15284 2017-07-26 Richard Biener <rguenther@suse.de>
15286 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15287 (decision_tree::gen): Make 'type' const.
15288 (write_predicate): Likewise.
15290 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15292 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15293 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15294 (rs6000_option_override_internal): Likewise.
15295 (rs6000_expand_vector_set): Likewise.
15296 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15297 (TARGET_UPPER_REGS_SF): Likewise.
15298 (TARGET_UPPER_REGS_DI): Likewise.
15299 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15300 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15301 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15302 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15303 (Splitters for DI constants in Altivec registers): Likewise.
15304 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15305 (vsx_set_v4sf_p9): Likewise.
15306 (vsx_set_v4sf_p9_zero): Likewise.
15307 (vsx_insert_extract_v4sf_p9): Likewise.
15308 (vsx_insert_extract_v4sf_p9_2): Likewise.
15310 2017-07-25 Carl Love <cel@us.ibm.com>
15312 * doc/extend.texi: Update the built-in documentation file for the
15313 existing built-in functions
15314 vector signed char vec_cnttz (vector signed char);
15315 vector unsigned char vec_cnttz (vector unsigned char);
15316 vector signed short vec_cnttz (vector signed short);
15317 vector unsigned short vec_cnttz (vector unsigned short);
15318 vector signed int vec_cnttz (vector signed int);
15319 vector unsigned int vec_cnttz (vector unsigned int);
15320 vector signed long long vec_cnttz (vector signed long long);
15321 vector unsigned long long vec_cnttz (vector unsigned long long);
15323 2017-07-25 Andrew Pinski <apinski@cavium.com>
15325 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15326 accesses where the use is for the first operand of a BIT_INSERT.
15328 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15331 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15332 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15334 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15336 * config/i386/gstabs.h: Delete.
15337 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15339 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15341 * config/i386/i386.c (ix86_decompose_address): Do not check for
15342 register RTX when looking at index_reg or base_reg.
15343 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15345 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15347 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15348 to update EH info here.
15350 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15352 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15354 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15356 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15358 2017-07-25 Torsten Duwe <duwe@suse.de>
15360 * common.opt: Introduce -fpatchable-function-entry
15361 command line option, and its variables function_entry_patch_area_size
15362 and function_entry_patch_area_start.
15363 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15364 including a two-value parser.
15365 * target.def (print_patchable_function_entry): New target hook.
15366 * targhooks.h (default_print_patchable_function_entry): New function.
15367 * targhooks.c (default_print_patchable_function_entry): Likewise.
15368 * toplev.c (process_options): Switch off IPA-RA if
15369 patchable function entries are being generated.
15370 * varasm.c (assemble_start_function): Look at the
15371 patchable-function-entry command line switch and current
15372 function attributes and maybe generate NOP instructions by
15373 calling the print_patchable_function_entry hook.
15374 * doc/extend.texi: Document patchable_function_entry attribute.
15375 * doc/invoke.texi: Document -fpatchable_function_entry
15376 command line option.
15377 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15379 * doc/tm.texi: Re-generate.
15381 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15384 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15385 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15387 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15389 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15391 2017-07-25 Richard Biener <rguenther@suse.de>
15393 PR tree-optimization/81455
15394 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15395 not walk in cycles when looking for guards.
15397 2017-07-25 Richard Biener <rguenther@suse.de>
15399 PR tree-optimization/81529
15400 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15401 when optimizing backedge uses.
15403 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15405 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15407 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15408 to dl_section_ref. On AIX, append an expression to subtract
15409 the size of the section length to dl_section_ref.
15411 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15413 * configure.ac: If any of the config.* scripts fail, exit 1.
15414 * configure: Regenerate.
15416 2017-07-25 Richard Biener <rguenther@suse.de>
15418 PR middle-end/81546
15419 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15420 of immediate uses, be more verbose on errors.
15422 2017-07-25 Richard Biener <rguenther@suse.de>
15424 PR tree-optimization/81510
15425 * tree-vect-loop.c (vect_is_simple_reduction): When the
15426 reduction stmt is not inside the loop bail out.
15428 2017-07-25 Richard Biener <rguenther@suse.de>
15430 PR tree-optimization/81303
15431 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15432 profitability check against LOOP_VINFO_NITERSM1.
15434 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15436 * domwalk.c (cmp_bb_postorder): Simplify.
15437 (sort_bbs_postorder): New function. Use it...
15438 (dom_walker::walk): ...here to optimize common cases.
15440 2017-07-25 Martin Liska <mliska@suse.cz>
15443 * ipa-visibility.c (function_and_variable_visibility): Make the
15444 redirection just on target that supports aliasing.
15445 Fix GNU coding style.
15447 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15450 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15452 * config/arm/rtems.h: Likewise.
15453 * config/bfin/rtems.h: Likewise.
15454 * config/i386/rtemself.h: Likewise.
15455 * config/lm32/rtems.h: Likewise.
15456 * config/m32c/rtems.h: Likewise.
15457 * config/m68k/rtemself.h: Likewise.
15458 * config/microblaze/rtems.h: Likewise.
15459 * config/mips/rtems.h: Likewise.
15460 * config/moxie/rtems.h: Likewise.
15461 * config/nios2/rtems.h: Likewise.
15462 * config/powerpcspe/rtems.h: Likewise.
15463 * config/rs6000/rtems.h: Likewise.
15464 * config/rtems.h: Likewise.
15465 * config/sh/rtems.h: Likewise.
15466 * config/sh/rtemself.h: Likewise.
15467 * config/sparc/rtemself.h: Likewise.
15469 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15472 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15473 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15474 * tree-ssa-structalias.c (alias_get_name): Same.
15476 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15479 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15480 instructions if no du chain is found.
15482 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15484 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15486 2017-07-25 Richard Biener <rguenther@suse.de>
15488 PR middle-end/81505
15489 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15492 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15494 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15495 upper-regs options.
15496 (ISA_2_7_MASKS_SERVER): Likewise.
15497 (ISA_3_0_MASKS_IEEE): Likewise.
15498 (OTHER_P8_VECTOR_MASKS): Likewise.
15499 (OTHER_VSX_VECTOR_MASKS): Likewise.
15500 (POWERPC_MASKS): Likewise.
15501 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15502 duplicate list of options.
15503 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15504 explicit -mupper-regs options.
15505 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15506 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15507 alias for -mupper-regs-df.
15508 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15509 (rs6000_init_hard_regno_mode_ok): Likewise.
15510 (rs6000_option_override_internal): Likewise.
15511 (rs6000_opt_masks): Likewise.
15512 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15513 options in terms of whether -mvsx or -mpower8-vector was used.
15514 (TARGET_UPPER_REGS_DI): Likewise.
15515 (TARGET_UPPER_REGS_SF): Likewise.
15516 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15517 -mupper-regs-* options.
15519 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15521 * passes.c (emergency_dump_function): Print some empty lines and a
15522 header before the RTL dump.
15524 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15526 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15528 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15531 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15532 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15534 2017-07-24 Carl Love <cel@us.ibm.com>
15536 * config/rs6000/rs6000-c.c: Add support for built-in functions
15537 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15538 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15539 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15540 vec_extract_fp_from_shortl): Add defines for the two builtins.
15541 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15542 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15544 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15545 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15546 * doc/extend.texi: Update the built-in documentation file for the
15547 new built-in function.
15549 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15552 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15554 * doc/generic.texi: Likewise.
15555 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15556 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15558 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15560 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15561 (aarch64_mls_elt_merge<mode>): Likewise.
15563 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15565 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15566 having __cxa_atexit.
15568 2017-07-23 Michael Collison <michael.collison@arm.com>
15570 * config/arm/arm.c (arm_option_override): Deprecate
15571 use of -mstructure-size-boundary.
15572 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15573 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15575 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15578 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15579 Reduce cost estimate for direct moves.
15581 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15584 * config/i386/i386.c (ix86_option_override_internal): Disable
15585 BMI, BMI2 and TBM instructions for -m16.
15587 2017-07-21 Carl Love <cel@us.ibm.com>
15589 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15590 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15591 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15592 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15593 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15594 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15595 VMULOSW): New enum "unspec" values.
15596 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15597 altivec_vmulosw): New patterns.
15598 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15599 VMULOSW): Add definitions.
15601 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15603 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15604 (qdf24xx): Likewise.
15605 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15606 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15607 (AARCH64_FL_V8_1): Renumber.
15608 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15609 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15610 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15611 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15612 rdma to feature modifiers list.
15614 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15616 PR middle-end/56727
15617 * ipa-visibility (function_and_variable_visibility): Convert
15618 recursive PLT call to direct call if appropriate.
15620 2017-07-21 Andrew Pinski <apinski@cavium.com>
15622 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15623 operand 1 to see if the types precision matches.
15624 * fold-const.c (operand_equal_p): Likewise.
15626 2017-07-21 Richard Biener <rguenther@suse.de>
15628 PR tree-optimization/81303
15629 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15630 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15631 (vect_peeling_hash_get_lowest_cost): Adjust.
15632 (vect_enhance_data_refs_alignment): Likewise. Use
15633 vect_get_peeling_costs_all_drs to compute the penalty for no
15634 peeling to match up costs.
15636 2017-07-21 Richard Biener <rguenther@suse.de>
15638 PR tree-optimization/81500
15639 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15640 we didn't identify a reduction path.
15642 2017-07-21 Tom de Vries <tom@codesourcery.com>
15643 Cesar Philippidis <cesar@codesourcery.com>
15645 PR gcov-profile/81442
15646 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15649 2017-07-21 Tom de Vries <tom@codesourcery.com>
15652 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15654 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15655 nvptx_override_options_after_change.
15657 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15659 * dwarf2out.c (output_file_names): Avoid double testing for
15660 dwarf_version >= 5.
15662 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15664 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15666 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15668 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15670 (try_crossjump_to_edge): Do not punt on partitioned functions.
15672 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15674 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15675 Put all BBs reachable only via paths crossing cold region to cold
15677 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15679 2016-07-21 Richard Biener <rguenther@suse.de>
15681 PR tree-optimization/81303
15682 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15683 into account prologue and epilogue iterations when raising
15684 min_profitable_iters to sth at least covering one vector iteration.
15686 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15688 * config/arm/arm.c (arm_test_cpu_arch_dat):
15691 2017-07-20 Nathan Sidwell <nathan@acm.org>
15693 Remove TYPE_METHODS.
15694 * tree.h (TYPE_METHODS): Delete.
15695 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15696 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15697 (dbxout_type_methods): Scan TYPE_FIELDS.
15698 (dbxout_type): Don't check TYPE_METHODS here.
15699 * function.c (use_register_for_decl): Always ignore register for
15700 class types when not optimizing.
15701 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15702 * tree.c (free_lang_data_in_type): Stitch out member functions and
15703 templates from TYPE_FIELDS.
15704 (build_distinct_type_copy, verify_type_variant,
15705 verify_type): Member fns are on TYPE_FIELDS.
15706 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15707 * tree-pretty-print.c (dump_generic_node): Likewise.
15709 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15712 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15713 V2TImode and V4TImode.
15714 (ix86_expand_vector_extract): Likewise.
15715 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15716 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15717 (ssescalarmode): Handle V4TImode and V2TImode.
15718 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15719 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15720 (VEXTRACTI128_MODE): New mode iterator.
15721 (splitter for *vec_extractv?ti first element): New.
15722 (VEC_INIT_MODE): New mode iterator.
15723 (vec_init<mode>): Consolidate 3 expanders into one using
15724 VEC_INIT_MODE mode iterator.
15726 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15728 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15729 non_spilled_static_chain_regno_p.
15731 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15733 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15735 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15737 * bb-reorder.c (connect_traces): Allow copying of blocks within
15740 2017-07-20 Richard Biener <rguenther@suse.de>
15742 * gimple.h (gimple_phi_result): Add gphi * overload.
15743 (gimple_phi_result_ptr): Likewise.
15744 (gimple_phi_arg): Likewise. Adjust index assert to only
15745 allow actual argument accesses rather than all slots available
15747 (gimple_phi_arg_def): Add gphi * overload.
15748 * tree-phinodes.c (make_phi_node): Initialize only actual
15750 (resize_phi_node): Clear memory not covered by old node,
15751 do not initialize excess argument slots.
15752 (reserve_phi_args_for_new_edge): Initialize new argument slot
15755 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15757 PR tree-optimization/81388
15759 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15761 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15762 by removing computation of may_be_zero.
15764 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15765 Tom de Vries <tom@codesourcery.com>
15767 PR middle-end/81030
15768 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15769 when gimple level profile disagrees with what RTL expander did.
15771 2017-07-20 Richard Biener <rguenther@suse.de>
15773 PR tree-optimization/61171
15774 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15775 (vect_analyze_stmt): Add slp instance parameter.
15776 (vectorizable_reduction): Likewise.
15777 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15778 (vect_is_simple_reduction): Deal with chains not detected
15779 as SLP reduction chain, specifically not properly associated
15780 chains containing a mix of plus/minus.
15781 (get_reduction_op): Remove.
15782 (get_initial_defs_for_reduction): Simplify, pass in whether
15783 this is a reduction chain, pass in the SLP node for the PHIs.
15784 (vect_create_epilog_for_reduction): Get the SLP instance as
15786 (vectorizable_reduction): Get the SLP instance as arg.
15787 During analysis remember the SLP node with the PHIs in the
15788 instance. Simplify getting at the vectorized reduction PHIs.
15789 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15790 through SLP instance.
15791 (vect_slp_analyze_operations): Likewise.
15792 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15793 (vect_transform_stmt): Likewise.
15795 2017-07-20 Tom de Vries <tom@codesourcery.com>
15797 PR tree-optimization/81489
15798 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15799 read of phi arg location to before loop that modifies phi.
15801 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15803 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15806 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15808 PR middle-end/81331
15809 * except.c (execute): Fix ordering issue.
15811 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15813 PR rtl-optimization/81423
15814 * combine.c (make_compound_operation_int): Don't try to optimize
15815 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15817 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15819 PR rtl-optimization/81423
15820 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15821 with a constant that is -1 in the truncated to mode.
15823 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15825 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15826 (determine_unlikely_bbs): ... here.
15827 * predict.h (propagate_unlikely_bbs_forward): Declare.
15828 * cfgexpand.c (pass_expand::execute): Use it.
15829 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15831 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15832 propagate_unlikely_bbs_forward.
15834 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15836 PR middle-end/81331
15837 * except.c (maybe_add_nop_after_section_switch): New function.
15840 2017-07-19 Tom de Vries <tom@codesourcery.com>
15842 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15844 2017-07-19 Tom de Vries <tom@codesourcery.com>
15846 * gimple.h (gimple_phi_arg): Make assert more strict.
15848 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15850 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15851 * config/rs6000/mmintrin.h: New file.
15852 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15854 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15856 PR tree-optimization/81346
15857 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15859 2017-07-19 Tom de Vries <tom@codesourcery.com>
15861 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15863 2017-07-19 Tom de Vries <tom@codesourcery.com>
15865 * config/nvptx/nvptx-modes.def: Add V2DImode.
15866 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15867 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15868 (nvptx_output_mov_insn): Handle lack of mov.b128.
15869 (nvptx_print_operand): Handle 'H' and 'L' codes.
15870 (nvptx_vector_mode_supported): Allow V2DImode.
15871 (nvptx_preferred_simd_mode): New function.
15872 (nvptx_data_alignment): New function.
15873 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15874 nvptx_preferred_simd_mode.
15875 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15877 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15879 2017-07-19 Tom de Vries <tom@codesourcery.com>
15881 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15882 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15883 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15884 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15885 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15886 (mov<VECIM>_insn): New define_insn.
15887 (define_expand "mov<VECIM>): New define_expand.
15889 2017-07-19 Tom de Vries <tom@codesourcery.com>
15891 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15893 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15895 PR tree-optimization/81346
15896 * fold-const.h (fold_div_compare, range_check_type): Declare.
15897 * fold-const.c (range_check_type): New function.
15898 (build_range_check): Use range_check_type.
15899 (fold_div_compare): No longer static, rewritten into
15900 a match.pd helper function.
15901 (fold_comparison): Don't call fold_div_compare here.
15902 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15903 as helper function.
15905 2017-07-19 Nathan Sidwell <nathan@acm.org>
15907 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15908 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15909 * tree.c (find_decls_types_r, verify_type): Use
15910 TYPE_{MIN,MAX}_VALUE_RAW.
15911 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15912 (hash_tree): Likewise.
15913 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15915 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15918 2017-07-18 Tom de Vries <tom@codesourcery.com>
15920 PR middle-end/81464
15921 * omp-expand.c (expand_omp_for_static_chunk): Handle
15922 equal-argument loop exit phi.
15924 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15927 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15928 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15929 operand 2 predicate.
15930 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15931 operand 2 predicate.
15932 (ror,rol -> rorx splitters): Use const_int_operand as
15933 operand 2 predicate.
15935 2017-06-18 Richard Biener <rguenther@suse.de>
15937 PR tree-optimization/81410
15938 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15939 the gap in the ! slp_perm SLP case after each group.
15941 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15943 PR middle-end/81463
15944 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15947 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15949 PR middle-end/81462
15950 * predict.c (set_even_probabilities): Cleanup; do not affect
15951 probabilities that are already known.
15952 (combine_predictions_for_bb): Call even when count is set.
15954 2017-07-18 Nathan Sidwell <nathan@acm.org>
15956 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15959 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15962 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15963 optimization for loop niter analysis.
15965 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15968 * config/avr/avr.c (avr_optimize_casesi): Don't use
15969 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15971 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15973 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15974 body_cost_vec from _vect_peel_extended_info.
15975 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15976 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15979 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15981 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15983 2017-07-18 Richard Biener <rguenther@suse.de>
15985 PR tree-optimization/80620
15986 PR tree-optimization/81403
15987 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15988 info when re-using a VN table entry.
15990 2017-07-18 Richard Biener <rguenther@suse.de>
15992 PR tree-optimization/81418
15993 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15994 vectype_in. Verify that with lane-reducing reduction operations
15995 we have a single def-use cycle.
15997 2017-07-17 Carl Love <cel@us.ibm.com>
15999 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16001 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16002 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16003 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16004 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16005 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16006 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16007 VMULOSW): New enum "unspec" values.
16008 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16009 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16010 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16011 altivec_vmulosw): New patterns.
16012 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16013 VMULOSW): Add definitions.
16014 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16016 * config/alpha/alpha.c: Include predict.h.
16018 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16020 * tree-vrp.c (compare_assert_loc): Fix comparison function
16021 to return predictable results.
16023 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16025 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16027 (subdi3): Likewise.
16028 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16029 * doc/invoke.texi (mexpand-adddi): Update text.
16031 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16033 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16034 that also clobbers the CC register. The old expand code is moved
16036 (*arc_clzsi2): ... here.
16037 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16038 the CC register. The old expand code is moved to ...
16039 (arc_ctzsi2): ... here.
16041 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16043 * config/arc/arc.opt (mindexed-loads): Use initial value
16044 TARGET_INDEXED_LOADS_DEFAULT.
16045 (mauto-modify-reg): Use initial value
16046 TARGET_AUTO_MODIFY_REG_DEFAULT.
16047 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16048 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16049 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16050 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16052 2017-07-17 Martin Liska <mliska@suse.cz>
16055 * opts.c (finish_options): Do not allow -fgnu-tm
16056 w/ -fsanitize={kernel-,}address. Say sorry.
16058 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16061 * tree-loop-distribution.c (classify_partition): Only assert on
16062 numer of iterations.
16063 (merge_dep_scc_partitions): Delete prameter. Update function call.
16064 (distribute_loop): Remove code handling loop with unknown niters.
16065 (pass_loop_distribution::execute): Skip loop with unknown niters.
16067 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16070 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16071 function sort_partitions_by_post_order.
16073 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16075 PR tree-optimization/81374
16076 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16077 the max index of basic blocks, rather than number of basic blocks.
16079 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16081 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16083 (arc_legitimate_pic_operand_p): Likewise.
16084 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16086 (arc_needs_pcl_p): Likewise.
16087 (arc_legitimate_pc_offset_p): Likewise.
16088 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16089 function is also used in constrains.md.
16090 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16091 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16092 PLUS. Only return true/false in known cases, otherwise assert.
16093 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16094 is already called in arc_legitimate_constant_p.
16095 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16097 (LEGITIMATE_PIC_OPERAND_P): Use
16098 arc_raw_symbolic_reference_mentioned_p function.
16099 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16104 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16105 Andrew Burgess <andrew.burgess@embecosm.com>
16107 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16108 (arc_return_address_register): New function.
16109 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16110 (arc_handle_fndecl_attribute): Add naked attribute.
16111 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16112 (TARGET_WARN_FUNC_RETURN): Likewise.
16113 (arc_allocate_stack_slots_for_args): New function.
16114 (arc_warn_func_return): Likewise.
16115 (machine_function): Change type fn_type.
16116 (arc_compute_function_type): Consider new naked function type,
16117 change function return type.
16118 (arc_must_save_register): Adapt to handle new
16119 arc_compute_function_type's return type.
16120 (arc_expand_prologue): Likewise.
16121 (arc_expand_epilogue): Likewise.
16122 (arc_return_address_regs): Delete.
16123 (arc_return_address_register): New function.
16124 (arc_epilogue_uses): Use above function.
16125 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16126 (arc_function_type): Change encoding, add naked type.
16127 (ARC_INTERRUPT_P): Change to handle the new encoding.
16128 (ARC_FAST_INTERRUPT_P): Likewise.
16129 (ARC_NORMAL_P): Define.
16130 (ARC_NAKED_P): Likewise.
16131 (arc_compute_function_type): Delete prototype.
16132 * config/arc/arc.md (in_ret_delay_slot): Use
16133 arc_return_address_register function.
16134 (simple_return): Likewise.
16135 (p_return_i): Likewise.
16137 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16139 PR tree-optimization/81428
16140 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16141 can't be built for those types.
16143 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16145 Remove stuff dead since r239246.
16147 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16148 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16149 (avr_inform_devices): Remove dead stuff.
16151 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16153 * config/arm/arm_neon.h: Fix softp typo.
16155 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16157 PR tree-optimization/81365
16158 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16159 aggregate moves onto bb predecessor edges, make sure there are no
16160 loads that could alias the lhs in between the start of bb and the
16163 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16166 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16167 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16168 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16170 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16172 PR tree-optimization/81396
16173 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16174 (init_symbolic_number): Initialize it to 1.
16175 (perform_symbolic_merge): Add n_ops from both operands into the new
16177 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16178 without base_addr as useless if they need more than one operation.
16179 (bswap_replace): Handle !bswap case for NULL base_addr.
16181 2017-07-17 Tom de Vries <tom@codesourcery.com>
16184 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16187 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16189 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16190 conditional builtin define __FIX_LEON3FT_B2BST.
16192 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16194 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16195 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16198 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16200 PR rtl-optimization/81424
16201 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16202 to remove potential trapping from operands if -fnon-call-exceptions.
16204 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16206 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16207 profile_proability for scalling.
16208 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16210 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16212 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16214 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16216 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16217 fixpoint arithmetics.
16219 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16221 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16222 fixpoint arithmetics.
16224 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16226 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16227 fixpoint arithmetics.
16229 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16231 * profile-count.h (profile_probability::from_reg_br_prob_note,
16232 profile_probability::to_reg_br_prob_note): New functions.
16233 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16234 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16235 * predict.c (probability_reliable_p): Update.
16236 (edge_probability_reliable_p): Update.
16237 (br_prob_note_reliable_p): Update.
16238 (invert_br_probabilities): Update.
16239 (add_reg_br_prob_note): New function.
16240 (combine_predictions_for_insn): Update.
16241 * asan.c (asan_clear_shadow): Update.
16242 * cfgbuild.c (compute_outgoing_frequencies): Update.
16243 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16244 (update_br_prob_note): Update.
16245 (rtl_verify_edges): Update.
16246 (purge_dead_edges): Update.
16247 (fixup_reorder_chain): Update.
16248 * emit-rtl.c (try_split): Update.
16249 * ifcvt.c (cond_exec_process_insns): Update.
16250 (cond_exec_process_if_block): Update.
16251 (dead_or_predicable): Update.
16252 * internal-fn.c (expand_addsub_overflow): Update.
16253 (expand_neg_overflow): Update.
16254 (expand_mul_overflow): Update.
16255 * loop-doloop.c (doloop_modify): Update.
16256 * loop-unroll.c (compare_and_jump_seq): Update.
16257 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16258 * predict.h: Update.
16259 * reorg.c (mostly_true_jump): Update.
16261 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16262 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16263 * config/arc/arc.c: (emit_unlikely_jump): Update.
16264 * config/arm/arm.c: (emit_unlikely_jump): Update.
16265 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16266 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16267 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16268 (ix86_print_operand): Update.
16269 (ix86_split_fp_branch): Update.
16270 (predict_jump): Update.
16271 * config/ia64/ia64.c (ia64_print_operand): Update.
16272 * config/mmix/mmix.c (mmix_print_operand): Update.
16273 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16274 (rs6000_expand_split_stack_prologue): Update.
16275 * config/rs6000/rs6000.c: Update.
16276 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16277 (s390_expand_vec_movstr): Update.
16278 (s390_expand_cs_tdsi): Update.
16279 (s390_expand_split_stack_prologue): Update.
16280 * config/sh/sh.c (sh_print_operand): Update.
16281 (expand_cbranchsi4): Update.
16282 (expand_cbranchdi4): Update.
16283 * config/sparc/sparc.c (output_v9branch): Update.
16284 * config/spu/spu.c (get_branch_target): Update.
16285 (ea_load_store_inline): Update.
16286 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16287 * config/tilepro/tilepro.c: Update.
16289 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16291 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16292 (gimplify_save_expr): Likewise.
16294 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16296 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16298 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16300 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16302 * timevar.def (TV_IPA_FNSUMMARY): Define.
16304 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16306 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16307 to back store errata sensitive sequence from being generated.
16308 (sqrtdf2_fix): Likewise.
16310 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16312 * tree-ssa-threadupdate.c (compute_path_counts,
16313 update_joiner_offpath_counts): Use profile_probability.
16315 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16318 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16320 * config/arm/arm-c.c (arm_cpu_builtins): Define
16321 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16323 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16325 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16326 array entries to represent __ieee128 versions of the
16327 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16328 scalar_extract_sig, and scalar_insert_exp built-in functions.
16329 (altivec_resolve_overloaded_builtin): Add special case handling
16330 for the __builtin_scalar_insert_exp function, as represented by
16331 the P9V_BUILTIN_VEC_VSIEDP constant.
16332 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16333 exponent support for __ieee128 argument.
16334 (VSESQP): Add scalar extract signature support for __ieee128
16336 (VSTDCNQP): Add scalar test negative support for __ieee128
16338 (VSIEQP): Add scalar insert exponent support for __int128 argument
16339 with __ieee128 result.
16340 (VSIEQPF): Add scalar insert exponent support for __ieee128
16341 argument with __ieee128 result.
16342 (VSTDCQP): Add scalar test data class support for __ieee128
16344 (VSTDCNQP): Add overload support for scalar test negative with
16345 __ieee128 argument.
16346 (VSTDCQP): Add overload support for scalar test data class
16347 __ieee128 argument.
16348 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16349 UNSPEC_VSX_SXSIGDP.
16350 (UNSPEC_VSX_SIEXPQP): New constant.
16351 (xsxexpqp): New insn for VSX scalar extract exponent quad
16353 (xsxsigqp): New insn for VSX scalar extract significand quad
16355 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16356 precision with floating point argument.
16357 (xststdcqp): New expand for VSX scalar test data class quad
16359 (xststdcnegqp): New expand for VSX scalar test negative quad
16361 (xststdcqp): New insn to match expansions for VSX scalar test data
16362 class quad precision and VSX scalar test negative quad precision.
16363 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16364 special case operand checking to enforce that second operand of
16365 VSX scalar test data class with quad precision argument is a 7-bit
16367 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16368 prototypes and descriptions of __ieee128 versions of
16369 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16370 scalar_test_data_class, and scalar_test_neg built-in functions.
16372 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16374 PR tree-optimization/81162
16375 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16376 replace a negate with an add.
16378 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16380 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16382 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16384 * config/arm/arm-c.c (arm_cpu_builtins): Define
16385 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16387 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16389 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16390 (armv8-r): Set ARM Cortex-R52 as default CPU.
16391 * config/arm/arm-tables.opt: Regenerate.
16392 * config/arm/arm-tune.md: Regenerate.
16393 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16395 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16396 extension for -mcpu=cortex-r52.
16398 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16400 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16401 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16402 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16403 (fp-armv8): Define it as FP_ARMv8 only.
16404 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16405 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16407 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16408 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16409 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16410 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16411 than TARGET_FPU_ARMV8.
16412 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16413 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16414 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16416 * config/arm/neon.md (neon_vrint): Likewise.
16417 (neon_vcvt): Likewise.
16418 (neon_<fmaxmin_op><mode>): Likewise.
16419 (<fmaxmin><mode>3): Likewise.
16420 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16421 * config/arm/predicates.md (arm_cond_move_operator): Check against
16422 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16424 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16426 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16427 to top of function.
16429 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16431 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16432 loop in comment with memset.
16434 2017-07-14 Martin Liska <mliska@suse.cz>
16436 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16437 * dwarf2out.c (is_java): Remove the function.
16438 (output_pubname): Remove usage of the function.
16439 (lower_bound_default): Remove usage of DW_LANG_Java.
16440 (gen_compile_unit_die): Likewise.
16441 * gcc.c: Remove compiler defaults for .java and .zip files.
16442 * gimple-expr.c (remove_suffix): Change as there's no longer
16443 extension than 4-letter one.
16444 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16445 (gimplify_save_expr): Likewise.
16446 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16447 as it's possible even for other languages than Java.
16448 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16449 * lto-opts.c (lto_write_options): Remove reference to Java.
16450 * opts.c (strip_off_ending): Update file extension handling.
16451 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16452 * tree-eh.c (lower_resx): Likewise.
16453 * tree.c (free_lang_data_in_type): Remove dead code.
16454 (find_decls_types_r): Likewise.
16455 (build_common_builtin_nodes): Remove Java from a comment.
16456 (verify_type): Remove dead code.
16457 * varasm.c (assemble_external): Remove Java from a comment.
16459 2017-07-14 Martin Liska <mliska@suse.cz>
16461 * opts.c (finish_options): Add quotes.
16462 (common_handle_option): Likewise.
16464 2017-07-14 Martin Liska <mliska@suse.cz>
16466 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16467 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16468 Remove N_SO_PASCAL.
16469 * dwarf2out.c (lower_bound_default): Do not handle
16471 (gen_compile_unit_die): Likewise.
16472 * gcc.c: Remove default extension binding for GNU Pascal.
16473 * stmt.c: Remove Pascal language from a comment.
16474 * xcoffout.c: Likewise.
16476 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16479 * diagnostic-show-locus.c (fixit_cmp): New function.
16480 (layout::layout): Sort m_fixit_hints.
16481 (column_range::column_range): Assert that the values are valid.
16482 (struct char_span): New struct.
16483 (correction::overwrite): New method.
16484 (struct source_line): New struct.
16485 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16486 calls in terms of classes source_line and char_span, and
16487 correction::overwrite.
16488 (selftest::test_overlapped_fixit_printing_2): New function.
16489 (selftest::diagnostic_show_locus_c_tests): Call it.
16491 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16493 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16494 early if there is no lhs.
16496 2017-07-13 Martin Liska <mliska@suse.cz>
16498 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16499 (gen_reference_type_die): Likewise.
16500 * stor-layout.c: Remove Pascal-related comment.
16502 2017-07-13 Martin Liska <mliska@suse.cz>
16504 * opts.c (finish_options): Add quotes to error messages.
16505 (parse_sanitizer_options): Likewise.
16507 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16509 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16511 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16513 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16515 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16517 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16519 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16521 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16524 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16525 provides the hardware capability bits, define the macro
16526 __BUILTIN_CPU_SUPPORTS__.
16527 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16528 if GLIBC does not provide the hardware capability bits. Add a
16529 gcc_unreachable call if the built-in cpu function is neither
16530 __builtin_cpu_is nor __builtin_cpu_supports.
16531 (rs6000_get_function_versions_dispatcher): Change the warning
16532 that an old GLIBC is used which does not export the capability
16533 bits to be an error.
16534 * doc/extend.texi (target_clones attribute): Document the
16535 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16536 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16537 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16538 the macros defined by GCC if the newer GLIBC is available.
16540 2017-07-12 Jeff Law <law@redhat.com>
16542 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16543 remaining includes slightly.
16544 * config/riscv/riscv-builtins.c: Include profile-count.h.
16546 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16549 * config/avr/avr.c (avr_set_current_function): In diagnostic
16550 messages: Quote keywords and (parts of) identifiers.
16551 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16554 2017-07-12 Carl Love <cel@us.ibm.com>
16556 * config/rs6000/rs6000-c.c: Add support for built-in functions
16557 vector bool char vec_revb (vector bool char);
16558 vector bool short vec_revb (vector short char);
16559 vector bool int vec_revb (vector bool int);
16560 vector bool long long vec_revb (vector bool long long);
16561 * doc/extend.texi: Update the built-in documentation file for the
16562 new built-in functions.
16564 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16566 * config/s390/s390.md: Remove movcc splitter.
16568 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16570 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16571 load/store on condition.
16573 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16576 * config/avr/avr.c (avr_encode_section_info)
16577 [progmem && !TREE_READONLY]: Error if progmem object needs
16580 2017-07-11 Michael Collison <michael.collison@arm.com>
16582 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16585 2017-07-11 Carl Love <cel@us.ibm.com>
16587 * config/rs6000/rs6000-c.c: Add support for builtins
16588 vector unsigned int vec_parity_lsbb (vector signed int);
16589 vector unsigned int vec_parity_lsbb (vector unsigned int);
16590 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16591 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16592 vector unsigned long long vec_parity_lsbb (vector signed long long);
16593 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16594 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16595 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16596 * doc/extend.texi: Update the built-in documentation file for the
16597 new built-in functions.
16599 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16601 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16602 (layout::m_primary_loc): New field.
16603 (layout::layout): Initialize new field. Move location filtering
16604 logic from here to...
16605 (layout::maybe_add_location_range): ...this new method. Add
16606 support for filtering to just the lines already specified by other
16608 (layout::will_show_line_p): New method.
16609 (gcc_rich_location::add_location_if_nearby): New method.
16610 (selftest::test_add_location_if_nearby): New test function.
16611 (selftest::diagnostic_show_locus_c_tests): Call it.
16612 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16615 2017-07-11 Tom de Vries <tom@codesourcery.com>
16617 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16618 (bb_first_real_insn): New function.
16619 (nvptx_single): Add extra initialization of broadcasted condition
16622 2017-07-11 Nathan Sidwell <nathan@acm.org>
16624 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16626 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16628 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16629 Binutils doc as TEXI will mess them up.
16630 * doc/invoke.texi (AVR Options): Same here.
16632 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16634 * config/sparc/sparc.opt (mfix-ut700): New option.
16635 (mfix-gr712rc): Likewise.
16636 (sparc_fix_b2bst): New variable.
16637 * doc/invoke.texi (SPARC options): Document them.
16638 (ARM options): Fix warnings.
16639 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16640 instructions to prevent sequences that can trigger the store-store
16641 errata for certain LEON3FT processors.
16642 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16643 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16644 * config/sparc/sparc.md (fix_b2bst): New attribute.
16645 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16647 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16650 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16652 (*rsqrtsf2_sse): Ditto.
16654 (div<mode>3): Macroize insn from divdf3 and divsf3
16655 using MODEF mode iterator.
16657 2017-07-10 Martin Sebor <msebor@redhat.com>
16659 PR tree-optimization/80397
16660 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16661 instead of testing for equality to INTEGER_TYPE.
16663 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16665 * config.gcc: Remove uclibc from arc target spec.
16667 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16669 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16671 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16674 * lto-wrapper.c (remove_option): New function.
16675 (merge_and_complain): Merge PIC/PIE options more realistically.
16677 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16679 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16683 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16684 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16685 * config.in: Regenerate.
16686 * configure: Regenerate.
16687 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16688 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16689 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16690 (TARGET_GASISR_PROLOGUES): ...target mask.
16691 * common/config/avr/avr-common.c
16692 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16693 Set -mgas-isr-prologues.
16694 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16695 INSERT_PASS_BEFORE for it.
16696 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16697 * config/avr/avr.c (avr_option_override)
16698 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16699 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16700 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16701 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16702 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16704 (make_avr_pass_pre_proep): New function.
16705 (emit_push_sfr) <treg>: Add argument to function and use it
16706 instead of TMP_REG.
16707 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16708 and set machine->gasisr.yes.
16709 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16710 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16711 __gcc_isr.n_pushed to .L__stack_usage.
16712 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16713 (avr_asm_final_postscan_insn): ...this new static function.
16714 * config/avr/avr.h (machine_function)
16715 <is_no_gccisr, use_L__stack_usage>: New fields.
16716 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16717 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16718 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16719 (gasisr, *gasisr): New expander and insn.
16720 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16721 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16722 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16724 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16726 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16729 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16731 Move jump-tables out of .text again.
16734 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16735 (ASM_OUTPUT_ADDR_VEC): New function.
16736 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16737 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16738 INSN_ADDRESSes as asm comment.
16739 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16740 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16741 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16742 * config/avr/avr.md (*tablejump): Adjust comment.
16743 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16744 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16746 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16747 (avr_output_addr_vec): New proto.
16748 (avr_log_t) <insn_addresses>: New field.
16750 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16753 * config/i386/i386.c (ix86_function_arg_advance): Set
16754 outgoing_args_on_stack to true if there are outgoing arguments
16756 (ix86_function_arg): Likewise.
16757 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16758 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16759 * config/i386/i386.h (machine_function): Add
16760 outgoing_args_on_stack.
16762 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16764 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16765 supporting pthreds.
16766 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16768 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16770 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16771 (REAL_H): Remove $(MACHMODE_H).
16772 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16774 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16775 $(MACHMODE_H) and double-int.h.
16776 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16778 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16781 2017-07-07 Andrew Pinski <apinski@cavium.com>
16783 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16784 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16786 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16788 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16789 Add warning if GCC was not configured to link against a GLIBC that
16790 exports the hardware capability bits.
16791 (make_resolver_func): Make resolver function private and not a
16792 COMDAT function. Create the name with clone_function_name instead
16793 of make_unique_name.
16796 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16797 correct operand in doing the split.
16799 2017-07-07 Carl Love <cel@us.ibm.com>
16801 * config/rs6000/rs6000-c: Add support for built-in function
16802 vector unsigned short vec_pack_to_short_fp32 (vector float,
16804 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16805 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16806 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16807 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16808 (convert_4f32_8i16): Add define_expand.
16809 * doc/extend.texi: Update the built-in documentation file for the
16810 new built-in function.
16812 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16814 * config/sparc/m8.md: New file.
16815 * config/sparc/sparc.md: Include m8.md.
16817 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16819 * config/sparc/sparc.opt: New option -mvis4b.
16820 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16821 (sparc_option_override): Handle VIS4B.
16822 (enum sparc_builtins): Define
16823 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16824 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16825 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16826 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16827 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16828 (check_constant_argument): New function.
16829 (sparc_vis_init_builtins): Define builtins
16830 __builtin_vis_dictunpack{8,16,32},
16831 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16832 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16833 __builtin_vis_fpcmpde{8,16,32}shl and
16834 __builtin_vis_fpcmpur{8,16,32}shl.
16835 (sparc_expand_builtin): Check that the constant operands to
16836 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16837 constant and in range.
16838 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16840 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16841 (SPARC_IMM5_P): Likewise.
16842 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16843 (enabled): Handle vis4b.
16844 (UNSPEC_DICTUNPACK): New unspec.
16845 (UNSPEC_FPCMPSHL): Likewise.
16846 (UNSPEC_FPUCMPSHL): Likewise.
16847 (UNSPEC_FPCMPDESHL): Likewise.
16848 (UNSPEC_FPCMPURSHL): Likewise.
16849 (cpu_feature): New CPU feature `vis4b'.
16850 (dictunpack{8,16,32}): New insns.
16851 (FPCSMODE): New mode iterator.
16852 (fpcscond): New code iterator.
16853 (fpcsucond): Likewise.
16854 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16855 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16856 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16857 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16858 * config/sparc/constraints.md: Define constraints `q' for unsigned
16859 2-bit integer constants and `t' for unsigned 5-bit integer
16861 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16863 (imm5_operand_dictunpack16): Likewise.
16864 (imm5_operand_dictunpack32): Likewise.
16865 (imm2_operand): Likewise.
16866 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16867 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16868 ditunpack* and fpcmp*shl builtins.
16870 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16872 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16873 * config.in: Add HAVE_AS_SPARC6 define.
16874 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16876 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16878 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16879 (CPP_CPU_SPEC): Handle m8.
16880 (ASM_CPU_SPEC): Likewise.
16881 * config/sparc/sparc-opts.h (enum processor_type): Add
16883 * config/sparc/sparc.c (m8_costs): New struct.
16884 (sparc_option_override): Handle TARGET_CPU_m8.
16885 (sparc32_initialize_trampoline): Likewise.
16886 (sparc64_initialize_trampoline): Likewise.
16887 (sparc_issue_rate): Likewise.
16888 (sparc_register_move_cost): Likewise.
16889 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16890 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16891 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16892 (CPP_CPU_SPEC): Handle M8.
16893 (ASM_CPU_SPEC): Likewise.
16894 (AS_M8_FLAG): Define.
16895 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16896 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16897 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16899 * configure: Regenerate.
16900 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16903 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16905 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16907 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16908 ("*movdi_insn_sp32"): Do not set v3pipe.
16909 ("*movsi_insn"): Likewise.
16910 ("*movdi_insn_sp64"): Likewise.
16911 ("*movsf_insn"): Likewise.
16912 ("*movdf_insn_sp32"): Likewise.
16913 ("*movdf_insn_sp64"): Likewise.
16914 ("*zero_extendsidi2_insn_sp64"): Likewise.
16915 ("*sign_extendsidi2_insn"): Likewise.
16916 ("*mov<VM32:mode>_insn"): Likewise.
16917 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16918 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16919 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16920 ("<vlop:code><VL:mode>3"): Likewise.
16921 ("*not_<vlop:code><VL:mode>3"): Likewise.
16922 ("*nand<VL:mode>_vis"): Likewise.
16923 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16924 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16925 ("one_cmpl<VL:mode>2"): Likewise.
16926 ("faligndata<VM64:mode>_vis"): Likewise.
16927 ("alignaddrsi_vis"): Likewise.
16928 ("alignaddrdi_vis"): Likweise.
16929 ("alignaddrlsi_vis"): Likewise.
16930 ("alignaddrldi_vis"): Likewise.
16931 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16932 ("bmaskdi_vis"): Likewise.
16933 ("bmasksi_vis"): Likewise.
16934 ("bshuffle<VM64:mode>_vis"): Likewise.
16935 ("cmask8<P:mode>_vis"): Likewise.
16936 ("cmask16<P:mode>_vis"): Likewise.
16937 ("cmask32<P:mode>_vis"): Likewise.
16938 ("pdistn<P:mode>_vis"): Likewise.
16939 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16941 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16943 * config/sparc/sparc.md ("subtype"): New insn attribute.
16944 ("*wrgsr_sp64"): Set insn subtype.
16945 ("*rdgsr_sp64"): Likewise.
16946 ("alignaddrsi_vis"): Likewise.
16947 ("alignaddrdi_vis"): Likewise.
16948 ("alignaddrlsi_vis"): Likewise.
16949 ("alignaddrldi_vis"): Likewise.
16950 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16951 ("fexpand_vis"): Likewise.
16952 ("fpmerge_vis"): Likewise.
16953 ("faligndata<VM64:mode>_vis"): Likewise.
16954 ("bshuffle<VM64:mode>_vis"): Likewise.
16955 ("cmask8<P:mode>_vis"): Likewise.
16956 ("cmask16<P:mode>_vis"): Likewise.
16957 ("cmask32<P:mode>_vis"): Likewise.
16958 ("fchksm16_vis"): Likewise.
16959 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16960 ("fmean16_vis"): Likewise.
16961 ("fp<plusminus_insn>64_vis"): Likewise.
16962 ("<plusminus_insn>v8qi3"): Likewise.
16963 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16964 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16965 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16966 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16967 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16968 ("*movqi_insn"): Likewise.
16969 ("*movhi_insn"): Likewise.
16970 ("*movsi_insn"): Likewise.
16971 ("movsi_pic_gotdata_op"): Likewise.
16972 ("*movdi_insn_sp32"): Likewise.
16973 ("*movdi_insn_sp64"): Likewise.
16974 ("movdi_pic_gotdata_op"): Likewise.
16975 ("*movsf_insn"): Likewise.
16976 ("*movdf_insn_sp32"): Likewise.
16977 ("*movdf_insn_sp64"): Likewise.
16978 ("*zero_extendhisi2_insn"): Likewise.
16979 ("*zero_extendqihi2_insn"): Likewise.
16980 ("*zero_extendqisi2_insn"): Likewise.
16981 ("*zero_extendqidi2_insn"): Likewise.
16982 ("*zero_extendhidi2_insn"): Likewise.
16983 ("*zero_extendsidi2_insn_sp64"): Likewise.
16984 ("ldfsr"): Likewise.
16985 ("prefetch_64"): Likewise.
16986 ("prefetch_32"): Likewise.
16987 ("tie_ld32"): Likewise.
16988 ("tie_ld64"): Likewise.
16989 ("*tldo_ldub_sp32"): Likewise.
16990 ("*tldo_ldub1_sp32"): Likewise.
16991 ("*tldo_ldub2_sp32"): Likewise.
16992 ("*tldo_ldub_sp64"): Likewise.
16993 ("*tldo_ldub1_sp64"): Likewise.
16994 ("*tldo_ldub2_sp64"): Likewise.
16995 ("*tldo_ldub3_sp64"): Likewise.
16996 ("*tldo_lduh_sp32"): Likewise.
16997 ("*tldo_lduh1_sp32"): Likewise.
16998 ("*tldo_lduh_sp64"): Likewise.
16999 ("*tldo_lduh1_sp64"): Likewise.
17000 ("*tldo_lduh2_sp64"): Likewise.
17001 ("*tldo_lduw_sp32"): Likewise.
17002 ("*tldo_lduw_sp64"): Likewise.
17003 ("*tldo_lduw1_sp64"): Likewise.
17004 ("*tldo_ldx_sp64"): Likewise.
17005 ("*mov<VM32:mode>_insn"): Likewise.
17006 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17007 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17009 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17011 * config/sparc/sparc.md ("type"): New insn type viscmp.
17012 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17014 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17015 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17016 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17017 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17019 ("n7_vis_logical_11cycle"): Likewise.
17020 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17021 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17022 * config/sparc/niagara.md ("niag_vis"): Likewise.
17023 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17024 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17026 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17028 * config/sparc/sparc.md: New instruction type `bmask'.
17029 (bmaskdi_vis): Use the `bmask' type.
17030 (bmasksi_vis): Likewise.
17031 * config/sparc/ultra3.md (us3_array): Likewise.
17032 * config/sparc/niagara7.md (n7_array): Likewise.
17033 * config/sparc/niagara4.md (n4_array): Likewise.
17034 * config/sparc/niagara2.md (niag2_vis): Likewise.
17035 (niag3_vis): Likewise.
17036 * config/sparc/niagara.md (niag_vis): Likewise.
17038 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17040 * ipa-comdats.c: Remove optimize check from gate.
17041 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17042 for functions not optimized.
17043 (ipa_fn_summary_read): Skip optimize check.
17044 (ipa_fn_summary_write): Likewise.
17045 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17047 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17049 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17050 (check_callers): Give up on caller which is not optimized.
17051 (inline_small_functions): Likewise.
17052 (ipa_inline): Do not give up when not optimizing.
17053 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17054 away unoptimizes cdtors.
17055 (whole_program_function_and_variable_visibility): Do
17056 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17057 * ipa.c (process_references): Do not check optimize.
17058 (symbol_table::remove_unreachable_nodes): Update optimize check.
17059 (set_writeonly_bit): Update optimize check.
17060 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17061 (pass_ipa_single_use::gate): Remove.
17063 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17065 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17066 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17067 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17068 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17069 permute_load, permute_store, adjust_extract, adjust_splat,
17070 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17071 replace_swap_with_copy, dump_swap_insn_table,
17072 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17073 recombine_lvx_pattern, recombine_stvx_pattern,
17074 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17075 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17076 to file rs6000-p8swap.c.
17077 * config/rs6000/rs6000-p8swap.c: New file.
17078 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17079 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17080 and rs6000*-*-* targets.
17082 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17084 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17086 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17088 * lto-wrapper.c (merge_and_complain): Do not merge
17089 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17090 fsigned_zeros, ftrapping_math, fwrapv.
17091 (append_compiler_options): Do not track these options.
17092 (append_linker_options): Likewie
17094 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17096 * cgraphunit.c (cgraph_node::finalize_function): When
17097 !flag_toplevel_reorde set no_reorder flag.
17098 (varpool_node::finalize_decl): Likewise.
17099 (symbol_table::compile): Drop no toplevel reorder path.
17101 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17103 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17104 edges; zero probability is not better than uninitialized.
17106 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17108 * asan.h (asan_sanitize_allocas_p): Declare.
17109 * asan.c (asan_sanitize_allocas_p): New function.
17110 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17111 (handle_builtin_alloca): Likewise.
17112 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17113 if !asan_sanitize_allocas_p.
17114 * params.def (asan-instrument-allocas): Add new option.
17115 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17116 * opts.c (common_handle_option): Disable allocas sanitization for
17119 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17121 * asan.c: Include gimple-fold.h.
17122 (get_last_alloca_addr): New function.
17123 (handle_builtin_stackrestore): Likewise.
17124 (handle_builtin_alloca): Likewise.
17125 (asan_emit_allocas_unpoison): Likewise.
17126 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17127 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17128 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17129 (instrument_builtin_call): Pass gimple iterator to
17130 get_mem_refs_of_builtin_call.
17131 (last_alloca_addr): New global.
17132 * asan.h (asan_emit_allocas_unpoison): Declare.
17133 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17134 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17135 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17136 if function calls alloca.
17137 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17138 * gimple-fold.h (replace_call_with_value): Declare.
17139 * internal-fn.c: Include asan.h.
17140 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17141 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17143 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17145 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17146 (C_SELFTEST_FLAGS): New.
17147 (CPP_SELFTEST_FLAGS): New.
17148 (SELFTEST_DEPS): New, from deps of s-selftest.
17149 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17150 (CPP_SELFTEST_DEPS): New.
17151 (selftest): Add dependency on s-selftest-c++.
17152 (s-selftest): Rename to...
17153 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17154 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17155 than SELFTEST_FLAGS.
17156 (selftest-gdb): Rename to...
17157 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17159 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17160 (selftest-valgrind): Rename to...
17161 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17163 (selftest-valgrind): Reintroduce as an alias for
17164 selftest-c-valgrind.
17165 (s-selftest-c++): New.
17166 (selftest-c++-gdb): New.
17167 (selftest-c++-valgrind): New.
17169 2017-07-06 Olivier Hainque <hainque@adacore.com>
17171 * gcc.c (process_command): When deciding if undefined variables
17172 should be ignored when processing specs, accept "gcc -v" as well.
17174 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17176 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17177 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17179 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17181 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17182 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17183 * config/arm/arm-tables.opt: Regenerate.
17184 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17186 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17188 2017-07-06 Carl Love <cel@us.ibm.com>
17190 * ChangeLog: Clean up from mid air collision
17192 2017-07-06 Carl Love <cel@us.ibm.com>
17194 * config/rs6000/rs6000-c.c: Add support for built-in functions
17195 vector signed int vec_subc (vector signed int, vector signed int);
17196 vector signed __int128 vec_subc (vector signed __int128,
17197 vector signed __int128);
17198 vector unsigned __int128 vec_subc (vector unsigned __int128,
17199 vector unsigned __int128);
17200 vector signed int vec_sube (vector signed int, vector signed int,
17201 vector signed int);
17202 vector unsigned int vec_sube (vector unsigned int,
17203 vector unsigned int,
17204 vector unsigned int);
17205 vector signed __int128 vec_sube (vector signed __int128,
17206 vector signed __int128,
17207 vector signed__int128);
17208 vector unsigned __int128 vec_sube (vector unsigned __int128,
17209 vector unsigned __int128,
17210 vector unsigned __int128);
17211 vector signed int vec_subec (vector signed int, vector signed int,
17212 vector signed int);
17213 vector unsigned int vec_subec (vector unsigned int,
17214 vector unsigned int,
17215 vector unsigned int);
17216 vector signed __int128 vec_subec (vector signed __int128,
17217 vector signed __int128,
17218 vector signed__int128);
17219 vector unsigned __int128 vec_subec (vector unsigned __int128,
17220 vector unsigned __int128,
17221 vector unsigned __int128);
17222 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17223 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17224 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17225 BU_ALTIVEC_OVERLOAD_X definitions.
17226 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17227 * doc/extend.texi: Update the built-in documentation file for the new
17228 built-in functions.
17230 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17233 * diagnostic-show-locus.c (layout::layout): Use start and finish
17234 spelling location for the start and finish of each range.
17235 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17236 Add unused aspect param.
17237 * input.c (expand_location_1): Add "aspect" param, and use it
17238 to access the correct part of the location.
17239 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17241 (expand_location_to_spelling_point): Likewise.
17242 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17243 param, and pass it to expand_location_1.
17245 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17247 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17248 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17249 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17250 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17251 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17252 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17253 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17254 _mm_maskz_getmant_ss): New intrinsics.
17255 (__builtin_ia32_getexpss128_mask): Changed to ...
17256 __builtin_ia32_getexpss128_round ... this.
17257 (__builtin_ia32_getexpsd128_mask): Changed to ...
17258 __builtin_ia32_getexpsd128_round ... this.
17259 * config/i386/i386-builtin-types.def
17260 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17261 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17262 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17263 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17264 __builtin_ia32_getmantss_mask_round): New builtins.
17265 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17266 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17267 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17268 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17269 * config/i386/sse.md
17270 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17271 avx512f_sgetexp<mode><mask_scalar_name>
17272 <round_saeonly_scalar_name> ... this.
17273 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17274 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17275 vgetexp<ssescalarmodesuffix>
17276 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17277 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17278 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17279 avx512f_vgetmant<mode><mask_scalar_name>
17280 <round_saeonly_scalar_name> ... this.
17281 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17282 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17283 vgetmant<ssescalarmodesuffix>
17284 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17285 %0<mask_scalar_operand4>, %1, %2
17286 <round_saeonly_scalar_mask_op4>, %3} ... this.
17287 * config/i386/subst.md (mask_scalar_operand4,
17288 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17289 round_saeonly_scalar_nimm_predicate): New subst attributes.
17291 2017-07-06 Julia Koval <julia.koval@intel.com>
17293 * config/i386/i386.c (ix86_erase_embedded_rounding):
17294 Remove code for old rounding pattern.
17296 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17298 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17300 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17302 * doc/sourcebuild.texi (Test Directives, Variants of
17303 dg-require-support): Add documentation for dg-require-stack-check.
17305 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17307 * config/i386/subst.md (mask_scalar, round_scalar,
17308 round_saeonly_scalar): New meta-templates.
17309 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17310 round_scalar_mask_operand3, round_scalar_mask_op3,
17311 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17312 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17313 round_saeonly_scalar_constraint,
17314 round_saeonly_scalar_prefix): New subst attribute.
17315 * config/i386/sse.md
17316 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17317 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17318 <round_scalar_name> ... this.
17319 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17320 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17321 <round_scalar_name> ... this.
17322 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17323 <sse>_vm<code><mode>3<mask_scalar_name>
17324 <round_saeonly_scalar_name> ... this.
17325 (v<plusminus_mnemonic><ssescalarmodesuffix>
17326 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17327 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17328 v<plusminus_mnemonic><ssescalarmodesuffix>
17329 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17330 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17331 (v<multdiv_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<multdiv_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<maxmin_float><ssescalarmodesuffix>
17338 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17339 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17340 v<maxmin_float><ssescalarmodesuffix>
17341 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17342 %0<mask_scalar_operand3>, %1, %<iptr>2
17343 <round_saeonly_scalar_mask_op3>} ... this.
17345 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17347 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17348 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17350 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17351 Alan Hayward <alan.hayward@arm.com>
17352 David Sherwood <david.sherwood@arm.com>
17354 * combine.c (simplify_if_then_else): Remove "enum" before
17356 * compare-elim.c (can_eliminate_compare): Likewise.
17357 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17359 (aarch64_lookup_simd_builtin_type): Likewise.
17360 (aarch64_simd_builtin_type): Likewise.
17361 (aarch64_init_simd_builtin_types): Likewise.
17362 (aarch64_simd_expand_args): Likewise.
17363 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17365 (aarch64_reverse_mask): Likewise.
17366 (aarch64_simd_emit_reg_reg_move): Likewise.
17367 (aarch64_gen_adjusted_ldpstp): Likewise.
17368 (aarch64_ccmp_mode_to_code): Likewise.
17369 (aarch64_operands_ok_for_ldpstp): Likewise.
17370 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17371 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17373 (aarch64_min_divisions_for_recip_mul): Likewise.
17374 (aarch64_reassociation_width): Likewise.
17375 (aarch64_get_condition_code_1): Likewise.
17376 (aarch64_simd_emit_reg_reg_move): Likewise.
17377 (aarch64_simd_attr_length_rglist): Likewise.
17378 (aarch64_reverse_mask): Likewise.
17379 (aarch64_operands_ok_for_ldpstp): Likewise.
17380 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17381 (aarch64_gen_adjusted_ldpstp): Likewise.
17382 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17384 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17385 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17386 (arm_lookup_simd_builtin_type): Likewise.
17387 (arm_simd_builtin_type): Likewise.
17388 (arm_init_simd_builtin_types): Likewise.
17389 (arm_expand_builtin_args): Likewise.
17390 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17391 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17392 (ft32_setup_incoming_varargs): Likewise.
17393 (ft32_function_arg): Likewise.
17394 (ft32_function_arg_advance): Likewise.
17395 (ft32_pass_by_reference): Likewise.
17396 (ft32_arg_partial_bytes): Likewise.
17397 (ft32_valid_pointer_mode): Likewise.
17398 (ft32_addr_space_pointer_mode): Likewise.
17399 (ft32_addr_space_legitimate_address_p): Likewise.
17400 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17402 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17403 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17404 (iamcu_alignment): Likewise.
17405 (canonicalize_vector_int_perm): Likewise.
17406 (ix86_noce_conversion_profitable_p): Likewise.
17407 (ix86_mpx_bound_mode): Likewise.
17408 (ix86_operands_ok_for_move_multiple): Likewise.
17409 * config/microblaze/microblaze-protos.h
17410 (microblaze_expand_conditional_branch_reg): Likewise.
17411 * config/microblaze/microblaze.c
17412 (microblaze_expand_conditional_branch_reg): Likewise.
17413 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17415 (rs6000_reassociation_width): Likewise.
17416 (rs6000_invalid_binary_op): Likewise.
17417 (fusion_p9_p): Likewise.
17418 (emit_fusion_p9_load): Likewise.
17419 (emit_fusion_p9_store): Likewise.
17420 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17422 (riscv_hard_regno_mode_ok_p): Likewise.
17423 (riscv_address_insns): Likewise.
17424 (riscv_split_symbol): Likewise.
17425 (riscv_legitimize_move): Likewise.
17426 (riscv_function_value): Likewise.
17427 (riscv_hard_regno_nregs): Likewise.
17428 (riscv_expand_builtin): Likewise.
17429 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17430 (riscv_build_integer): Likewise.
17431 (riscv_split_integer): Likewise.
17432 (riscv_legitimate_constant_p): Likewise.
17433 (riscv_cannot_force_const_mem): Likewise.
17434 (riscv_regno_mode_ok_for_base_p): Likewise.
17435 (riscv_valid_base_register_p): Likewise.
17436 (riscv_valid_offset_p): Likewise.
17437 (riscv_valid_lo_sum_p): Likewise.
17438 (riscv_classify_address): Likewise.
17439 (riscv_legitimate_address_p): Likewise.
17440 (riscv_address_insns): Likewise.
17441 (riscv_load_store_insns): Likewise.
17442 (riscv_force_binary): Likewise.
17443 (riscv_split_symbol): Likewise.
17444 (riscv_force_address): Likewise.
17445 (riscv_legitimize_address): Likewise.
17446 (riscv_move_integer): Likewise.
17447 (riscv_legitimize_const_move): Likewise.
17448 (riscv_legitimize_move): Likewise.
17449 (riscv_address_cost): Likewise.
17450 (riscv_subword): Likewise.
17451 (riscv_output_move): Likewise.
17452 (riscv_canonicalize_int_order_test): Likewise.
17453 (riscv_emit_int_order_test): Likewise.
17454 (riscv_function_arg_boundary): Likewise.
17455 (riscv_pass_mode_in_fpr_p): Likewise.
17456 (riscv_pass_fpr_single): Likewise.
17457 (riscv_pass_fpr_pair): Likewise.
17458 (riscv_get_arg_info): Likewise.
17459 (riscv_function_arg): Likewise.
17460 (riscv_function_arg_advance): Likewise.
17461 (riscv_arg_partial_bytes): Likewise.
17462 (riscv_function_value): Likewise.
17463 (riscv_pass_by_reference): Likewise.
17464 (riscv_setup_incoming_varargs): Likewise.
17465 (riscv_print_operand): Likewise.
17466 (riscv_elf_select_rtx_section): Likewise.
17467 (riscv_save_restore_reg): Likewise.
17468 (riscv_for_each_saved_reg): Likewise.
17469 (riscv_register_move_cost): Likewise.
17470 (riscv_hard_regno_mode_ok_p): Likewise.
17471 (riscv_hard_regno_nregs): Likewise.
17472 (riscv_class_max_nregs): Likewise.
17473 (riscv_memory_move_cost): Likewise.
17474 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17475 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17476 (rl78_addr_space_address_mode): Likewise.
17477 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17479 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17480 (rs6000_reassociation_width): Likewise.
17481 (rs6000_invalid_binary_op): Likewise.
17482 (fusion_p9_p): Likewise.
17483 (emit_fusion_p9_load): Likewise.
17484 (emit_fusion_p9_store): Likewise.
17485 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17486 (ok_for_simple_move_operands): Likewise.
17487 (ok_for_simple_move_strict_operands): Likewise.
17488 (ok_for_simple_arith_logic_operands): Likewise.
17489 (visium_legitimize_reload_address): Likewise.
17490 (visium_select_cc_mode): Likewise.
17491 (output_cbranch): Likewise.
17492 (visium_split_double_move): Likewise.
17493 (visium_expand_copysign): Likewise.
17494 (visium_expand_int_cstore): Likewise.
17495 (visium_expand_fp_cstore): Likewise.
17496 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17497 (visium_function_arg): Likewise.
17498 (visium_function_arg_advance): Likewise.
17499 (visium_libcall_value): Likewise.
17500 (visium_setup_incoming_varargs): Likewise.
17501 (visium_legitimate_constant_p): Likewise.
17502 (visium_legitimate_address_p): Likewise.
17503 (visium_legitimize_address): Likewise.
17504 (visium_secondary_reload): Likewise.
17505 (visium_register_move_cost): Likewise.
17506 (visium_memory_move_cost): Likewise.
17507 (prepare_move_operands): Likewise.
17508 (ok_for_simple_move_operands): Likewise.
17509 (ok_for_simple_move_strict_operands): Likewise.
17510 (ok_for_simple_arith_logic_operands): Likewise.
17511 (visium_function_value_1): Likewise.
17512 (rtx_ok_for_offset_p): Likewise.
17513 (visium_legitimize_reload_address): Likewise.
17514 (visium_split_double_move): Likewise.
17515 (visium_expand_copysign): Likewise.
17516 (visium_expand_int_cstore): Likewise.
17517 (visium_expand_fp_cstore): Likewise.
17518 (visium_split_cstore): Likewise.
17519 (visium_select_cc_mode): Likewise.
17520 (visium_split_cbranch): Likewise.
17521 (output_cbranch): Likewise.
17522 (visium_print_operand_address): Likewise.
17523 * expmed.c (flip_storage_order): Likewise.
17524 * expmed.h (emit_cstore): Likewise.
17525 (flip_storage_order): Likewise.
17526 * genrecog.c (validate_pattern): Likewise.
17527 * hsa-gen.c (gen_hsa_addr): Likewise.
17528 * internal-fn.c (expand_arith_overflow): Likewise.
17529 * ira-color.c (allocno_copy_cost_saving): Likewise.
17530 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17531 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17532 (process_invariant_for_inheritance): Likewise.
17533 * lra-eliminations.c (move_plus_up): Likewise.
17534 * omp-low.c (lower_oacc_reductions): Likewise.
17535 * simplify-rtx.c (simplify_subreg): Likewise.
17536 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17537 (TARGET_CHKP_BOUND_MODE): Likewise..
17538 * targhooks.c (default_chkp_bound_mode): Likewise.
17539 (default_setup_incoming_vararg_bounds): Likewise.
17540 * targhooks.h (default_chkp_bound_mode): Likewise.
17541 (default_setup_incoming_vararg_bounds): Likewise.
17542 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17543 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17544 (have_whole_vector_shift): Likewise.
17545 * tree-vect-stmts.c (vectorizable_load): Likewise.
17546 * doc/tm.texi: Regenerate.
17548 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17550 Graceful degrade if Binutils PR21472 is not available.
17553 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17554 .rodata in flash test fails.
17555 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17556 * confgure: Regenerate.
17557 * config.in: Regenerate.
17558 * config/avr/avr.c (avr_asm_named_section)
17559 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17560 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17561 (avr_asm_init_sections): Same.
17563 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17565 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17566 (fma<VH:mode>4_intrinsic): Likewise.
17567 (*fmsub<VCVTF:mode>4): Likewise.
17568 (*fmsub<VH:mode>4_intrinsic): Likewise.
17570 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17573 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17574 Don't depend on "optimize > 0".
17575 (out_movhi_r_mr, out_movqi_mr_r): Same.
17576 (out_movhi_mr_r, out_movqi_r_mr): Same.
17577 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17578 io_address_operand on "optimize > 0".
17580 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17582 * tree-loop-distribution.c: Add general explanantion on the pass.
17583 (generate_loops_for_partition): Mark distributed loop.
17584 (pg_add_dependence_edges): New parameter. Handle alias data
17585 dependence specially and record it in the parameter if asked.
17586 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17587 (init_partition_graph_vertices, add_partition_graph_edge): New.
17588 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17589 (free_partition_graph_vdata, build_partition_graph): New.
17590 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17591 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17592 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17593 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17594 (version_for_distribution_p, finalize_partitions): New.
17595 (distribute_loop): Handle alias data dependence specially. Factor
17596 out loop fusion code as functions and call these functions.
17598 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17600 * tree-loop-distribution.c (classify_partition): New parameter and
17601 better handle reduction statement.
17602 (rdg_build_partitions): Revise comment.
17603 (distribute_loop): Compute statements in all partitions and pass it
17604 to classify_partition.
17606 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17608 * tree-loop-distribution.c (enum partition_type): New.
17609 (struct partition): New field type.
17610 (partition_merge_into): Add parameter. Update partition type.
17611 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17612 (build_rdg_partition_for_vertex): Compute partition type.
17613 (rdg_build_partitions): Dump partition type.
17614 (distribute_loop): Update calls to partition_merge_into.
17616 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17618 * tree-loop-distribution.c (struct ddr_hasher): New.
17619 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17621 (classify_partition): Call get_data_dependence.
17622 (pg_add_dependence_edges): Ditto.
17623 (distribute_loop): Release data dependence hash table.
17625 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17627 * tree-loop-distribution.c (ref_base_address): Delete.
17628 (similar_memory_accesses): Rename ...
17629 (share_memory_accesses): ... to this. Check if partitions access
17630 the same memory reference.
17631 (distribute_loop): Call share_memory_accesses.
17633 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17635 * tree-loop-distribution.c (struct partition): New field recording
17636 its data reference.
17637 (partition_alloc, partition_free): Init and release data refs.
17638 (partition_merge_into): Merge data refs.
17639 (build_rdg_partition_for_vertex): Collect data refs for partition.
17640 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17642 (distribute_loop): Remve data refs from vertice data of partition
17645 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17647 * tree-loop-distribution.c (params.h): Include header file.
17648 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17649 (datarefs_vec): New global var.
17650 (create_rdg_vertices): Use datarefs_vec directly.
17651 (free_rdg): Don't free data references.
17652 (build_rdg): Update use. Don't free data references.
17653 (distribute_loop): Compute global variable for data references.
17654 Bail out if there are too many data references.
17656 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17658 * tree-loop-distribution.c (loop_nest): New global var.
17659 (build_rdg): Use loop directly, rather than loop nest.
17660 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17662 (distribute_loop): Compute global variable loop nest. Update use.
17664 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17666 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17667 (partition_merge_into): New parameter. Dump reason for fusion.
17668 (distribute_loop): Update use of partition_merge_into.
17670 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17672 * tree-loop-distribution.c (bb_top_order_index): New.
17673 (bb_top_order_index_size, bb_top_order_cmp): New.
17674 (stmts_from_loop): Use topological order.
17675 (pass_loop_distribution::execute): Compute and release topological
17676 order for basic blocks.
17678 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17680 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17683 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17685 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17686 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17687 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17688 * internal-fn.def (LOOP_DIST_ALIAS): New.
17689 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17690 (fold_loop_internal_call): ... this.
17691 (vect_loop_dist_alias_call): New function.
17692 (set_uid_loop_bbs): Call fold_loop_internal_call.
17693 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17696 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17699 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17700 Require dead FLAGS_REG at the beginning of a peephole.
17702 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17705 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17706 arguments in the call to __builtin_ia32_sbb_u32.
17707 (_subborrow_u64): Swap _X and _Y arguments in the call to
17708 __builtin_ia32_sbb_u64.
17710 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17713 * tree-vrp.c (compare_assert_loc): Turn into a function template
17714 with stable template parameter. Only test if a->e is NULL,
17715 !a->e == !b->e has been verified already. Use e == NULL or
17716 e != NULL instead of e or ! e tests. If stable is true, don't use
17717 iterative_hash_expr, on the other side allow a or b or both NULL
17718 and sort the NULLs last.
17719 (process_assert_insertions): Sort using compare_assert_loc<false>
17720 instead of compare_assert_loc, later sort using
17721 compare_assert_loc<true> before calling process_assert_insertions_for
17722 in a loop. Use break instead of continue once seen NULL pointer.
17724 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17726 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17727 Cortex-R7 and Cortex-R8 processors.
17729 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17731 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17732 uninitialized while src is not.
17734 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17736 * common/config/arm/arm-common.c: Adjust include path for
17738 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17739 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17740 (arm-cpu-data.h): Likewise.
17741 (arm-cpu-cdata.h): Likewise.
17742 * config/arm/arm-cpu.h: Delete.
17743 * config/arm/arm-cpu-cdata.h: Delete.
17744 * config/arm/arm-cpu-data.h: Delete.
17746 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17748 * config/arm/arm-cpus.in (cortex-a55): New.
17749 (cortex-a75): Likewise.
17750 (cortex-a75.cortex-a55): Likewise.
17751 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17753 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17754 * config/arm/arm-cpu-cdata.h: Regenerate.
17755 * config/arm/arm-cpu-data.h: Regenerate.
17756 * config/arm/arm-cpu.h: Regenerate.
17757 * config/arm/arm-tables.opt: Regenerate.
17758 * config/arm/arm-tune.md: Regenerate.
17760 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17762 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17764 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17766 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17769 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17771 PR tree-optimization/81292
17772 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17773 full_string_p, also call adjust_related_strinfos if the adjustment
17774 is simple, otherwise invalidate related strinfos.
17776 2017-07-04 Martin Liska <mliska@suse.cz>
17779 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17780 newly created variable as DECL_IGNORED_P.
17782 2017-07-04 Martin Liska <mliska@suse.cz>
17785 * ipa-inline.c (inline_small_functions):
17786 Use xstrdup_for_dump.
17788 2017-07-04 Tom de Vries <tom@codesourcery.com>
17790 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17792 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17795 * config/darwin.c (darwin_function_switched_text_sections):
17798 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17800 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17802 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17804 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17806 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17808 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17809 min_profitable_iters, and th as inclusive lower bounds.
17810 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17811 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17812 for min_profitable_iters and min_profitable_estimate.
17813 (vect_transform_loop): Treat th as an inclusive lower bound.
17814 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17816 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17819 * config/darwin.c (darwin_function_switched_text_sections):
17820 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17821 in two pieces, and suppress the use of buf.
17823 2017-07-03 Nathan Sidwell <nathan@acm.org>
17825 * hash-table.h (hash_table_mod1): Fix indentation.
17827 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17829 PR middle-end/81290
17830 * predict.c (force_edge_cold): Be more careful about propagation
17832 * profile-count.h (profile_probability::guessed,
17833 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17835 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17837 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17839 * doc/invoke.texi (rcpc architecture extension): Document it.
17841 2017-07-03 Richard Biener <rguenther@suse.de>
17843 PR tree-optimization/60510
17844 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17845 the scalar reduction PHI and use it.
17846 (vectorizable_reduction): Properly guard the single_defuse_cycle
17847 path for non-SLP reduction chains where we cannot use it.
17848 Rework reduc_def/index and vector type deduction. Rework
17849 vector operand gathering during reduction op code-gen.
17850 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17851 chains dissolve the chain and leave it to non-SLP reduction
17854 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17856 * tree-data-ref.h (dr_alignment): Declare.
17857 * tree-data-ref.c (dr_alignment): New function.
17858 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17859 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17861 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17863 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17865 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17866 and base_misalignment fields.
17867 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17868 * tree-data-ref.c: Include builtins.h.
17869 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17870 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17871 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17872 * tree-vect-data-refs.c: Include tree-cfg.h.
17873 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17874 fields instead of calculating an alignment here.
17875 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17876 innermost_loop_behavior fields.
17878 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17880 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17882 (DR_STEP_ALIGNMENT): New macro.
17883 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17884 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17885 (create_data_ref): Print it.
17886 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17887 to tell whether the step preserves vector (mis)alignment.
17888 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17889 Move the check for an integer step and generalise to all INTEGER_CST.
17890 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17891 Print the outer step alignment.
17893 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17895 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17896 with offset_alignment.
17897 (DR_ALIGNED_TO): Delete.
17898 (DR_OFFSET_ALIGNMENT): New macro.
17899 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17900 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17901 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17902 (create_data_ref): Likewise.
17903 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17904 (vect_analyze_data_refs): Likewise.
17905 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17906 creating dummy innermost behavior.
17908 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17910 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17911 with a "innermost_loop_behavior *" and refeence tree.
17912 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17913 (create_data_ref): Update call accordingly.
17914 * tree-predcom.c (find_looparound_phi): Likewise.
17916 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17918 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17919 fields with dr_wrt_vec_loop.
17920 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17921 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17922 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17923 (vect_dr_behavior): New function.
17924 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17925 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17926 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17927 track whether the step preserves the misalignment.
17928 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17929 Use vect_dr_behavior.
17930 (vect_setup_realignment): Update call accordingly.
17931 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17932 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17933 call to vect_create_addr_base_for_vector_ref.
17934 (vect_create_cond_for_align_checks): Likewise.
17935 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17936 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17937 (vect_recog_mask_conversion_pattern): Likewise.
17938 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17939 (new_stmt_vec_info): Remove redundant zeroing.
17941 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17943 * common/config/arm/arm-common.c (arm_be8_option): New function.
17944 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17945 (ISA_ARMv6): Add isa_bit_be8.
17946 * config/arm/arm.h (arm_be8_option): Add prototype.
17947 (BE8_SPEC_FUNCTION): New define.
17948 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17949 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17950 (mlittle-endian): Similarly.
17951 (mbe8, mbe32): New options.
17952 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17953 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17955 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17957 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17959 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17961 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17962 (cleanup_tree_cfg_bb): Use it.
17963 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17965 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17967 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17970 * loop-doloop.c (add_test): Update profile.
17972 2017-07-03 Martin Liska <mliska@suse.cz>
17975 * sanopt.c (rewrite_usage_of_param): New function.
17976 (sanitize_rewrite_addressable_params): Likewise.
17977 (pass_sanopt::execute): Call rewrite_usage_of_param.
17979 2017-07-03 Richard Biener <rguenther@suse.de>
17981 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17982 back to using VIEW_CONVERT_EXPR.
17984 2017-07-03 Martin Liska <mliska@suse.cz>
17987 * doc/extend.texi: Document when a resolver function is
17988 generated for target_clones.
17990 2017-07-03 Martin Liska <mliska@suse.cz>
17992 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17993 and shadow memory of auto variables which are subject of
17994 use-after-scope sanitization.
17995 (asan_expand_mark_ifn): Add do set only when is_poison.
17997 2016-07-03 Richard Biener <rguenther@suse.de>
17999 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18001 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18002 (vectorizable_reduction): Perform reduction PHI creation when
18003 visiting a reduction PHI and adjust and simplify code generation
18004 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18005 (vect_transform_loop): Visit reduction PHIs.
18006 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18007 defs into the SLP tree.
18008 (vect_build_slp_tree): Reduction defs terminate the recursion.
18009 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18011 (vect_get_vec_defs_for_stmt_copy): Export.
18012 (vect_get_vec_defs): Likewise.
18013 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18015 (vect_get_vec_defs_for_stmt_copy): Declare.
18016 (vect_get_vec_defs): Likewise.
18018 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18020 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18021 parameter with a "loop" parameter and use it instead of the
18022 loop containing DR_STMT. Don't check simple_iv when doing
18023 BB analysis. Describe the two analysis modes in the comment.
18025 2017-07-03 Tom de Vries <tom@codesourcery.com>
18027 PR tree-optimization/69468
18028 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18029 (find_same_succ_bb): Handle ignore_edge_flags.
18031 2017-07-03 Tom de Vries <tom@codesourcery.com>
18033 PR tree-optimization/81192
18034 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18036 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18038 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18040 2017-07-03 Tom de Vries <tom@codesourcery.com>
18042 PR tree-optimization/81192
18043 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18044 BB_SAME_SUCC (bb) == NULL.
18046 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18048 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18051 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18053 * dumpfile.c: Include profile-count.h
18054 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18056 (insert_cond_bb): Update profile.
18057 * tree-cfg.h (insert_cond_bb): Update prototype.
18058 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18059 * tree-dump.c: Do not include tree-cfg.
18061 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18063 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18065 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18067 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18070 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18072 * tree-complex.c (expand_complex_div_wide): update profile.
18074 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18075 Alan Hayward <alan.hayward@arm.com>
18076 David Sherwood <david.sherwood@arm.com>
18078 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18079 (CORETYPES_H): New define.
18080 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18081 (insn-modes-inline.h, s-modes-inline-h): New rules.
18082 (generated_files): Add insn-modes-inline.h.
18083 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18084 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18085 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18086 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18087 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18088 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18089 (build/gencodes.o, build/genconditions.o): Likewise.
18090 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18091 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18092 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18093 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18094 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18095 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18096 * coretypes.h: Include everything up to real.h for generators.
18097 Include insn-modes.h first. Include wide-int-print.h after
18098 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18099 * machmode.h: Don't include insn-modes.h here.
18100 * function-tests.c: Remove includes of signop.h, machmode.h,
18101 double-int.h and wide-int.h.
18103 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18105 * optc-save-gen.awk: Likewise.
18106 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18107 * godump.c: Remove include of wide-int-print.h.
18108 * pretty-print.h: Likewise.
18109 * wide-int-print.cc: Likewise.
18110 * wide-int.cc: Likewise.
18111 * hash-map-tests.c: Remove include of signop.h.
18112 * hash-set-tests.c: Likewise.
18113 * rtl-tests.c: Likewise.
18114 * mkconfig.sh: Remove include of machmode.h.
18115 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18117 (emit_insn_modes_inline_h): ...this new function. Emit the code
18118 into an insn-modes-inline.h header file, adding appropriate
18119 include guards and end comments.
18120 (emit_insn_modes_c_header): Remove include of machmode.h.
18121 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18123 (main): Handle -i flag and call emit_insn_modes_inline_h when
18126 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18128 * tree-ssa-strlen.c (strinfo): Rename the length field to
18129 nonzero_chars. Add a full_string_p field.
18130 (compare_nonzero_chars, zero_length_string_p): New functions.
18131 (get_addr_stridx): Add an offset_out parameter.
18132 Use compare_nonzero_chars.
18133 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18134 (new_strinfo): Update after above changes to strinfo.
18135 (set_endptr_and_length): Set full_string_p.
18136 (get_string_length): Update after above changes to strinfo.
18137 (unshare_strinfo): Update call to new_strinfo.
18138 (maybe_invalidate): Likewise.
18139 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18140 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18141 as a uhwi instead of an shwi. Update after above changes to
18142 strinfo and new_strinfo.
18143 (zero_length_string): Assert that chainsi contains full strings.
18144 Use zero_length_string_p. Update call to new_strinfo.
18145 (adjust_related_strinfos): Update after above changes to strinfo.
18146 Copy full_string_p from origsi.
18147 (adjust_last_stmt): Use zero_length_string_p.
18148 (handle_builtin_strlen): Update after above changes to strinfo and
18149 new_strinfo. Install the lhs as the string length if the previous
18150 entry didn't describe a full string.
18151 (handle_builtin_strchr): Update after above changes to strinfo
18153 (handle_builtin_strcpy): Likewise.
18154 (handle_builtin_strcat): Likewise.
18155 (handle_builtin_malloc): Likewise.
18156 (handle_pointer_plus): Likewise.
18157 (handle_builtin_memcpy): Likewise. Track nonzero characters
18158 that aren't necessarily followed by a nul terminator.
18159 (handle_char_store): Likewise.
18161 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18163 PR tree-optimization/80769
18164 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18165 for malloc and calloc. Document the new invariant that all related
18166 strinfos have delayed lengths or none do.
18167 (verify_related_strinfos): Move earlier in file.
18168 (set_endptr_and_length): New function, split out from...
18169 (get_string_length): ...here. Also set the lengths of related
18171 (zero_length_string): Assert that chainsi has known (rather than
18173 (adjust_related_strinfos): Likewise.
18175 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18177 PR tree-optimization/81136
18178 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18179 assert that two references with the same misalignment have the same
18180 compile-time misalignment if those compile-time misalignments
18183 2017-07-01 Andi Kleen <ak@linux.intel.com>
18185 * print-tree.c (print_node): Print all attributes.
18187 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18189 * cfg.c (scale_bbs_frequencies): New function.
18190 * cfg.h (scale_bbs_frequencies): Declare it.
18191 * cfgloopanal.c (single_likely_exit): Cleanup.
18192 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18194 (scale_loop_profile): Likewise.
18195 (loop_version): Likewise.
18196 (create_empty_loop_on_edge): Update.
18197 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18198 scale_loop_frequencies, scale_loop_profile, loopify,
18199 loop_version): Update prototypes.
18200 * modulo-sched.c (sms_schedule): Update.
18201 * predict.c (unlikely_executed_edge_p): Also check probability.
18202 (probably_never_executed_edge_p): Fix typo.
18203 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18204 * tree-parloops.c (gen_parallel_loop): Update.
18205 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18206 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18207 * tree-ssa-loop-split.c (split_loop): Update.
18208 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18209 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18210 (vect_loop_versioning): Update.
18211 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18213 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18215 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18217 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18219 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18220 to keep profile consistent.
18222 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18224 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18225 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18226 * profile-count.h (max_safe_multiplier): Make unsigned.
18227 (profile_count::guessed_zero): New.
18229 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18231 * bb-reorder.c (fix_up_crossing_landing_pad,
18232 fix_crossing_conditional_branches): Use make_single_succ_edge
18233 to keep profile consistent.
18235 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18237 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18240 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18243 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18244 the right scopes, make sure cond_jump isn't preserved between multiple
18245 iterations. Search for fallthru edge whenever there are 3+ edges and
18246 use find_fallthru_edge for it.
18248 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18250 Patch by Alexander Monakov <amonakov@ispras.ru>
18251 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18252 probabilities consistently.
18254 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18256 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18257 emit_cmp_and_jump_insns.
18259 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18262 * tree-inline.c (expand_call_inline): Combine profile statuses.
18264 2017-06-30 Andrew Pinski <apinski@cavium.com>
18266 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18267 fold_stmt returned true.
18269 2017-06-30 Nathan Sidwell <nathan@acm.org>
18271 * ggc.h (empty_string): Delete.
18272 * cfgexpand.c (expand_asm_stmt): Use plain "".
18273 * optabs.c (expand_asm_memory_barrier): Likewise.
18274 * stringpool.c (empty_string): Delete.
18275 (digit_vector, digit_string): Delete.
18276 (ggc_alloc_string): Use plain "", don't optimize single digit
18277 strings. Use ggc_alloc_atomic.
18279 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18281 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18282 comparison set and one other set, use the cost of the non-comparison
18285 2017-06-30 Nathan Sidwell <nathan@acm.org>
18287 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18290 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18292 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18293 loops. Remove now unneeded calls to gimple_switch_set_label() that
18294 just set removed labels to NULL_TREE.
18296 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18298 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18299 (set_range_info): ...here. Only call set_range_info_raw if domain
18301 (set_nonzero_bits): Call set_range_info_raw.
18302 * tree-ssanames.h (set_range_info_raw): New.
18304 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18307 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18308 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18309 of nonimmediate_operand and <store_mask_constraint> instead of m
18310 for the input operand. For V8FI iterator, always split if input
18311 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18312 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18313 <store_mask_predicate> instead of register_operand and
18314 <store_mask_constraint> instead of v for the input operand. Make
18315 sure both operands aren't MEMs for if not <mask_applied>.
18317 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18319 * lto-wrapper.c (copy_file) Close both file descriptors before
18322 2017-06-30 Martin Liska <mliska@suse.cz>
18325 * multiple_target.c (create_dispatcher_calls): Make ifunc
18326 also for function that don't have calls or are not referenced.
18328 2017-06-30 Richard Biener <rguenther@suse.de>
18330 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18331 analyze the first scalar stmt. Move vector type computation
18332 for the BB case here from ...
18333 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18334 live operation processing in the SLP case properly.
18336 2017-06-30 Richard Biener <rguenther@suse.de>
18338 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18340 2017-06-30 Martin Liska <mliska@suse.cz>
18343 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18344 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18346 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18348 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18351 2017-06-30 Richard Biener <rguenther@suse.de>
18353 PR tree-optimization/81249
18354 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18355 condition reduction result to original scalar type.
18357 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18359 * profile-count.h (enum profile_quality): Fix typos and whitespace
18362 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18364 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18365 type for branch probabilities.
18367 2017-06-29 Julian Brown <julian@codesourcery.com>
18368 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18370 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18371 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18372 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18373 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18375 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18377 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18378 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18379 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18380 CC usage from generic code to here.
18381 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18382 CC usage into the target macros.
18384 2017-06-29 Maya Rashish <coypu@sdf.org>
18386 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18389 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18391 * arm/arm-builtins.c: Include profile-count.h
18392 * except.c (sjlj_emit_function_enter): Use
18393 profile_probability::unlikely.
18395 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18397 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18398 and tocrel_offset be pointer args rather than implicitly using
18400 (legitimate_constant_pool_address_p, rs6000_emit_move,
18401 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18402 tocrel_offset and use in toc_relative_expr_p call.
18403 (print_operand, print_operand_address): Use static tocrel_base_oac
18404 and tocrel_offset_oac.
18405 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18408 2017-06-29 Maya Rashish <coypu@sdf.org>
18410 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18412 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18414 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18415 objects, take into account only the alignment of 'op0' and 'mode1' if
18418 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18420 * ccmp.c (ccmp_tree_comparison_p): New function.
18421 (ccmp_candidate_p): Update to use above function.
18422 (get_compare_parts): New function.
18423 (expand_ccmp_next): Update to use new functions.
18424 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18426 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18427 take mode as argument.
18428 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18429 * expr.c (expand_expr_real_1): Pass mode as argument.
18431 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18433 * combine.c (combine_instructions): Print insns to dump_file, together
18436 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18438 * asan.c (asan_emit_stack_protection): Update.
18439 (create_cond_insert_point): Update.
18440 * auto-profile.c (afdo_propagate_circuit): Update.
18441 * basic-block.h (struct edge_def): Turn probability to
18442 profile_probability.
18443 (EDGE_FREQUENCY): Update.
18444 * bb-reorder.c (find_traces_1_round): Update.
18445 (better_edge_p): Update.
18446 (sanitize_hot_paths): Update.
18447 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18448 (make_single_succ_edge): Update.
18449 (check_bb_profile): Update.
18450 (dump_edge_info): Update.
18451 (update_bb_profile_for_threading): Update.
18452 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18453 probabilitycount to 0.
18454 * cfgbuild.c (compute_outgoing_frequencies): Update.
18455 * cfgcleanup.c (try_forward_edges): Update.
18456 (outgoing_edges_match): Update.
18457 (try_crossjump_to_edge): Update.
18458 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18459 (expand_gimple_tailcall): Update.
18460 (construct_init_block): Use make_single_succ_edge.
18461 (construct_exit_block): Use make_single_succ_edge.
18462 * cfghooks.c (verify_flow_info): Update.
18463 (redirect_edge_succ_nodup): Update.
18464 (split_edge): Update.
18465 (account_profile_record): Update.
18466 * cfgloopanal.c (single_likely_exit): Update.
18467 * cfgloopmanip.c (scale_loop_profile): Update.
18468 (set_zero_probability): Remove.
18469 (duplicate_loop_to_header_edge): Update.
18470 * cfgloopmanip.h (loop_version): Update prototype.
18471 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18472 (force_nonfallthru_and_redirect): Update.
18473 (update_br_prob_note): Update.
18474 (rtl_verify_edges): Update.
18475 (purge_dead_edges): Update.
18476 (rtl_lv_add_condition_to_bb): Update.
18477 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18478 * cgraphunit.c (init_lowered_empty_function): Update.
18479 (cgraph_node::expand_thunk): Update.
18480 * cilk-common.c: Include profile-count.h
18481 * dojump.c (inv): Remove.
18482 (jumpifnot): Update.
18483 (jumpifnot_1): Update.
18484 (do_jump_1): Update.
18486 (do_jump_by_parts_greater_rtx): Update.
18487 (do_compare_rtx_and_jump): Update.
18488 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18489 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18490 * dwarf2cfi.c: Include profile-count.h
18491 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18492 (sjlj_emit_dispatch_table): Likewise.
18493 * explow.c: Include profile-count.h
18494 * expmed.c (emit_store_flag_force): Update.
18495 (do_cmp_and_jump): Update.
18496 * expr.c (compare_by_pieces_d::generate): Update.
18497 (compare_by_pieces_d::finish_mode): Update.
18498 (emit_block_move_via_loop): Update.
18499 (store_expr_with_bounds): Update.
18500 (store_constructor): Update.
18501 (expand_expr_real_2): Update.
18502 (expand_expr_real_1): Update.
18503 * expr.h (try_casesi, try_tablejump): Update prototypes.
18504 * gimple-pretty-print.c (dump_probability): Update.
18505 (dump_profile): New.
18506 (dump_gimple_label): Update.
18507 (dump_gimple_bb_header): Update.
18508 * graph.c (draw_cfg_node_succ_edges): Update.
18509 * hsa-gen.c (convert_switch_statements): Update.
18510 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18511 (find_if_case_1): Update.
18512 (find_if_case_2): Update.
18513 * internal-fn.c (expand_arith_overflow_result_store): Update.
18514 (expand_addsub_overflow): Update.
18515 (expand_neg_overflow): Update.
18516 (expand_mul_overflow): Update.
18517 (expand_vector_ubsan_overflow): Update.
18518 * ipa-cp.c (good_cloning_opportunity_p): Update.
18519 * ipa-split.c (split_function): Use make_single_succ_edge.
18520 * ipa-utils.c (ipa_merge_profiles): Update.
18521 * loop-doloop.c (add_test): Update.
18522 (doloop_modify): Update.
18523 * loop-unroll.c (compare_and_jump_seq): Update.
18524 (unroll_loop_runtime_iterations): Update.
18525 * lra-constraints.c (lra_inheritance): Update.
18526 * lto-streamer-in.c (input_cfg): Update.
18527 * lto-streamer-out.c (output_cfg): Update.
18528 * mcf.c (adjust_cfg_counts): Update.
18529 * modulo-sched.c (sms_schedule): Update.
18530 * omp-expand.c (expand_omp_for_init_counts): Update.
18531 (extract_omp_for_update_vars): Update.
18532 (expand_omp_ordered_sink): Update.
18533 (expand_omp_for_ordered_loops): Update.
18534 (expand_omp_for_generic): Update.
18535 (expand_omp_for_static_nochunk): Update.
18536 (expand_omp_for_static_chunk): Update.
18537 (expand_cilk_for): Update.
18538 (expand_omp_simd): Update.
18539 (expand_omp_taskloop_for_outer): Update.
18540 (expand_omp_taskloop_for_inner): Update.
18541 * omp-simd-clone.c (simd_clone_adjust): Update.
18542 * optabs.c (expand_doubleword_shift): Update.
18543 (expand_abs): Update.
18544 (emit_cmp_and_jump_insn_1): Update.
18545 (expand_compare_and_swap_loop): Update.
18546 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18547 * predict.c (predictable_edge_p): Update.
18548 (edge_probability_reliable_p): Update.
18549 (set_even_probabilities): Update.
18550 (combine_predictions_for_insn): Update.
18551 (combine_predictions_for_bb): Update.
18552 (propagate_freq): Update.
18553 (estimate_bb_frequencies): Update.
18554 (force_edge_cold): Update.
18555 * profile-count.c (profile_count::dump): Add missing space into dump.
18556 (profile_count::debug): Add newline.
18557 (profile_count::differs_from_p): Explicitly convert to unsigned.
18558 (profile_count::stream_in): Update.
18559 (profile_probability::dump): New member function.
18560 (profile_probability::debug): New member function.
18561 (profile_probability::differs_from_p): New member function.
18562 (profile_probability::differs_lot_from_p): New member function.
18563 (profile_probability::stream_in): New member function.
18564 (profile_probability::stream_out): New member function.
18565 * profile-count.h (profile_count_quality): Rename to ...
18566 (profile_quality): ... this one.
18567 (profile_probability): New.
18568 (profile_count): Update.
18569 * profile.c (compute_branch_probabilities): Update.
18570 * recog.c (peep2_attempt): Update.
18571 * sched-ebb.c (schedule_ebbs): Update.
18572 * sched-rgn.c (find_single_block_region): Update.
18573 (compute_dom_prob_ps): Update.
18574 (schedule_region): Update.
18575 * sel-sched-ir.c (compute_succs_info): Update.
18576 * stmt.c (struct case_node): Update.
18577 (do_jump_if_equal): Update.
18578 (get_outgoing_edge_probs): Update.
18579 (conditional_probability): Update.
18580 (emit_case_dispatch_table): Update.
18581 (expand_case): Update.
18582 (expand_sjlj_dispatch_table): Update.
18583 (emit_case_nodes): Update.
18584 * targhooks.c: Update.
18585 * tracer.c (better_p): Update.
18586 (find_best_successor): Update.
18587 * trans-mem.c (expand_transaction): Update.
18588 * tree-call-cdce.c: Update.
18589 * tree-cfg.c (gimple_split_edge): Upate.
18590 (move_sese_region_to_fn): Upate.
18591 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18592 * tree-eh.c (lower_resx): Upate.
18593 (cleanup_empty_eh_move_lp): Upate.
18594 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18595 * tree-inline.c (copy_edges_for_bb): Update.
18596 (copy_cfg_body): Update.
18597 * tree-parloops.c (gen_parallel_loop): Update.
18598 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18599 (gimple_gen_time_profiler): Update.
18600 * tree-ssa-dce.c (remove_dead_stmt): Update.
18601 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18602 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18603 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18604 (unloop_loops): Update.
18605 (try_peel_loop): Update.
18606 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18607 * tree-ssa-loop-split.c (connect_loops): Update.
18608 (split_loop): Update.
18609 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18610 (hoist_guard): Update.
18611 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18612 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18613 (value_replacement): Update.
18614 * tree-ssa-reassoc.c (branch_fixup): Update.
18615 * tree-ssa-tail-merge.c (replace_block_by): Update.
18616 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18617 (create_edge_and_update_destination_phis): Update.
18618 (compute_path_counts): Update.
18619 (recompute_probabilities): Update.
18620 (update_joiner_offpath_counts): Update.
18621 (freqs_to_counts_path): Update.
18622 (duplicate_thread_path): Update.
18623 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18624 (struct switch_conv_info): Update.
18625 (gen_inbound_check): Update.
18626 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18627 (vect_do_peeling): Update.
18628 (vect_loop_versioning): Update.
18629 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18630 (optimize_mask_stores): Update.
18631 * ubsan.c (ubsan_expand_null_ifn): Update.
18632 * value-prof.c (gimple_divmod_fixed_value): Update.
18633 (gimple_divmod_fixed_value_transform): Update.
18634 (gimple_mod_pow2): Update.
18635 (gimple_mod_pow2_value_transform): Update.
18636 (gimple_mod_subtract): Update.
18637 (gimple_mod_subtract_transform): Update.
18638 (gimple_ic): Update.
18639 (gimple_stringop_fixed_value): Update.
18640 (gimple_stringops_transform): Update.
18641 * value-prof.h: Update.
18643 2017-06-29 Carl Love <cel@us.ibm.com>
18645 * config/rs6000/rs6000-c.c: Add support for built-in functions
18646 vector signed int vec_signed (vector float);
18647 vector signed long long vec_signed (vector double);
18648 vector signed int vec_signed2 (vector double, vector double);
18649 vector signed int vec_signede (vector double);
18650 vector signed int vec_signedo (vector double);
18651 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18652 instruction generator.
18653 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18654 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18655 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18657 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18658 vunsignede_v2df): Add define_expands.
18659 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18660 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18661 VEC_UNSIGNEDO): Add definitions.
18662 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18663 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18664 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18665 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18666 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18667 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18668 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18669 vec_unsignede, vec_unsignedo): Add builtin defines.
18670 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18672 * doc/extend.texi: Update the built-in documentation file for the
18673 new built-in functions.
18675 2017-06-29 Richard Biener <rguenther@suse.de>
18677 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18678 reduction chains to LOOP_VINFO_REDUCTIONS.
18679 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18680 SLP reductions after processing reduction chains.
18682 2017-06-29 Nathan Sidwell <nathan@acm.org>
18684 * builtins.c (fold_builtin_FUNCTION): Use
18685 lang_hooks.decl_printable_name.
18687 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18689 PR middle-end/81194
18690 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18691 with only one label.
18692 * stmt.c (expand_case): Assert NCASES is greater than one.
18694 2017-06-29 Richard Biener <rguenther@suse.de>
18696 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18698 (group_case_labels): Likewise.
18699 (find_taken_edge): Push sanity checking on val to workers...
18700 (find_taken_edge_cond_expr): ... here
18701 (find_taken_edge_switch_expr): ... and here, handle cases
18702 with just a default label.
18703 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18704 (group_case_labels): Likewise.
18705 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18706 group_case_labels does anything cleanup the CFG again.
18708 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18710 PR tree-optimization/81196
18711 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18712 exit condition comparing two IVs.
18714 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18716 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18717 profile to the dummy entry at the end of the list of architectures.
18718 * config/arm/arm-cpu-cdata.h: Regenerated.
18720 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18721 Michael Collison <michael.collison@arm.com>
18724 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18726 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18727 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18728 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18729 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18730 of shift when the shift amount is masked with constant equal to
18731 the size of the mode.
18732 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18735 2017-06-29 Martin Liska <mliska@suse.cz>
18737 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18739 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18742 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18745 (edge_growth_cache_entry::edge_growth_cache_entry): New
18747 (reset_edge_growth_cache): Update to use constructor.
18749 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18751 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18752 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18753 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18755 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18757 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18758 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18760 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18762 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18763 (*-linux-uclibc*): Add t-uclibc tmake_file.
18764 * config/t-musl: New.
18765 * config/t-uclibc: New.
18767 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18769 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18771 (gen_comm_data): Emit architectural setting of arch_prof.
18772 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18774 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18775 (armv8-m.base, armv8-m.main): Likewise.
18776 * arm-protos.h (arm_build_target): Add profile field.
18777 (arch_option): Likewise.
18778 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18780 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18781 arm_active_target.profile.
18783 2017-06-28 Richard Biener <rguenther@suse.de>
18785 PR middle-end/81227
18786 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18787 TYPE_OVERFLOW_WRAPS.
18788 * match.pd (negate_expr_p): Likewise.
18789 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18790 fold_build2, not fold_binary.
18792 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18794 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18795 Convert memory address to Pmode.
18796 (aarch64_print_operand): Assert MEM operands are always Pmode.
18798 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18801 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18802 Remove redundant if.
18803 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18804 * config/arm/aarch-common-protos.h
18805 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18806 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18808 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18811 * multiple_target.c (create_dispatcher_calls): Set the default
18812 clone to be static, not public.
18814 2017-06-28 Richard Biener <rguenther@suse.de>
18816 * tree-vect-loop.c (vectorizable_reduction): Move special
18817 cond reduction IV var creation ...
18818 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18819 parameter. Use STMT_VINFO_VECTYPE.
18820 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18823 2017-06-28 Martin Liska <mliska@suse.cz>
18826 * ipa-visibility.c (non_local_p): Handle visibility.
18828 2017-06-28 Martin Liska <mliska@suse.cz>
18831 * common.opt: Add IntegerRange to various options.
18832 * opt-functions.awk (integer_range_info): New function.
18833 * optc-gen.awk: Add integer_range_info to cl_options struct.
18834 * opts-common.c (decode_cmdline_option): Handle
18835 CL_ERR_INT_RANGE_ARG.
18836 (cmdline_handle_error): Likewise.
18837 * opts.c (print_filtered_help): Show valid interval in
18838 when --help is provided.
18839 * opts.h (struct cl_option): Add range_min and range_max fields.
18840 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18842 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18844 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18845 (x * C EQ/NE y * C): New transformation.
18847 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18849 * genmultilib (combination_space): Accept '+' in option names.
18851 2017-06-28 Martin Liska <mliska@suse.cz>
18854 * asan.c (instrument_derefs): Bail out inner references
18855 that are hard register variables.
18857 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18860 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18861 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18863 2017-06-28 Richard Biener <rguenther@suse.de>
18865 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18866 (vect_get_slp_defs): Adjust.
18867 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18869 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18871 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18872 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18873 (vectorizable_reduction): Adjust.
18874 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18876 (vect_get_slp_defs): Likewise.
18877 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18878 (vectorizable_bswap): Adjust.
18879 (vectorizable_call): Likewise.
18880 (vectorizable_conversion): Likewise.
18881 (vectorizable_assignment): Likewise.
18882 (vectorizable_shift): Likewise.
18883 (vectorizable_operation): Likewise.
18884 (vectorizable_store): Likewise.
18885 (vectorizable_condition): Likewise.
18886 (vectorizable_comparison): Likewise.
18888 2017-06-28 Michael Collison <michael.collison@arm.com>
18891 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18893 (arm_gen_movmemqi): Removed unused variable 'i'.
18894 Convert 'for' loop into 'while' loop.
18895 (arm_expand_prologue): Remove last unnecessary set of insn.
18896 (thumb_pop): Remove unused variable 'pushed_words'.
18897 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18899 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18901 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18902 * config/s390/s390-protos.h: Add prototype of
18903 s390_rel_address_ok_p.
18904 * config/s390/s390.c (s390_got_symbol): New function.
18905 (s390_rel_address_ok_p): New function.
18906 (legitimize_pic_address): Use s390_rel_address_ok_p.
18907 (s390_load_got): Use s390_got_symbol.
18908 (s390_option_override): Issue error if
18909 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18910 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18912 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18914 2017-06-27 Andrew Pinski <apinski@cavium.com>
18916 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18917 (X * copysign (1.0, X)): New pattern.
18918 (X * copysign (1.0, -X)): New pattern.
18919 (copysign (-1.0, CST)): New pattern.
18921 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18923 * genmultilib (combination_space): Remove variable.
18924 Validate reuse rules against regular expression for any sequence
18925 of multilib options in any order.
18927 2017-06-27 Michael Collison <michael.collison@arm.com>
18929 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18930 call aarch64_split_simd_combine.
18931 * (aarch64_combine_internal<mode>): Delete pattern.
18932 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18933 Allow register and subreg operands.
18935 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18937 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18938 specific need, just fallback on defaults.
18939 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18941 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18942 Olivier Hainque <hainque@adacore.com>
18944 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18946 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18947 targets. Pick a default if no particular attempt applied.
18948 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18951 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18953 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18954 (x86_64-wrs-vxworks7): Likewise.
18956 2017-06-27 Marek Polacek <polacek@redhat.com>
18959 * ubsan.c (instrument_null): Check get_base_address's result for null.
18961 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18963 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18965 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18967 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18968 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18969 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18970 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18971 New function types.
18972 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18973 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18974 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18975 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18976 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18977 BUILT_IN_FEUPDATEENV): New builtins.
18978 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18979 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18980 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18981 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18983 (builtin_structptr_types): Adjust size.
18984 * tree.c (builtin_structptr_types): Add four entries.
18986 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18987 Olivier Hainque <hainque@adacore.com>
18989 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18990 (TLS_SYM): New local macro, forcing reference to __tls__ on
18991 link command lines for VxWorks 7 RTPs, triggering initialization
18993 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18994 OS features TLS support, true for RTPs on VxWorks 7.
18995 * config/vxworks.c (vxworks_override_options): Setup emutls
18998 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19000 * predict.c (test_prediction_value_range): Use -1U instead of -1
19001 to avoid narrowing conversion warning.
19002 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19003 to avoid narrowing conversion warning.
19004 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19006 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19008 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19010 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19011 64bit configurations.
19012 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19013 (SIZE_TYPE): Likewise.
19014 * config/vxworks.c (vxworks_emutls_var_fields): Use
19015 long_unsigned_type_node instead of unsigned_type_node as the offset
19016 field type, which is "pointer" mode in emutls.c.
19018 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19021 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19023 PR middle-end/81207
19024 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19025 gimple_vuse copying separately from gimple_vdef copying.
19027 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19029 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19032 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19033 Olivier Hainque <hainque@adacore.com>
19035 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19036 for all vxworks7 targets.
19037 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19038 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19039 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19040 variations for VX6/VX7 and 32/64bits later on in ...
19041 (VXWORKS_LIB_SPEC): Leverage new macros.
19042 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19043 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19045 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19047 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19048 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19050 2017-06-26 Carl Love <cel@us.ibm.com>
19052 * config/rs6000/rs6000-c.c: Add support for built-in functions
19053 vector bool char vec_reve (vector bool char);
19054 vector signed char vec_reve (vector signed char);
19055 vector unsigned char vec_reve (vector unsigned char);
19056 vector bool int vec_reve (vector bool int);
19057 vector signed int vec_reve (vector signed int);
19058 vector unsigned int vec_reve (vector unsigned int);
19059 vector bool long long vec_reve (vector bool long long);
19060 vector signed long long vec_reve (vector signed long long);
19061 vector unsigned long long vec_reve (vector unsigned long long);
19062 vector bool short vec_reve (vector bool short);
19063 vector signed short vec_reve (vector signed short);
19064 vector double vec_reve (vector double);
19065 vector float vec_reve (vector float);
19066 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19067 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19068 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19069 (altivec_vreve): New pattern.
19070 * config/rs6000/altivec.h (vec_reve): New define.
19071 * doc/extend.texi (vec_rev): Update the built-in documentation file
19072 for the new built-in functions.
19074 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19076 PR tree-optimization/71815
19077 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19079 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19081 (slsr_process_phi): Likewise.
19082 (replace_uncond_cands_and_profitable_phis): Don't replace a
19083 multiply candidate with a stride of 1 (copy or cast).
19084 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19086 (lowest_cost_path): Likewise.
19087 (total_savings): Likewise.
19089 2017-06-26 Richard Biener <rguenther@suse.de>
19092 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19093 Use def_builtin_pure for all gather builtins.
19095 2017-06-26 Richard Biener <rguenther@suse.de>
19097 PR tree-optimization/81203
19098 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19099 non-dominating BBs.
19101 2017-06-26 Marek Polacek <polacek@redhat.com>
19104 * doc/invoke.texi: Document -Wmultistatement-macros.
19106 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19108 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19109 arm_neon_ok_no_float_abi effective target.
19111 2017-06-26 Richard Biener <rguenther@suse.de>
19113 PR tree-optimization/80928
19114 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19115 (copy_bbs): Set BB_DUPLICATED flag early.
19116 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19118 (execute_on_shrinking_pred): Likewise.
19119 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19120 BB_DUPLICATED blocks.
19121 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19122 iterate over all PHIs considering removal of *gsi.
19124 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19126 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19129 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19131 * config/rs6000/rs6000-string.c: (expand_block_clear,
19132 do_load_for_compare, select_block_compare_mode,
19133 compute_current_alignment, expand_block_compare,
19134 expand_strncmp_align_check, expand_strn_compare,
19135 expand_block_move, rs6000_output_load_multiple)
19136 Move functions related to string/block move/compare
19137 to a separate file.
19138 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19139 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19140 for this function which is now used in two files.
19141 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19142 * config.gcc: Add rs6000-string.o to extra_objs for
19143 targets powerpc*-*-* and rs6000*-*-*.
19145 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19148 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19149 32-bit, since indexed is not valid for DImode.
19150 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19151 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19152 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19153 (define_peephole2 for Altivec d-form store): Likewise.
19156 * multiple_target.c (create_dispatcher_calls): Only create the
19157 dispatcher call if the function is the default clone of a
19158 versioned function.
19160 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19162 PR middle-end/80902
19163 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19164 a call, force the call to not be a tail call.
19166 2017-06-23 Jeff Law <law@redhat.com>
19168 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19171 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19173 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19174 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19175 handling for early expansion of vector shifts (sl,sr,sra,rl).
19176 (builtin_function_type): Add vector shift right instructions
19177 to the unsigned argument list.
19179 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19181 rtl-optimizatoin/79286
19182 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19183 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19184 trap. PIC register plus a const unspec without offset can never trap.
19186 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19188 * tree.h (builtin_structptr_type): New type.
19189 (builtin_structptr_types): Declare new array.
19190 * tree.c (builtin_structptr_types): New array.
19191 (free_lang_data, build_common_tree_nodes): Use it.
19193 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19196 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19199 2017-06-22 Matt Turner <mattst88@gmail.com>
19201 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19202 Lake models to skylake case. Assume skylake for unknown
19203 models with clflushopt.
19205 2017-06-22 Jeff Law <law@redhat.com>
19207 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19208 frame sizes that do not satisfy aarch64_uimm12_shift.
19210 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19212 * profile-count.h (apply_probability,
19213 apply_scale, probability_in): Fix checks for zero.
19215 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19217 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19218 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19220 2016-06-22 Richard Biener <rguenther@suse.de>
19222 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19223 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19224 REDUC_MAX_EXPR support.
19225 (vectorizable_reduction): Likewise.
19226 (vect_create_epilog_for_reduction): Likewise.
19228 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19230 * match.pd (A / (1 << B) -> A >> B): New.
19231 * generic-match-head.c: Include optabs-tree.h.
19232 * gimple-match-head.c: Likewise.
19233 * optabs-tree.h (target_supports_op_p): New.
19234 * optabs-tree.c (target_supports_op_p): New.
19236 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19238 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19239 $gcc_cv_ld --help output.
19240 (gcc_cv_ld_demangle): Likewise.
19241 (gcc_cv_ld_eh_frame_hdr): Likewise.
19242 (gcc_cv_ld_pie): Likewise.
19243 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19244 (gcc_cv_ld_buildid): Likewise.
19245 (gcc_cv_ld_sysroot): Likewise.
19246 (ld_bndplt_support): Likewise.
19247 (ld_pushpopstate_support): Likewise.
19248 * configure: Regenerate.
19249 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19251 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19254 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19255 operands indexes to avoid gap between operands and match_dups.
19257 2017-06-21 Andrew Pinski <apinski@cavium.com>
19259 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19260 Increment Arith_shift and Arith_shift_reg by 1.
19261 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19263 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19264 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19265 (aarch64_strip_extend): Add new argument and test for it.
19266 (aarch64_cheap_mult_shift_p): New function.
19267 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19268 add a cost if it is true.
19269 Update calls to aarch64_strip_extend.
19270 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19272 2017-06-21 Andrew Pinski <apinski@cavium.com>
19274 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19276 (thunderxt88): Likewise.
19277 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19278 (thunderx_prefetch_tune): New variable.
19279 (thunderx2t99_prefetch_tune): Update for the correct values.
19280 (thunderxt88_tunings): New variable.
19281 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19282 generic_prefetch_tune.
19283 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19285 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19287 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19288 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19289 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19290 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19291 (aarch64_atomic_cas<mode>, GPI): Likewise.
19293 2017-06-21 Martin Liska <mliska@suse.cz>
19295 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19296 statements on cold and hot labels.
19297 * predict.c (tree_estimate_probability_bb): Remove the
19298 prediction from this place.
19300 2017-06-21 Martin Liska <mliska@suse.cz>
19302 PR tree-optimization/79489
19303 * gimplify.c (maybe_add_early_return_predict_stmt): New
19305 (gimplify_return_expr): Call the function.
19306 * predict.c (tree_estimate_probability_bb): Remove handling
19308 * predict.def: Update comment about early return predictor.
19309 * gimple-predict.h (is_gimple_predict): New function.
19310 * predict.def: Change default value of early return to 66.
19311 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19313 * passes.def: Put pass_strip_predict_hints to the beginning of
19316 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19318 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19319 FUNCTION_DECL declarations.
19320 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19322 (dwaf2out_decl): Likewise.
19323 * godump.c (go_early_global_decl): Skip call to the real debug hook
19324 for FUNCTION_DECL declarations.
19325 * passes.c (rest_of_decl_compilation): Skip call to the
19326 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19327 -fdump-go-spec is passed.
19329 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19331 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19332 Reorder for compactness.
19333 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19334 (def_builtin_pure, def_builtin_pure2): New functions.
19335 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19337 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19339 * match.pd (nop_convert): New predicate.
19340 ((A +- CST1) +- CST2): Allow some NOP conversions.
19342 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19345 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19346 with ctors/dtors if GOVD_SHARED is set.
19348 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19350 * config/aarch64/aarch64.md (movti_aarch64):
19351 Emit mov rather than orr.
19352 (movtf_aarch64): Likewise.
19353 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19354 Emit mov rather than orr.
19356 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19358 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19359 Swap alternatives, make integer dup more expensive.
19361 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19363 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19364 Return true for non-tls symbols.
19366 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19368 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19369 (cortex-a75): Likewise.
19370 (cortex-a75.cortex-a55): Likewise.
19371 * config/aarch64/aarch64-tune.md: Regenerate.
19372 * doc/invoke.texi (-mtune): Document new values for -mtune.
19374 2017-06-21 Tom de Vries <tom@codesourcery.com>
19376 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19377 stack_size feature.
19378 (Effective-Target Keywords, Other attributes): Suggest using
19379 dg-add-options stack_size feature to get stack limit in stack_size
19380 effective target documentation.
19382 2017-06-21 Julian Brown <julian@codesourcery.com>
19383 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19385 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19386 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19387 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19389 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19390 attribute type list for neon_multiply.
19391 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19392 attribute type list for neon_multiply.
19393 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19394 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19395 attribute type list for neon_multiply.
19396 * config/arm/types.md (crypto_pmull): Add.
19397 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19398 attribute type list.
19400 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19402 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19405 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19407 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19408 to make sure not to dereference a NULL cost_classes_ptr pointer.
19410 2017-06-20 Carl Love <cel@us.ibm.com>
19412 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19413 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19414 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19415 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19416 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19417 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19418 VMULOSW): New enum "unspec" values.
19419 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19420 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19421 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19422 altivec_vmulosw): New patterns.
19423 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19424 VMULOSW): Add definitions.
19426 2017-06-20 Julia Koval <julia.koval@intel.com>
19428 * config/i386/i386.c: Fix rounding expand for new pattern.
19429 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19431 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19433 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19434 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19436 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19438 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19441 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19443 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19444 architecture, then by implementer ID.
19445 * config/aarch64/aarch64-tune.md: Regenerate.
19447 2017-06-20 Richard Biener <rguenther@suse.de>
19449 PR middle-end/81097
19450 * fold-const.c (split_tree): Fold to type before negating.
19452 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19454 * diagnostic-show-locus.c
19455 (selftest::test_fixit_deletion_affecting_newline): New function.
19456 (selftest::diagnostic_show_locus_c_tests): Call it.
19458 2017-06-20 Andreas Schwab <schwab@suse.de>
19461 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19464 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19466 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19467 __ARM_FEATURE_COPROC according to support.
19469 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19471 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19472 Rewritten to avoid overflow for > 32-bit pointers.
19475 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19476 by removing enum keyword.
19477 (ubsan_type_descriptor): Likewise. Formatting fix.
19480 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19481 splitter): Require TARGET_SSE2 in the condition.
19483 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19486 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19487 for doing vector set of SFmode on ISA 3.0.
19488 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19489 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19491 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19492 SFmode value into a V4SF variable that was extracted from another
19493 V4SF variable without converting the element to double precision
19494 and back to single precision vector format.
19495 (vsx_insert_extract_v4sf_p9_2): Likewise.
19497 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19499 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19500 in UWHI to avoid undefined overflow.
19503 * ubsan.h (enum ubsan_encode_value_phase): New.
19504 (ubsan_encode_value): Change second argument to
19505 enum ubsan_encode_value_phase with default value of
19506 UBSAN_ENCODE_VALUE_GENERIC.
19507 * ubsan.c (ubsan_encode_value): Change second argument to
19508 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19509 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19510 create_tmp_var_raw instead of create_tmp_var and use a
19512 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19513 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19514 ubsan_encode_value callers.
19517 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19518 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19519 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19521 2017-06-19 Richard Biener <rguenther@suse.de>
19523 PR middle-end/81118
19524 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19525 estimates if we changed anything.
19527 2017-06-19 Richard Biener <rguenther@suse.de>
19529 PR tree-optimization/80887
19530 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19531 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19532 simplified lookups, then reset mprts_hook.
19533 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19535 (try_to_simplify): Likewise.
19537 2017-06-19 Martin Liska <mliska@suse.cz>
19540 * gimplify.c (gimplify_switch_expr):
19541 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19543 2017-06-19 Martin Liska <mliska@suse.cz>
19545 * doc/install.texi: Document that PGO runs in 4 stages.
19547 2017-06-19 Martin Liska <mliska@suse.cz>
19550 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19551 to dispatcher function name.
19552 * multiple_target.c (replace_function_decl): New function.
19553 (create_dispatcher_calls): Redirect both edges and references.
19555 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19557 * profile-count.c (profile_count::dump): Dump quality.
19558 (profile_count::differs_from_p): Update for unsigned val.
19559 * profile-count.h (profile_count_quality): New enum.
19560 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19562 2017-06-19 Richard Biener <rguenther@suse.de>
19564 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19565 struct function as arg.
19566 (estimate_numbers_of_iterations): Export overload with loop arg.
19567 (free_numbers_of_iterations_estimates_loop): Use an overload of
19568 free_numbers_of_iterations_estimates instead.
19569 * tree-cfg.c (remove_bb): Adjust.
19570 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19571 * tree-parloops.c (gen_parallel_loop): Likewise.
19572 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19574 (tree_unroll_loops_completely): Likewise.
19575 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19576 Use an overload instead and export.
19577 (estimated_loop_iterations): Adjust.
19578 (max_loop_iterations): Likewise.
19579 (likely_max_loop_iterations): Likewise.
19580 (estimate_numbers_of_iterations): Take struct function as arg
19582 (loop_exits_before_overflow): Adjust.
19583 (free_numbers_of_iterations_estimates_loop): Use an overload.
19584 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19585 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19587 2017-06-19 Richard Biener <rguenther@suse.de>
19590 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19591 RANGE_EXPR conservatively.
19593 2017-06-16 Carl Love <cel@us.ibm.com>
19595 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19596 definitions for vec_float, vec_float2, vec_floato,
19597 vec_floate built-ins.
19598 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19599 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19601 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19602 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19603 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19604 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19605 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19606 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19607 vec_floato): Add builtin defines.
19608 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19609 Update the built-in documentation file for the new built-in
19612 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19614 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19615 (mthumb): Mark as the negative of -marm.
19617 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19619 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19621 (ARM Options, -mtune): Document that this accepts the same
19622 extension options as -mcpu.
19623 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19625 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19627 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19628 permitted extensions.
19630 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19632 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19633 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19634 (armv8-m.main): Add option +nodsp.
19635 * config/arm/arm-cpu-cdata.h: Regenerated.
19637 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19639 * config/arm/t-fuchsia: New file.
19640 * config.gcc (arm*-*-fuchsia*): Use it.
19642 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19644 * config/arm/t-symbian: Rewrite for new option infrastructure.
19646 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19648 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19649 (MULTILIB_REQUIRED): Likewise.
19651 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19653 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19654 (MULTILIB_RESUE): Likewise.
19655 (MULTILIB_MATCHES): Likewise.
19656 (MULTLIB_REQUIRED): Likewise.
19658 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19660 * config/arm/t-rtems: Rewrite for new option framework.
19662 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19664 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19665 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19666 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19667 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19668 * config/arm/t-multilib: ... here.
19669 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19670 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19671 armv7-a and armv8*-a when A-profile libraries have not been built.
19672 * config/arm/t-rmprofile: Rewrite.
19674 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19676 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19677 with a backslash. Remove the backslash after substituting unescaped
19679 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19681 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19683 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19684 the driver to pass through the default setting of -mfloat-abi.
19685 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19687 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19688 (all_feat_combs): New rule.
19689 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19691 * config/arm/t-aprofile: Rewrite.
19693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19695 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19696 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19697 fpu is not specified by the user/command-line.
19698 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19699 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19700 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19701 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19702 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19703 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19705 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19707 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19708 * config/arm/t-arm-elf: Rewritten.
19710 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19712 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19713 have some floating-point instructions.
19714 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19715 (TARGET_MAYBE_HARD_FLOAT): New macro.
19716 * config/arm/arm-builtins.c (arm_init_builtins): Use
19717 TARGET_MAYBE_HARD_FLOAT.
19718 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19720 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19722 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19723 (configargs.h): Include it.
19724 (arm_print_hint_for_fpu_option): New function.
19725 (arm_parse_fpu_option): New function.
19726 (candidate_extension): New class.
19727 (arm_canon_for_multilib): New function.
19728 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19729 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19730 (ARCH_CANONICAL_SPECS): New macro.
19731 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19733 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19735 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19736 are set after handling multilib fragments. Set target_cpu_default2
19739 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19741 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19743 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19744 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19746 (check_arch): Likewise.
19747 * config/arm/arm.c (arm_configure_build_target): Handle
19748 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19749 options in the default.
19751 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19753 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19754 when an option is an alias of another.
19755 * config/arm/parsecpu.awk (optalias): New parser token.
19756 (gen_comm_data): Mark non-alias options as such. Emit entries
19757 for extension aliases.
19758 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19759 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19760 (armv6kz, armv6zk, armv6t2): Likewise.
19761 (armv7): Make vfpv3-d16 an alias.
19762 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19764 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19765 Sort in canonical order.
19766 (armv8-a): Sort in canonical order.
19767 (armv8.1-a, armv8.2-a): Likewise.
19768 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19770 (cortex-a9): Sort in canonical order.
19771 * config/arm/arm.c (selftests.h): Include it.
19772 (arm_test_cpu_arch_data): New function.
19773 (arm_run_self_tests): New function.
19774 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19775 (targetm): Move declaration to the end of the file.
19776 * arm-cpu-cdata.h: Regenerated.
19778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19780 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19781 call to target_mode_check describing the type of option passed.
19782 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19783 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19784 arm_parse_cpu_option_name to match parameters against list of
19786 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19787 arm_arch_core_flags data structure.
19788 * config/arm/arm-cpu_cdata.h: Regenerated.
19790 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19792 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19794 (arm_print_hint_for_cpu_option): Likewise.
19795 (arm_print_hint_for_arch_option): Likewise.
19796 (arm_parse_cpu_option_name): Likewise.
19797 (arm_parse_arch_option_name): Likewise.
19798 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19799 of entries in the all_fpus list.
19800 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19801 (arm_parse_cpu_option_name): Declare.
19802 (arm_parse_arch_option_name): Declare.
19803 (arm_parse_option_features): Declare.
19804 (arm_intialize_isa): Declare.
19805 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19807 (gen_comm_data): ... here. Make definitions non-static.
19808 * config/arm/arm-cpu-data.h: Regenerated.
19809 * config/arm/arm-cpu-cdata.h: Regenerated.
19811 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19813 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19814 (cpu_arch_extension): New structure.
19815 (cpu_arch_option, arch_option, cpu_option): New structures.
19816 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19817 architecture types.
19818 (gen_data): Generate new format data tables.
19819 * config/arm/arm.c (cpu_tune): New structure.
19820 (cpu_option, processors): Delete.
19821 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19822 (arm_print_hint_for_cpu_option): ... this and ...
19823 (arm_print_hint_for_arch_option): ... this.
19824 (arm_parse_arch_cpu_name): Delete. Replace with ...
19825 (arm_parse_cpu_option_name): ... this and ...
19826 (arm_parse_arch_option_name): ... this.
19827 (arm_unrecognized_feature): Change type of target parameter to
19829 (arm_parse_arch_cpu_features): Delete. Replace with ...
19830 (arm_parse_option_features): ... this.
19831 (arm_configure_build_target): Rework to use new configuration data
19833 (arm_print_tune_info): Rework for new configuration data tables.
19834 * config/arm/arm-cpu-data.h: Regenerated.
19835 * config/arm/arm-cpu.h: Regenerated.
19837 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19839 * Makefile.in (OBJS): Move sbitmap.o from here ...
19840 (OBJS-libcommon): ... to here.
19842 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19844 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19845 (ISA_ALL_CRYPTO): New macro.
19846 (ISA_ALL_SIMD): New macro
19847 (ISA_ALL_FP): New macro.
19848 * config/arm/arm.c (fpu_bitlist): Update initializer.
19849 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19851 (arm9e): Add fpu. Add option for nofp
19852 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19853 (arm926ej-s, arm1026ej-s): Likewise.
19854 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19855 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19856 neon-fp16, neon-vfpv4, nofp and nosimd.
19857 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19858 (cortex-a8): Add fpu. Add option for nofp.
19859 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19860 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19861 (cortex-r4f): Add fpu.
19862 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19863 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19865 (cortex-r8): Likewise.
19866 (cortex-m4): Add fpu. Add option for nofp.
19867 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19868 (cortex-a17.cortex-a7): Likewise.
19869 (cortex-a32): Add fpu. Add options for crypto and nofp.
19870 (cortex-a35, cortex-a53): Likewise.
19871 (cortex-a57): Add fpu. Add option for crypto.
19872 (cortex-a72, cortex-a73): Likewise.
19873 (exynos-m1): Likewise.
19874 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19875 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19876 (cortex-m33): Add fpu. Add option for nofp.
19877 * config/arm/arm-cpu-cdata.h: Regenerated
19878 * config/arm/arm-cpu-data.h: Regenerated.
19880 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19882 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19883 (armv5te, armv5tej): Likewise.
19884 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19885 (armv7): Add options fp and vfpv3-d16.
19886 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19887 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19889 (armv7ve): Likewise.
19890 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19891 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19892 (armv8-a): Add nocrypto option.
19893 (armv8.1-a, armv8.2-a): Likewise.
19894 (armv8-m.main): add options fp, fp.dp and nofp.
19896 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19898 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19900 (armv8-a+crc): Delete.
19901 (armv8.1-a): Add options simd, crypto and nofp.
19902 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19903 (armv8.2-a+fp16): Delete.
19904 (armv8-m.main): Add option dsp.
19905 (armv8-m.main+dsp): Delete.
19906 (cortex-a8): Add fpu. Add nofp option.
19907 (cortex-a9): Add fpu. Add nofp and nosimd options.
19908 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19909 link to main cpu and architecture data structures.
19910 (gen_comm_data): Only put isa attributes from the main architecture
19912 (option): New statement for architecture and CPU entries.
19913 * arm.c (struct cpu_option): New structure.
19914 (struct processors): Add entry for options.
19915 (arm_unrecognized_feature): New function.
19916 (arm_parse_arch_cpu_name): Ignore any characters after the first
19918 (arm_parse_arch_cpu_feature): New function.
19919 (arm_configure_build_target): Separate out any CPU and architecture
19920 features and parse separately. Don't error out if -mfpu=auto is
19921 used with only an architecture string.
19922 (arm_print_asm_arch_directives): New function.
19923 (arm_file_start): Call it.
19924 * config/arm/arm-cpu-cdata.h: Regenerated.
19925 * config/arm/arm-cpu-data.h: Likewise.
19926 * config/arm/arm-tables.opt: Likewise.
19928 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19930 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19931 assembler when it is not -mfpu=auto.
19933 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19935 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19936 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19937 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19938 (ASM_CPU_SPEC): Rewrite.
19939 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19940 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19941 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19942 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19943 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19944 copied string is NUL-terminated. Also strip any characters prefixed
19946 (arm_rewrite_selected_arch): New function.
19947 (arm_rewrite_march): New function.
19949 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19951 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19952 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19953 (march, mcpu, mtune): Convert to string-based options.
19954 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19955 (arm_parse_arch_cpu_name): New function.
19956 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19957 identify selected architecture or CPU.
19958 (arm_option_save): New function.
19959 (TARGET_OPTION_SAVE): Redefine.
19960 (arm_option_restore): Restore string options.
19961 (arm_option_print): Print string options.
19963 2017-06-16 Martin Sebor <msebor@redhat.com>
19965 PR tree-optimization/80933
19966 PR tree-optimization/80934
19967 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19968 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19969 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19970 (gimple_fold_builtin): Call them.
19972 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19974 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19975 as unlikely; update profile.
19977 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19979 * predict.c (force_edge_cold): Handle declaring edges impossible
19982 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19984 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19986 (try_unroll_loop_completely): Fix reporting.
19988 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19990 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19992 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19995 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19996 if given a non-constant argument for an intrinsic which requires a
19999 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20001 * profile.c (compare_freqs): New function.
20002 (branch_prob): Sort edge list.
20003 (find_spanning_tree): Assume that the list is priority sorted.
20005 2017-06-16 Richard Biener <rguenther@suse.de>
20007 PR tree-optimization/81090
20008 * passes.def (pass_record_bounds): Remove.
20009 * tree-pass.h (make_pass_record_bounds): Likewise.
20010 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20011 make_pass_record_bounds): Likewise.
20012 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20013 not free niter estimates at the beginning but at the end.
20014 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20016 2017-06-16 Richard Biener <rguenther@suse.de>
20018 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20019 initializer to workaround ICE in host GCC 4.8.
20021 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20023 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20025 (clone_inlined_nodes): Update.
20027 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20029 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20030 prefetch settings, and enable prefetching by default at -O3.
20032 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20034 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20035 Set flag_prefetch_loop_arrays according to tuning data.
20037 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20039 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20040 New tune structure.
20041 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20042 [Unrelated to main purpose of the patch] Place the pointer field last
20043 to enable type checking errors when tune structure are wrongly merged.
20044 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20045 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20046 (thunderx2t99_prefetch_tune): New tune constants.
20047 (tune_params *_tunings): Update all tunings (no functional change).
20048 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20049 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20050 from tunings structures.
20052 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20055 * ubsan.c (instrument_null): Add T argument, use it instead
20056 of computing it based on IS_LHS.
20057 (instrument_object_size): Likewise.
20058 (pass_ubsan::execute): Adjust instrument_null and
20059 instrument_object_size callers to pass gimple_get_lhs or
20060 gimple_assign_rhs1 result to it. Use instrument_null instead of
20061 calling get_base_address and instrument_mem_ref. Handle
20062 aggregate call arguments for object-size sanitization.
20064 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20066 PR tree-optimization/81089
20067 * tree-vrp.c (is_masked_range_test): Validate operands of
20070 2017-06-15 Martin Sebor <msebor@redhat.com>
20073 * dumpfile.c (dump_register): Avoid calling memset to initialize
20074 a class with a default ctor.
20075 * gcc.c (struct compiler): Remove const qualification.
20076 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20077 * hash-table.h: Ditto.
20078 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20080 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20081 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20083 * params.h (struct param_info): Make struct members non-const.
20084 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20085 with default initialization.
20086 * vec.h (vec_copy_construct, vec_default_construct): New helper
20088 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20089 with vec_copy_construct.
20090 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20091 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20092 * doc/invoke.texi (-Wclass-memaccess): Document.
20094 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20096 * emit-rtl.h (is_leaf): Update comment about local
20097 register allocator.
20099 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20102 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20103 for a variable to have a section before checking if the section has a
20105 Set section to.persistent if persistent attribute is set.
20106 Warn if .persistent attribute is used on an automatic variable.
20108 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20110 PR rtl-optimization/80474
20111 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20113 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20115 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20118 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20120 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20121 (casesi): Emit bounds checking as RTL.
20122 (casesi_internal_mips16_<mode>): Remove bounds checking.
20124 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20126 * config/xtensa/xtensa.c (xtensa_option_override): Append
20127 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20128 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20129 xtensa_doloop_hooks): Define unconditionally.
20130 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20132 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20133 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20134 for it in xtensa_option_override.
20135 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20136 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20138 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20140 * doc/cppopts.texi: Document '-' special value to -MF.
20142 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20144 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20145 (cortex_a53_fconst): Likewise.
20146 (cortex_a53_fpmul): Likewise.
20147 (cortex_a53_f_load_64): Likewise.
20148 (cortex_a53_f_load_many): Likewise.
20149 (cortex_a53_advsimd_alu): Likewise.
20150 (cortex_a53_advsimd_alu_q): Likewise.
20151 (cortex_a53_advsimd_mul): Likewise.
20152 (cortex_a53_advsimd_mul_q): Likewise.
20153 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20154 Add missing fmul, r2f_cvt and fconst cases.
20156 2017-06-14 Richard Biener <rguenther@suse.de>
20158 PR middle-end/81088
20159 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20161 (fold_binary_loc): When associating do not treat pre-existing
20162 TREE_OVERFLOW on literal constants as a reason to allow
20163 TREE_OVERFLOW on associated literal constants.
20165 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20167 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20168 (MASK_FEATURES): New macro.
20169 * config/sparc/sparc.c (sparc_option_override): Remove the special
20170 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20172 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20174 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20175 a division of 0 if non-call exceptions are enabled.
20177 2017-06-14 Andrew Pinski <apinski@cavium.com>
20178 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20181 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20182 Improve vector initialization code gen for only variable case.
20184 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20186 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20188 2017-06-14 Richard Biener <rguenther@suse.de>
20190 PR tree-optimization/81083
20191 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20194 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20196 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20197 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20198 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20199 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20200 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20201 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20203 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20205 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20206 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20208 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20210 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20212 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20214 * config/rs6000/t-rtems: Don't handle SPE.
20216 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20218 * config/rs6000/t-linux: Don't handle SPE.
20220 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20222 * config/rs6000/eabispe.h: Delete file.
20224 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20226 * config/rs6000/t-spe: Delete file.
20228 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20230 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20231 (rs6000_legitimate_offset_address_p): Return false for anything in
20232 V2SImode or V2SFmode.
20234 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20236 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20237 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20238 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20240 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20241 (rs6000_legitimate_offset_address_p): Ditto.
20242 (rs6000_emit_move): Ditto.
20243 (rs6000_init_builtins): Remove V4HI_type_node.
20245 2017-06-13 Martin Liska <mliska@suse.cz>
20248 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20249 (gate_asan): Likewise.
20250 * asan.h (asan_no_sanitize_address_p): Remove the function.
20251 (sanitize_flags_p): New function.
20252 * builtins.def: Fix coding style.
20253 * common.opt: Use renamed enum value.
20254 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20255 * doc/extend.texi: Document no_sanitize attribute.
20256 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20257 to SANITIZE_UNDEFINED_NONDEFAULT.
20258 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20259 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20260 Use sanitize_flags_p.
20261 * gimplify.c (gimplify_function_tree): Likewise.
20262 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20263 * opts.c (parse_no_sanitize_attribute): New function.
20264 (common_handle_option): Use renamed enum value.
20265 * opts.h (parse_no_sanitize_attribute): Declare.
20266 * tree.c (sanitize_flags_p): New function.
20267 * tree.h: Declared here.
20268 * tsan.c: Use sanitize_flags_p.
20269 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20270 (instrument_mem_ref): Likewise.
20271 (instrument_bool_enum_load): Likewise.
20272 (do_ubsan_in_current_function): Remove the function.
20273 (pass_ubsan::execute): Use sanitize_flags_p.
20274 * ubsan.h: Remove do_ubsan_in_current_function
20275 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20276 (dump_function_to_file): Use it here.
20278 2017-06-13 Martin Jambor <mjambor@suse.cz>
20280 PR tree-optimization/80803
20281 PR tree-optimization/81063
20282 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20283 (propagate_subaccesses_across_link): Enqueue subtree whenever
20284 necessary instead of relying on the caller.
20286 2017-06-13 Martin Jambor <mjambor@suse.cz>
20288 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20289 that have a first_link.
20290 (sort_and_splice_var_accesses): Do not check first_link before
20292 (subtree_mark_written_and_enqueue): Likewise.
20293 (propagate_all_subaccesses): Likewise and do not stop at first
20294 parent with a first_link.
20296 2017-06-13 Martin Jambor <mjambor@suse.cz>
20298 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20301 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20303 * match.pd: New pattern.
20305 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20307 * tree-vrp.c (is_masked_range_test): New function.
20308 (register_edge_assert_for): Determine ranges for
20311 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20313 PR tree-optimization/67328
20314 * fold-const.c (maskable_range_p): New function.
20315 (build_range_check): Generate bittests if possible.
20317 2017-06-13 Martin Liska <mliska@suse.cz>
20319 * gimple-pretty-print.c (dump_probability): Add new argument.
20320 (dump_edge_probability): Dump both probability and count.
20321 (dump_gimple_label): Likewise.
20322 (dump_gimple_bb_header): Likewise.
20324 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20327 * config/avr/avr-devices.c: Fix indentation.
20328 * config/avr/gen-avr-mmcu-specs.c: Dito.
20330 2017-06-13 Richard Biener <rguenther@suse.de>
20332 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20333 instead get vector type from stmt_info.
20334 (vectorizable_reduction): Adjust. Remove dead code.
20336 2017-06-13 Richard Biener <rguenther@suse.de>
20338 PR middle-end/81065
20339 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20340 case of C * (x * C2 + C3).
20341 (fold_addr_of_array_ref_difference): Properly fold index difference.
20343 2017-06-12 David S. Miller <davem@davemloft.net>
20346 * config/sparc/sparc.md (return expander): Emit frame blockage if
20347 function uses alloca.
20349 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20351 * combine.c (make_field_assignment): Check len rather than the mode
20352 precision when calling force_to_mode.
20354 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20356 Support multilibs and devices that see flash in RAM address range.
20359 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20360 (avr_mcu_t) <flash_pm_offset>: New field.
20361 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20362 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20363 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20364 (AVR_TINY_PM_OFFSET): Remove macro.
20365 * config/avr/avr.opt (-mshort-calls): New option.
20366 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20367 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20368 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20369 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20370 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20371 instead of avr_arch->have_jmp_call.
20372 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20373 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20374 avr_arch->flash_pm_offset to define.
20375 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20376 new field flash_pm_offset. Add entry for avrxmega3.
20377 (avr_texinfo): Add entry for avrxmega3.
20378 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20379 attiny212, attiny214,
20380 attiny412, attiny414, attiny416, attiny417,
20381 attiny814, attiny816, attiny817,
20382 attiny1614, attiny1616, attiny1617,
20383 attiny3214, attiny3216, attiny3217.
20384 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20385 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20386 (avr_print_operand_address) [AVR_TINY]: Same.
20387 (avr_asm_init_sections) <readonly_data_section>: Only patch
20388 callback if avr_arch->flash_pm_offset = 0.
20389 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20390 for rodata if avr_arch->flash_pm_offset != 0.
20391 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20392 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20393 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20394 (m_options): Append opt_rcall.
20395 (m_dirnames): Append dir_rcall.
20396 * config/avr/t-multilib: Regenerate.
20398 * configure.ac [target=avr]: Check whether avrxmega3 default
20399 linker description file works as needed.
20400 * configure: Regenerate.
20401 * doc/avr-mmcu.texi: Regenerate.
20402 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20403 <__AVR_ARCH__>: Document avrxmega3 and 103.
20404 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20405 <__AVR_SHORT_CALLS__>: Document it.
20406 <__AVR_PM_BASE_ADDRESS__>: Document it.
20407 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20408 (AVR Variable Attributes) <progmem>: Document this is
20409 not needed for avrxmega3.
20410 (AVR Named Address Spaces) <__flash>: Dito.
20412 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20414 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20416 2017-06-12 Doug Rupp <rupp@adacore.com>
20418 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20419 Append vxworks-stdint.h to the tm_file list.
20420 * config/vxworks-stdint.h: New file.
20422 2017-06-12 Martin Liska <mliska@suse.cz>
20424 PR tree-optimization/81041
20425 * tree-profile.c (gimple_gen_ic_func_profiler):
20426 Create an extra BB in profile-generate
20427 (gimple_gen_time_profiler): Likewise.
20429 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20431 PR tree-optimization/81003
20432 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20433 (update_range_test): Use it instead of force_gimple_operand_gsi.
20435 2017-06-12 Richard Biener <rguenther@suse.de>
20437 PR tree-optimization/81053
20438 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20439 with backedge value not defined in loop. Simplify def stmt
20442 2017-06-11 Tom de Vries <tom@codesourcery.com>
20445 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20447 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20448 nvptx_cannot_force_const_mem.
20450 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20452 * opts.c (finish_options): Move test for flag_split_stack after
20453 it has been initialized.
20455 2017-06-11 Jason Merrill <jason@redhat.com>
20457 * tree.h (id_equal): New.
20458 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20459 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20460 instead of strcmp of IDENTIFIER_POINTER.
20462 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20464 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20465 (mark_all_inlined_calls_cdtor): Fix formating.
20466 (inline_transform): Rescale profile before inlining.
20468 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20470 * cgraph.h (cgraph_edge::clone): Update prototype.
20471 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20472 (cgraph_node::create_clone): Update.
20473 (cgraph_node::create_version_clone): Update.
20474 * tree-inline.c (copy_bb): Update.
20475 (expand_call_inline): Update.
20477 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20479 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20480 factored out from ...
20481 (rs6000_emit_prologue): ... here.
20483 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20485 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20486 factored out from ...
20487 (rs6000_emit_prologue): ... here.
20489 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20491 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20493 (handle_missing_profiles): Fix computation of tp_first_run.
20494 (counts_to_freqs): Do not touch freqs when count is 0.
20496 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20498 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20501 2017-06-10 Tom de Vries <tom@codesourcery.com>
20503 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20504 attributes): Document signal effective target.
20506 2017-06-10 Tom de Vries <tom@codesourcery.com>
20508 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20509 Document effective target stack_size.
20511 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20513 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20514 to the edit_context if they can be auto-applied.
20516 2017-06-9 Ian Lance Taylor <iant@golang.org>
20518 * opts.c (finish_options): If -fsplit-stack, disable implicit
20519 -forder-blocks-and-partition.
20520 * doc/invoke.texi (Optimize Options): Document that when using
20521 -fsplit-stack -forder-blocks-and-partition is not implicitly
20524 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20526 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20527 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20528 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20529 * builtins.def (abort, trap, unreachable): Declare cold.
20530 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20531 * tree-core.h (ECF_COLD): New.
20532 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20533 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20535 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20537 * predict.c (unlikely_executed_stmt_p): Cleanup.
20539 2017-06-09 Richard Biener <rguenther@suse.de>
20541 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20542 model if the ref is always written to.
20544 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20546 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20548 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20550 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20553 2017-06-09 Tom de Vries <tom@codesourcery.com>
20556 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20557 "target cannot support label values" when encountering LABEL_REF.
20559 2017-06-09 Martin Liska <mliska@suse.cz>
20561 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20562 (gimple_gen_ic_func_profiler): Emit direct comparison
20563 of __gcov_indirect_call_callee with NULL.
20564 (gimple_gen_time_profiler): Change probability from
20565 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20567 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20569 * profile.c (edge_gcov_counts): Turn to pointer.
20570 (compute_branch_probabilities, compute_branch_probabilities): Update.
20571 (branch_prob): Do not clear edge_gcov_count.
20572 * profile.h (edge_gcov_counts): Turn to pointer.
20573 (edge_gcov_count): Update.
20575 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20577 * gimple.h (gimple_check_failed): Mark cold.
20579 2017-06-09 Richard Biener <rguenther@suse.de>
20581 PR tree-optimization/66623
20582 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20583 refactor check_reduction into two parts, properly computing
20584 whether we have to check reduction validity for outer loop
20587 2017-06-09 Richard Biener <rguenther@suse.de>
20589 PR tree-optimization/79483
20590 * graphite-scop-detection.c (order): New global.
20591 (get_order): Compute bb to order mapping that satisfies code
20592 generation constraints.
20593 (cmp_pbbs): New helper.
20594 (build_scops): Start domwalk at entry block, sort generated
20597 2017-06-09 Richard Biener <rguenther@suse.de>
20599 PR middle-end/81007
20600 * ipa-polymorphic-call.c
20601 (ipa_polymorphic_call_context::restrict_to_inner_class):
20602 Skip FIELD_DECLs with error_mark_node type.
20603 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20606 2017-06-09 Martin Liska <mliska@suse.cz>
20608 * predict.c (struct branch_predictor): New struct.
20609 (test_prediction_value_range): New test.
20610 (predict_c_tests): New function.
20611 * selftest-run-tests.c (selftest::run_tests): Run the function.
20612 * selftest.h: Declare new tests.
20614 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20617 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20618 gen_add3_insn did not fail.
20619 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20620 r0, construct that number in a temporary reg and add that reg to r0.
20621 If asked to put the result in r0 as well, fail.
20623 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20625 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20626 for early expansion of vec_eqv.
20628 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20630 PR middle-end/81005
20631 * ubsan.c (instrument_null): Avoid pointless code temporary.
20632 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20634 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20638 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20641 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20644 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20646 * predict.c (unlikely_executed_edge_p): Move ahead.
20647 (probably_never_executed_edge_p): Use it.
20649 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20651 PR middle-end/79988
20652 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20653 gimple_call_builtin_p call.
20655 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20657 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20658 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20659 rtl_check_failed_type2, rtl_check_failed_code1,
20660 rtl_check_failed_code2, rtl_check_failed_code_mode,
20661 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20662 rtvec_check_failed_bounds, rtl_check_failed_flag,
20663 _fatal_insn_not_found, _fatal_insn): Likewise.
20664 * tree.h (tree_contains_struct_check_failed,
20665 tree_check_failed, tree_not_check_failed,
20666 tree_class_check_failed, tree_range_check_failed,
20667 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20668 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20669 tree_operand_check_failed, omp_clause_check_failed,
20670 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20673 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20675 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20676 flag_branch_probabilities.
20677 * ipa-inline.c (edge_badness): Likewise.
20678 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20679 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20680 * predict.c (maybe_hot_frequency_p): Likewise.
20681 (probably_never_executed): Likewise.
20682 * sched-ebb.c (schedule_ebbs): Likewise.
20683 * sched-rgn.c (find_single_block_region): Likewise.
20684 * tracer.c (tail_duplicate): Likewise.
20686 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20688 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20689 longer requires x_flag_profile_use.
20691 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20693 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20694 instead of flag_reorder_blocks_and_partition.
20695 * dbxout.c (dbxout_function_end): Likewise.
20696 * dwarf2out.c (gen_subprogram_die): Likewise.
20697 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20698 * hw-doloop.c (reorg_loops): Likewise.
20699 * varasm.c (assemble_start_function,
20700 assemble_end_function): Likewise.
20701 (decide_function_section): Do not check for
20702 flag_reorder_blocks_and_partition.
20704 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20706 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20708 (chkp_get_hard_register_fake_addr_expr): Ditto.
20709 (chkp_build_addr_expr): Add check for hard reg case.
20710 (chkp_parse_array_and_component_ref): Ditto.
20711 (chkp_find_bounds_1): Ditto.
20712 (chkp_process_stmt): Don't generate bounds store for
20715 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20717 * predict.c (maybe_hot_bb_p): Do not check profile status.
20718 (maybe_hot_edge_p): Likewise.
20719 (probably_never_executed): Check for zero counts even if profile
20721 (unlikely_executed_edge_p): New function.
20722 (unlikely_executed_stmt_p): New function.
20723 (unlikely_executed_bb_p): New function.
20724 (set_even_probabilities): Use unlikely predicates.
20725 (combine_predictions_for_bb): Likewise.
20726 (predict_paths_for_bb): Likewise.
20727 (predict_paths_leading_to_edge): Likewise.
20728 (determine_unlikely_bbs): New function.
20729 (estimate_bb_frequencies): Use it.
20730 (compute_function_frequency): Use zero counts even if profile is
20732 * profile-count.h: Fix typo.
20734 2017-08-08 Julia Koval <julia.koval@intel.com>
20736 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20737 _mm512_mask_cvtsepi16_storeu_epi8,
20738 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20739 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20740 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20741 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20742 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20743 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20744 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20745 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20746 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20747 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20748 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20749 __builtin_ia32_pmovuswb256mem_mask,
20750 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20751 __builtin_ia32_pmovwb512mem_mask): New builtins.
20753 2017-08-08 Julia Koval <julia.koval@intel.com>
20755 PR target/73350,80862
20756 * config/i386/subst.md (round): Fix round pattern.
20757 * config/i386/i386.c (ix86_erase_embedded_rounding):
20758 Fix erasing rounding for the fixed pattern.
20760 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20762 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20764 2017-06-08 Martin Liska <mliska@suse.cz>
20766 PR gcov-profile/80911
20767 * gcov.c (block_info::block_info): New constructor.
20769 2017-06-07 Carl Love <cel@us.ibm.com>
20771 * config/rs6000/rs6000-c: The return type of the following
20772 built-in functions was implemented as int not long long. Fix sign
20773 of return value for the unsigned version of vec_mulo and vec_mule.
20774 vector unsigned long long vec_bperm (vector unsigned long long,
20775 vector unsigned char)
20776 vector signed long long vec_mule (vector signed int,
20778 vector unsigned long long vec_mule (vector unsigned int,
20779 vector unsigned int)
20780 vector signed long long vec_mulo (vector signed int,
20782 vector unsigned long long vec_mulo (vector unsigned int,
20783 vector unsigned int)
20784 * doc/extend.texi: Fix the documentation for the built-in
20787 2017-06-07 Carl Love <cel@us.ibm.com>
20790 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20793 2017-06-07 Carl Love <cel@us.ibm.com>
20795 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20796 support, Generate doublehv for signed int/float for BE case only.
20798 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20800 * doc/invoke.texi (mcx16): Rewrite.
20802 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20804 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20805 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20806 *mov<mode>_softfloat, and an anonymous splitter): Use
20807 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20809 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20811 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20813 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20814 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20815 (rs6000_debug_reg_global): Adjust.
20816 (rs6000_init_hard_regno_mode_ok): Adjust.
20817 (rs6000_dbx_register_number): Adjust.
20818 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20819 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20820 Remove SPE_ACC and SPEFSCR.
20821 (REG_ALLOC_ORDER): Ditto.
20822 (FRAME_POINTER_REGNUM): Change to 111.
20823 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20824 (REG_CLASS_NAMES): Ditto.
20825 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20826 (REGISTER_NAMES): Ditto.
20827 (ADDITIONAL_REG_NAMES): Ditto.
20828 (rs6000_reg_names): Ditto.
20829 * config/rs6000/rs6000.md: Renumber some register number
20832 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20834 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20836 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20837 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20839 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20840 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20841 Delete the SPE high registers.
20842 (REG_ALLOC_ORDER): Ditto.
20843 (enum reg_class): Remove SPE_HIGH_REGS.
20844 (REG_CLASS_NAMES): Ditto.
20845 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20846 (REGISTER_NAMES): Ditto.
20847 (rs6000_reg_names): Ditto.
20848 * doc/tm.texi.in: Remove SPE as example.
20849 * doc/tm.texi: Regenerate.
20851 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20853 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20854 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20855 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20856 * config/rs6000/rs6000.md (type): Remove "brinc".
20858 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20860 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20861 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20862 * config/rs6000/linuxspe.h: Delete file.
20863 * config/rs6000/rs6000.md: Don't include spe.md.
20864 * config/rs6000/spe.h: Delete file.
20865 * config/rs6000/spe.md: Delete file.
20866 * config/rs6000/t-rs6000: Remove spe.md.
20868 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20870 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20871 (reg_or_none500mem_operand): Delete.
20872 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20873 instead of reg_or_none500mem_operand.
20875 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20877 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20878 handling of SPE flags.
20879 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20881 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20883 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20885 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20886 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20887 paired_divv2sf3): Similar.
20888 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20889 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20890 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20892 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20893 Rename the paired_* instruction patterns.
20894 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20896 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20897 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20898 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20899 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20900 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20901 PAIRED_VECTOR_MODE.
20902 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20903 (spe_func_has_64bit_regs_p): Delete.
20904 (spe_expand_predicate_builtin): Delete.
20905 (spe_expand_evsel_builtin): Delete.
20906 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20907 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20908 (invalid_e500_subreg): Delete.
20909 (rs6000_legitimize_address): Always force_reg op2 as well, for
20910 paired single memory accesses.
20911 (rs6000_member_type_forces_blk): Delete.
20912 (rs6000_spe_function_arg): Delete.
20913 (rs6000_expand_unop_builtin): Delete SPE handling.
20914 (rs6000_expand_binop_builtin): Ditto.
20915 (spe_expand_stv_builtin): Delete.
20916 (bdesc_2arg_spe): Delete.
20917 (spe_expand_builtin): Delete.
20918 (spe_expand_predicate_builtin): Delete.
20919 (spe_expand_evsel_builtin): Delete.
20920 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20921 (spe_init_builtins): Delete.
20922 (spe_func_has_64bit_regs_p): Delete.
20923 (savres_routine_name): Delete "info" parameter. Adjust callers.
20924 (rs6000_emit_stack_reset): Ditto.
20925 (rs6000_dwarf_register_span): Delete.
20926 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20927 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20928 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20930 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20932 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20933 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20934 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20935 mulv2sf3, divv2sf3): Delete expanders.
20937 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20939 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20941 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20943 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20944 * config/rs6000/rs6000.c: Ditto.
20946 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20948 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20949 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20950 comparison_operator.
20952 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20954 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20955 * config/rs6000/rs6000.opt: Ditto.
20956 * config/rs6000/t-rtems: Ditto.
20958 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20960 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20961 TARGET_E500_SINGLE by 0, simplify.
20962 * config/rs6000/rs6000.c: Ditto.
20963 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20964 (spe_build_register_parallel): Delete.
20965 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20966 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20967 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20968 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20969 (E500_CONVERT): Delete.
20970 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20972 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20974 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20975 * config/rs6000/dfp.md: Ditto.
20976 (negdd2, *negdd2_fpr): Merge.
20977 (absdd2, *absdd2_fpr): Merge.
20978 (negtd2, *negtd2_fpr): Merge.
20979 (abstd2, *abstd2_fpr): Merge.
20980 * config/rs6000/e500.h: Delete file.
20981 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20982 TARGET_FPRS by 1 and simplify.
20983 * config/rs6000/rs6000-c.c: Ditto.
20984 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20985 TARGET_DF_SPE by 0.
20986 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20988 * config/rs6000/rs6000.md: Ditto.
20989 (floatdidf2, *floatdidf2_fpr): Merge.
20990 (move_from_CR_gt_bit): Delete.
20991 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20992 (E500_CR_IOR_COMPARE): Delete.
20993 (All patterns that require !TARGET_FPRS): Delete.
20994 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20996 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20998 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21000 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21002 * graphds.c (add_edge): Intitialize edge's attached data.
21003 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21004 pointer parameter. Call pointed function on each edge during
21005 graph traversing. Skip traversing the edge when the function
21007 (graphds_dfs, graphds_scc): Ditto.
21008 (for_each_edge): New parameter. Pass the new parameter to callback
21010 * graphds.h (skip_edge_callback): New function pointer type.
21011 (graphds_dfs, graphds_scc): New function pointer parameter.
21012 (graphds_edge_callback, for_each_edge): New parameter.
21014 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21016 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21017 out code checking if runtime alias check is possible to below ...
21018 Call the new function.
21019 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21020 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21022 2017-06-07 Marek Polacek <polacek@redhat.com>
21025 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21026 TYPE_OVERFLOW_WRAPS check.
21028 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21030 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21031 if versioning is required.
21032 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21033 peeling with the check for versioning.
21035 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21037 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21038 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21039 Set true to new parameter if new ssa variable is defined.
21040 (vect_gen_vector_loop_niters): Refactor. Set range information
21041 for the new vector loop bound variable.
21042 (vect_do_peeling): Ditto.
21044 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21046 * tree-affine.c (ssa.h): Include header file.
21047 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21048 has wrapping overflow behavior.
21050 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21052 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21054 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21056 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21057 (tree_to_aff_combination): ... here.
21059 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21061 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21062 reg_pressure model function.
21063 (ivopts_global_cost_for_size): Delete.
21064 (determine_set_costs, iv_ca_recount_cost): Call new model function
21065 ivopts_estimate_reg_pressure.
21067 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21069 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21070 expensive than udiv. Remove floating point cases from mod.
21072 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21074 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21075 Increase idiv cost.
21077 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21079 * config/aarch64/aarch64.md
21080 (copysignsf3): Fix mask generation.
21082 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21084 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21086 (class dump_manager): Add register_dumps method.
21087 * dumpfile.c: Include langhooks.h.
21088 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21089 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21090 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21091 (dump_manager::dump_register): Start with 512 entries instead of 32.
21092 (dump_manager::register_dumps): New method.
21093 * toplev.c (general_init): Instead of invoking register_dumps
21094 langhook, invoke register_dumps method on the dump manager.
21095 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21098 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21100 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21101 Say that # requires an associated define_split to exist, and that
21102 the define_split must be suitable for use after register allocation.
21104 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21106 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21107 (compute_outgoing_frequencies): Also initialize zero counts.
21108 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21109 around loops; preserve more of profile when nothing changes.
21111 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21113 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21115 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21116 * config/arm/arm-cpu-cdata.h: Regenerate.
21117 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21118 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21119 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21120 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21121 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21123 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21125 * config/arm/t-rmprofile: Likewise.
21126 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21128 2017-06-06 David S. Miller <davem@davemloft.net>
21131 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21132 blockage if function uses alloca.
21134 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21136 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21137 New "uid" fields to hold pretty-print IDs of group and ref.
21138 Memory references are now identified as <group_id>:<ref_id>
21139 instead of using [random] addresses.
21140 (dump_mem_details): Simplify, no functional change.
21141 (dump_mem_ref): Simplify and make output more concise.
21142 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21143 (find_or_create_group): Initialize group uid.
21144 (record_ref): Initialize ref uid. Improve debug output.
21145 (prune_group_by_reuse, should_issue_prefetch_p,)
21146 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21147 (mark_nontemporal_store, determine_loop_nest_reuse):
21148 Improve debug output.
21150 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21152 * dbgcnt.def (prefetch): New debug counter.
21153 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21154 (schedule_prefetches): Stop issueing prefetches if debug counter
21157 2017-06-06 Tom de Vries <tom@codesourcery.com>
21159 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21160 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21162 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21164 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21165 Use aarch64_reg_or_zero predicate for operand 4.
21166 (aarch64_compare_and_swap<mode> define_insn_and_split):
21167 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21168 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21170 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21172 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21173 (arm_compute_save_core_reg_mask): This.
21174 (thumb1_compute_save_reg_mask): Rename into ...
21175 (thumb1_compute_save_core_reg_mask): This.
21176 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21177 (arm_compute_frame_layout): Likewise.
21179 2017-06-06 Richard Biener <rguenther@suse.de>
21181 PR tree-optimization/80974
21182 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21183 keep or clear leaders SSA info.
21185 2017-06-06 Tom de Vries <tom@codesourcery.com>
21187 * config/nvptx/nvptx.c (split_mode_p): New function.
21188 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21190 2017-06-06 Tom de Vries <tom@codesourcery.com>
21192 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21194 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21197 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21200 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21202 * shrink-wrap.c (handle_simple_exit): Update profile.
21203 (try_shrink_wrapping): Upate profile.
21205 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21207 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21208 (tree_guess_outgoing_edge_probabilities): New.
21209 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21210 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21212 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21214 * ipa-split.c (split_function): Initialize return bb profile.
21216 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21218 * profile.c (compute_branch_probabilities): Also initialize
21219 EXIT_BLOCK profile.
21221 2017-06-06 Richard Biener <rguenther@suse.de>
21223 PR tree-optimization/80928
21224 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21225 (vect_analyze_loop_operations): Properly guard analysis for
21227 (vect_transform_loop): Likewise.
21228 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21229 (vect_model_induction_cost): Do not cost for pure SLP.
21230 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21231 of induction in inner loop vectorization.
21232 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21233 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21234 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21236 (vect_analyze_slp_cost_1): Cost induction.
21237 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21238 (vect_get_slp_vect_defs): Likewise.
21239 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21240 (vect_transform_stmt): Handle SLP reductions.
21241 * tree-vectorizer.h (vectorizable_induction): Adjust.
21243 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21245 * config/rs6000/rs6000.c (make_resolver_func): Update
21246 init_lowered_empty_function call.
21248 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21250 * doc/invoke.texi: Document the -fprofile-abs-path option.
21251 * common.opt (fprofile-abs-path): New option.
21252 * gcov-io.h (gcov_write_filename): Declare.
21253 * gcov-io.c (gcov_write_filename): New function.
21254 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21255 * profile.c (output_location): Likewise.
21257 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21259 * shring-wrap.c: Revert accidental commit.
21261 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21263 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21265 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21267 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21269 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21270 profile in callgraph edge.
21271 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21272 (apply_scale): Likewise.
21273 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21274 Also scale profile when inlining function with zero profile.
21275 (initialize_cfun): Update exit block profile even when it is zero.
21276 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21277 when profile is read.
21279 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21281 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21282 (CLONE_*): New constants to define the processors we can generate
21283 code for with the target_clone attribute.
21284 (rs6000_clone_map): New array to identify which clone processors
21285 the current program is running on.
21286 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21287 target_clone attribute.
21288 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21289 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21290 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21291 (cpu_expand_builtin): Add support for target_clone attribute.
21292 (rs6000_valid_attribute_p): Allow "default" attribute.
21293 (get_decl_name): New debug function to simplify printing the
21294 current function name in debugging statements.
21295 (rs6000_clone_priority): New functions to support the target_clone
21296 attribute, and be able to generate code to switch between ISA 2.05
21297 through ISA 3.0 (power6 through power9).
21298 (rs6000_compare_version_priority): Likewise.
21299 (rs6000_get_function_versions_dispatcher): Likewise.
21300 (make_resolver_func): Likewise.
21301 (add_condition_to_bb): Likewise.
21302 (dispatch_function_versions): Likewise.
21303 (rs6000_generate_version_dispatcher_body): Likewise.
21304 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21305 (fusion_gpr_load_p): Fix a spacing issue.
21306 * doc/extend.texi (Common Function Attributes): Document that the
21307 PowerPC supports the target_clone attribute.
21309 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21311 * config/arm/arm.h: explain F symbol found in description of ARM
21312 register allocation in its legend.
21314 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21316 * config/mips/frame-header-opt.c: Include profile-count.h.
21317 * config/riscv/riscv.c: Include profile-count.h
21319 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21321 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21323 (sm_set_flag_if_changed): Add bbs field.
21324 (execute_sm_if_changed_flag_set): Pass BBS.
21325 (execute_sm): Update.
21327 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21329 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21332 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21334 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21335 (peephole2): New peephole2 to emit the above.
21336 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21338 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21340 * config/aarch64/aarch64.c (define_peephole2 above
21341 *sub_<shift>_<mode>): New peephole.
21343 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21345 * config/i386/i386.c (make_resolver_func): Update.
21346 * Makefile.in: Add profile-count.h and profile-count.o
21347 * auto-profile.c (afdo_indirect_call): Update to new API.
21348 (afdo_set_bb_count): Update.
21349 (afdo_propagate_edge): Update.
21350 (afdo_propagate_circuit): Update.
21351 (afdo_calculate_branch_prob): Update.
21352 (afdo_annotate_cfg): Update.
21353 * basic-block.h: Include profile-count.h
21354 (struct edge_def): Turn count to profile_count.
21355 (struct basic_block_def): Likewie.
21356 (REG_BR_PROB_BASE): Move to profile-count.h
21357 (RDIV): Move to profile-count.h
21358 * bb-reorder.c (max_entry_count): Turn to profile_count.
21359 (find_traces): Update.
21360 (rotate_loop):Update.
21361 (connect_traces):Update.
21362 (sanitize_hot_paths):Update.
21363 * bt-load.c (migrate_btr_defs): Update.
21364 * cfg.c (RDIV): Remove.
21365 (init_flow): Use alloc_block.
21366 (alloc_block): Uninitialize count.
21367 (unchecked_make_edge): Uninitialize count.
21368 (check_bb_profile): Update.
21369 (dump_edge_info): Update.
21370 (dump_bb_info): Update.
21371 (update_bb_profile_for_threading): Update.
21372 (scale_bbs_frequencies_int): Update.
21373 (scale_bbs_frequencies_gcov_type): Update.
21374 (scale_bbs_frequencies_profile_count): New.
21375 * cfg.h (update_bb_profile_for_threading): Update.
21376 (scale_bbs_frequencies_profile_count): Declare.
21377 * cfgbuild.c (compute_outgoing_frequencies): Update.
21378 (find_many_sub_basic_blocks): Update.
21379 * cfgcleanup.c (try_forward_edges): Update.
21380 (try_crossjump_to_edge): Update.
21381 * cfgexpand.c (expand_gimple_tailcall): Update.
21382 (construct_exit_block): Update.
21383 * cfghooks.c (verify_flow_info): Update.
21384 (dump_bb_for_graph): Update.
21385 (split_edge): Update.
21386 (make_forwarder_block): Update.
21387 (duplicate_block): Update.
21388 (account_profile_record): Update.
21389 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21390 (get_estimated_loop_iterations): Update.
21391 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21392 (single_likely_exit): Update.
21393 * cfgloopmanip.c (scale_loop_profile): Update.
21395 (set_zero_probability): Update.
21396 (lv_adjust_loop_entry_edge): Update.
21397 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21398 (purge_dead_edges): Update.
21399 (rtl_account_profile_record): Update.
21400 * cgraph.c (cgraph_node::create): Uninitialize count.
21401 (symbol_table::create_edge): Uninitialize count.
21402 (cgraph_update_edges_for_call_stmt_node): Update.
21403 (cgraph_edge::dump_edge_flags): Update.
21404 (cgraph_node::dump): Update.
21405 (cgraph_edge::maybe_hot_p): Update.
21406 * cgraph.h: Include profile-count.h
21407 (create_clone), create_edge, create_indirect_edge): Update.
21408 (cgraph_node): Turn count to profile_count.
21409 (cgraph_edge0: Likewise.
21410 (make_speculative, clone): Update.
21411 (create_edge): Update.
21412 (init_lowered_empty_function): Update.
21413 * cgraphclones.c (cgraph_edge::clone): Update.
21414 (duplicate_thunk_for_node): Update.
21415 (cgraph_node::create_clone): Update.
21416 * cgraphunit.c (cgraph_node::analyze): Update.
21417 (cgraph_node::expand_thunk): Update.
21418 * final.c (dump_basic_block_info): Update.
21419 * gimple-streamer-in.c (input_bb): Update.
21420 * gimple-streamer-out.c (output_bb): Update.
21421 * graphite.c (print_global_statistics): Update.
21422 (print_graphite_scop_statistics): Update.
21423 * hsa-brig.c: Include basic-block.h.
21424 * hsa-dump.c: Include basic-block.h.
21425 * hsa-gen.c (T sum_slice): Update.
21426 (convert_switch_statements):Update.
21427 * hsa-regalloc.c: Include basic-block.h.
21428 * ipa-chkp.c (chkp_produce_thunks): Update.
21429 * ipa-cp.c (struct caller_statistics): Update.
21430 (init_caller_stats): Update.
21431 (gather_caller_stats): Update.
21432 (ipcp_cloning_candidate_p): Update.
21433 (good_cloning_opportunity_p): Update.
21434 (get_info_about_necessary_edges): Update.
21435 (dump_profile_updates): Update.
21436 (update_profiling_info): Update.
21437 (update_specialized_profile): Update.
21438 (perhaps_add_new_callers): Update.
21439 (decide_about_value): Update.
21440 (ipa_cp_c_finalize): Update.
21441 * ipa-devirt.c (struct odr_type_warn_count): Update.
21442 (struct decl_warn_count): Update.
21443 (struct final_warning_record): Update.
21444 (possible_polymorphic_call_targets): Update.
21445 (ipa_devirt): Update.
21446 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21447 * ipa-icf.c (sem_function::merge): Update.
21448 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21449 * ipa-inline.c (compute_uninlined_call_time): Update.
21450 (compute_inlined_call_time): Update.
21451 (want_inline_small_function_p): Update.
21452 (want_inline_self_recursive_call_p): Update.
21453 (edge_badness): Update.
21454 (lookup_recursive_calls): Update.
21455 (recursive_inlining): Update.
21456 (inline_small_functions): Update.
21457 (dump_overall_stats): Update.
21458 (dump_inline_stats): Update.
21459 * ipa-profile.c (ipa_profile_generate_summary): Update.
21460 (ipa_propagate_frequency): Update.
21461 (ipa_profile): Update.
21462 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21463 * ipa-utils.c (ipa_merge_profiles): Update.
21464 * loop-doloop.c (doloop_modify): Update.
21465 * loop-unroll.c (report_unroll): Update.
21466 (unroll_loop_runtime_iterations): Update.
21467 * lto-cgraph.c (lto_output_edge): Update.
21468 (lto_output_node): Update.
21469 (input_node): Update.
21470 (input_edge): Update.
21471 (merge_profile_summaries): Update.
21472 * lto-streamer-in.c (input_cfg): Update.
21473 * lto-streamer-out.c (output_cfg): Update.
21474 * mcf.c (create_fixup_graph): Update.
21475 (adjust_cfg_counts): Update.
21476 (sum_edge_counts): Update.
21477 * modulo-sched.c (sms_schedule): Update.
21478 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21479 * predict.c (maybe_hot_count_p): Update.
21480 (probably_never_executed): Update.
21481 (dump_prediction): Update.
21482 (combine_predictions_for_bb): Update.
21483 (propagate_freq): Update.
21484 (handle_missing_profiles): Update.
21485 (counts_to_freqs): Update.
21486 (rebuild_frequencies): Update.
21487 (force_edge_cold): Update.
21488 * predict.h: Include profile-count.h
21489 (maybe_hot_count_p, counts_to_freqs): UPdate.
21490 * print-rtl-function.c: Do not include cfg.h
21491 * print-rtl.c: Include basic-block.h
21492 * profile-count.c: New file.
21493 * profile-count.h: New file.
21494 * profile.c (is_edge_inconsistent): Update.
21495 (correct_negative_edge_counts): Update.
21496 (is_inconsistent): Update.
21497 (set_bb_counts): Update.
21498 (read_profile_edge_counts): Update.
21499 (compute_frequency_overlap): Update.
21500 (compute_branch_probabilities): Update; Initialize and deinitialize
21502 (branch_prob): Update.
21503 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21504 (edge_gcov_count): New.
21505 (bb_gcov_count): New.
21506 * shrink-wrap.c (try_shrink_wrapping): Update.
21507 * tracer.c (better_p): Update.
21508 * trans-mem.c (expand_transaction): Update.
21509 (ipa_tm_insert_irr_call): Update.
21510 (ipa_tm_insert_gettmclone_call): Update.
21511 * tree-call-cdce.c: Update.
21512 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21513 (gimple_duplicate_sese_tail): Update.
21514 (gimple_account_profile_record): Update.
21515 (execute_fixup_cfg): Update.
21516 * tree-inline.c (copy_bb): Update.
21517 (copy_edges_for_bb): Update.
21518 (initialize_cfun): Update.
21519 (freqs_to_counts): Update.
21520 (copy_cfg_body): Update.
21521 (expand_call_inline): Update.
21522 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21523 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21524 (try_unroll_loop_completely): Update.
21525 (try_peel_loop): Update.
21526 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21527 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21528 * tree-ssa-loop-split.c (connect_loops): Update.
21529 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21530 * tree-ssa-reassoc.c (branch_fixup): Update.
21531 * tree-ssa-tail-merge.c (replace_block_by): Update.
21532 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21533 (compute_path_counts): Update.
21534 (update_profile): Update.
21535 (recompute_probabilities): Update.
21536 (update_joiner_offpath_counts): Update.
21537 (estimated_freqs_path): Update.
21538 (freqs_to_counts_path): Update.
21539 (clear_counts_path): Update.
21540 (ssa_fix_duplicate_block_edges): Update.
21541 (duplicate_thread_path): Update.
21542 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21543 (struct switch_conv_info): Update.
21544 * tree-tailcall.c (decrease_profile): Update.
21545 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21546 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21547 * value-prof.c (check_counter): Update.
21548 (gimple_divmod_fixed_value): Update.
21549 (gimple_mod_pow2): Update.
21550 (gimple_mod_subtract): Update.
21551 (gimple_ic_transform): Update.
21552 (gimple_stringop_fixed_value): Update.
21553 * value-prof.h (gimple_ic): Update.
21555 2017-06-02 Carl Love <cel@us.ibm.com>
21557 * config/rs6000/rs6000-c: Add support for built-in functions
21558 vector double vec_doublee (vector signed int);
21559 vector double vec_doublee (vector unsigned int);
21560 vector double vec_doublee (vector float);
21561 vector double vec_doubleh (vector signed int);
21562 vector double vec_doubleh (vector unsigned int);
21563 vector double vec_doubleh (vector float);
21564 vector double vec_doublel (vector signed int);
21565 vector double vec_doublel (vector unsigned int);
21566 vector double vec_doublel (vector float);
21567 vector double vec_doubleo (vector signed int);
21568 vector double vec_doubleo (vector unsigned int);
21569 vector double vec_doubleo (vector float);.
21570 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21571 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21573 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21574 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21575 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21577 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21578 vec_doublel, vec_doubleh.
21579 * doc/extend.texi: Update the built-in documentation file for the
21580 new built-in functions.
21582 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21585 * ipa-inline-analysis.c (free_growth_caches): Set
21586 edge_removal_hook_holder to NULL after removing it.
21588 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21590 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21591 comparision with zero.
21593 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21594 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21595 for early expansion of vec_min and vec_max builtins.
21596 (builtin_function_type): Add min/max unsigned variants to those
21597 identified as having unsigned arguments.
21599 2017-06-02 Olivier Hainque <hainque@adacore.com>
21601 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21603 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21605 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21606 Use VALL_F16 iterator rather than VALL.
21608 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21610 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21611 Emit CBNZ inside loop when doing a strong exchange and comparing
21612 against zero. Generate the CC flags after the loop.
21614 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21616 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21617 (dl_section_ref): New.
21618 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21619 On AIX, append an expression to subtract the size of the
21620 section length to dl_section_ref.
21622 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21624 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21625 for early expansion of vector absolute builtins.
21627 2017-06-02 Richard Biener <rguenther@suse.de>
21629 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21630 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21632 2017-06-02 Richard Biener <rguenther@suse.de>
21634 PR tree-optimization/80948
21635 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21636 stmt order as well.
21638 2017-06-02 Richard Biener <rguenther@suse.de>
21640 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21642 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21643 for inductions as relevant.
21645 2017-06-02 Richard Biener <rguenther@suse.de>
21647 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21648 (vectorizable_induction): ... this. Remove dead code.
21650 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21652 * builtins. (expand_builtin_alloca): Remove second parameter and
21653 infer its value from the first parameter instead.
21654 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21656 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21658 PR rtl-optimization/80903
21659 * loop-doloop.c (add_test): Unshare sequence.
21661 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21663 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21665 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21667 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21669 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21670 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21671 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21672 (xlogue_layout::get_stub_rtx): Make static.
21673 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21674 (xlogue_layout::compute_stub_managed_regs): Rename to...
21675 (xlogue_layout::count_stub_managed_regs): ...this.
21676 (xlogue_layout::is_stub_managed_reg): New function.
21677 (xlogue_layout::m_stub_names): Rename to...
21678 (xlogue_layout::s_stub_names): ...this, make static.
21679 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21680 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21681 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21682 xlogue_layout::s_stub_names): Instantiate statics.
21683 (stub_managed_regs): Remove.
21684 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21685 (disable_call_ms2sysv_xlogues): Rename to...
21686 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21687 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21689 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21690 change after reload_completed.
21691 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21693 (ix86_expand_prologue): Likewise.
21694 (ix86_expand_epilogue): Likewise.
21695 (ix86_expand_split_stack_prologue): Likewise.
21696 (ix86_compute_frame_layout): Remove frame parameter ...
21697 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21698 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21700 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21701 (ix86_frame): Move from here ...
21702 * config/i386/i386.h (ix86_frame): ... to here.
21703 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21704 complete ix86_frame data structure instead. Remove some_ld_name.
21706 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21708 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21709 symbols that hold a DECL_VALUE_EXPR.
21711 2017-06-01 Martin Jambor <mjambor@suse.cz>
21713 PR tree-optimization/80898
21714 * tree-sra.c (process_subtree_disqualification): Removed.
21715 (disqualify_candidate): Do not acll
21716 process_subtree_disqualification.
21717 (subtree_mark_written_and_enqueue): New function.
21718 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21719 RHS has been disqualified and re-queue LHS if necessary. Apart
21720 from that, ignore disqualified RHS.
21722 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21724 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21725 address fetch for z10 or later.
21727 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21729 * config/arc/arc.md (tst_movb): Add guard when splitting.
21731 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21733 * config/arc/arc.c (arc_can_eliminate): Test against
21734 arc_frame_pointer_needed.
21736 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21738 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21739 to prevent store reordering.
21740 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21741 (type): Add block type.
21742 (stack_tie): Define special instruction to be used in
21745 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21747 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21748 constraint. It is not valid for the pattern.
21749 (noncommutative_binary_comparison): Likewise.
21751 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21753 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21756 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21758 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21759 be used by the reg-alloc.
21761 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21763 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21764 reg-alloc when having mul64 or mul32x16 instructions.
21765 (mulsidi3): Likewise.
21766 (umulsidi3): Likewise.
21767 (mulsi32x16): New pattern.
21768 (mulsi64): Likewise.
21769 (mulsidi64): Likewise.
21770 (umulsidi64): Likewise.
21771 (MUL32x16_REG): Define.
21772 (mul64_600): Use MUL32x16_REG.
21773 (mac64_600): Likewise.
21774 (umul64_600): Likewise.
21775 (umac64_600): Likewise.
21777 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21779 * config/arc/arc.md (mulsi3_700): Make it commutative.
21781 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21783 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21785 (*sign_extendsidi2_insn): Likewise for movstosw.
21787 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21789 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21790 the type of the input discriminant value. Convert the
21791 discriminant value of signedness vary.
21793 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21795 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21796 Add to -Wall section.
21798 2017-06-01 Richard Biener <rguenther@suse.de>
21800 PR middle-end/66313
21801 * fold-const.c (fold_plusminus_mult_expr): If the factored
21802 factor may be zero use a wrapping type for the inner operation.
21803 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21804 and handle moved defs.
21805 (process_assignment): Properly guard the unary op case. Return a
21806 tri-state indicating that moving the stmt before the call may allow
21807 to continue. Pass through to_move.
21808 (find_tail_calls): Handle moving unrelated defs before
21811 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21814 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21815 splitter result in the canonical way.
21817 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21819 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21820 also for 32bit target. Update insn attributes.
21821 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21823 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21825 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21826 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21827 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21828 (_mm_maskz_min_ss): New intrinsics.
21830 2017-05-31 Martin Liska <mliska@suse.cz>
21832 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21833 Change comment style to one we normally use.
21834 (vectorizable_reduction): Likewise.
21835 (vectorizable_induction): Likewise.
21836 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21837 (vectorizable_call): Likewise.
21838 (vectorizable_simd_clone_call): Likewise.
21839 (vectorizable_conversion): Likewise.
21840 (vectorizable_assignment): Likewise.
21841 (vectorizable_shift): Likewise.
21842 (vectorizable_operation): Likewise.
21843 (vectorizable_store): Likewise.
21844 (vectorizable_load): Likewise.
21845 * tree-vectorizer.h: Likewise.
21847 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21849 * passes.c (emergency_dump_function): New.
21850 * tree-pass.h (emergency_dump_function): Declare.
21851 * plugin.c (plugins_internal_error_function): Remove.
21852 * plugin.h (plugins_internal_error_function): Remove declaration.
21853 * toplev.c (internal_error_function): New static function. Use it...
21854 (general_init): ...here.
21856 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21858 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21859 (arc_rtx_costs): Add costs for new patterns.
21860 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21861 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21863 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21865 * tree-ssa-strlen.c (get_next_strinfo): New function.
21866 (get_stridx_plus_constant): Use it.
21867 (zero_length_string): Likewise.
21868 (adjust_related_strinfos): Likewise.
21869 (adjust_last_stmt): Likewise.
21871 2017-05-31 Richard Biener <rguenther@suse.de>
21874 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21875 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21877 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21879 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21880 loop_vinfo argument and use of dependence distance vectors.
21881 Check instead whether the two references differ only in their
21882 initial value and assume that they have the same alignment if the
21883 difference is a multiple of the vector alignment.
21884 (vect_analyze_data_refs_alignment): Update call accordingly.
21886 2017-05-31 Martin Liska <mliska@suse.cz>
21889 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21891 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21893 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21894 (create_intersect_range_checks): Move from ...
21895 * tree-data-ref.c (create_intersect_range_checks_index)
21896 (create_intersect_range_checks): ... to here.
21897 (create_runtime_alias_checks): New function factored from ...
21898 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21899 here. Call above function.
21900 * tree-data-ref.h (create_runtime_alias_checks): New function.
21902 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21904 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21905 segment length for dr_b and compute it in wide_int.
21907 2017-05-31 Richard Biener <rguenther@suse.de>
21909 PR tree-optimization/80906
21910 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21911 and pass through iv_map.
21912 (copy_bb_and_scalar_dependences): Adjust.
21913 (translate_pending_phi_nodes): Likewise.
21914 (copy_loop_close_phi_args): Handle code-generating IVs instead
21917 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21919 * diagnostic-color.c (color_dict): Add "type-diff".
21920 (parse_gcc_colors): Update comment.
21921 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21922 -fdiagnostics-show-template-tree and -fno-elide-type.
21923 (GCC_COLORS): Add type-diff to example.
21925 (-fdiagnostics-show-template-tree): New.
21926 (-fno-elide-type): New.
21927 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21928 the pp_format_decoder callback. Call any m_format_postprocessor's
21930 (pretty_printer::pretty_printer): Initialize
21931 m_format_postprocessor.
21932 (pretty_printer::~pretty_printer): Delete any
21933 m_format_postprocessor.
21934 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21935 (class format_postprocessor): New class.
21936 (struct pretty_printer::format_decoder): Document the new parameters.
21937 (struct pretty_printer::m_format_postprocessor): New field.
21938 * tree-diagnostic.c (default_tree_printer): Update for new
21939 bool and const char ** params.
21940 * tree-diagnostic.h (default_tree_printer): Likewise.
21942 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21944 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21945 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21946 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21947 rs6000_gen_cell_microcode code.
21948 (rs6000_final_prescan_insn): Delete.
21949 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21950 "warn-cell-microcode" entries.
21951 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21952 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21953 throughout. Change cc_reg_not_micro_cr0_operand to
21954 cc_reg_not_cr0_operand throughout.
21955 (*extendhi<mode>2_noload): Delete.
21956 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21957 (mwarn-cell-microcode): Delete.
21958 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21959 -mgen-cell-microcode and -mwarn-cell-microcode.
21961 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21964 * config/i386/constraints.md (Yd): New constraint.
21966 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21967 and (?Yd, r) alternatives. Update insn attributes.
21968 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21969 and (?*Yd, r) alternatives. Update insn attributes.
21970 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21972 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21974 * gimplify.c (gimplify_modify_expr): Don't create a
21975 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21978 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21980 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21982 2017-05-30 Richard Biener <rguenther@suse.de>
21984 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21985 and reduc_def fields.
21986 (STMT_VINFO_REDUC_TYPE): New define.
21987 (STMT_VINFO_REDUC_DEF): Likewise.
21988 (vect_force_simple_reduction): Adjust prototype.
21989 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21990 (vect_is_simple_reduction): Remove check_reduction argument.
21991 (vect_force_simple_reduction): Adjust and set
21992 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21993 (vectorizable_reduction): Do not re-do reduction analysis
21994 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21995 * tree-parloops.c (gather_scalar_reductions): Adjust.
21997 2017-05-30 Richard Biener <rguenther@suse.de>
21999 PR middle-end/80901
22000 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22003 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22005 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22006 Introduce unknown_misalignment parameter and remove vf.
22007 (vect_peeling_hash_get_lowest_cost):
22008 Pass unknown_misalignment parameter.
22009 (vect_enhance_data_refs_alignment):
22010 Fix unsupportable data ref treatment.
22012 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22014 * tree-vect-data-refs.c (vect_get_data_access_cost):
22015 Workaround for SLP handling.
22016 (vect_enhance_data_refs_alignment):
22017 Compute costs for doing no peeling at all, compare to the best
22018 peeling costs so far and avoid peeling if cheaper.
22020 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22022 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22023 Return peeling info and set costs to zero for unlimited cost
22025 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22026 with unknown misalignment. Compute and costs for unknown
22027 misalignment, compare them to the costs for known misalignment
22028 and choose the cheapest for peeling.
22030 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22032 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22033 (vect_get_peeling_costs_all_drs): Create function.
22034 (vect_peeling_hash_get_lowest_cost):
22035 Use vect_get_peeling_costs_all_drs.
22036 (vect_peeling_supportable): Create function.
22037 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22039 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22041 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22042 DR_HAS_NEGATIVE_STEP.
22043 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22044 (vect_enhance_data_refs_alignment): Use.
22045 (vect_duplicate_ssa_name_ptr_info): Use.
22046 * tree-vectorizer.h (dr_misalignment): Use.
22047 (known_alignment_for_access_p): Use.
22049 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22052 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22054 (has_section_name): New function.
22056 2017-05-30 Martin Liska <mliska@suse.cz>
22059 * auto-profile.c (get_function_decl_from_block): Fix
22062 2017-05-30 Richard Biener <rguenther@suse.de>
22064 PR middle-end/80876
22065 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22067 2017-05-30 Martin Liska <mliska@suse.cz>
22069 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22070 * dumpfile.h (struct dump_file_info): Remove ctors.
22072 2017-05-30 Martin Liska <mliska@suse.cz>
22074 * predict.def: Fix GNU coding style.
22076 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22078 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22079 Mark 'to' argument with ATTRIBUTE_UNUSED.
22081 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22083 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22084 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22085 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22088 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22090 * doc/install.texi (Options specification): Restore entry of
22091 --enable-sjlj-exceptions.
22093 2017-05-27 Michael Eager <eager@eagercon.com>
22096 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22098 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22100 * config/microblaze/microblaze.h
22101 (FIXED_REGISTERS): Update in macro.
22102 (CALL_USED_REGISTERS): Update in macro.
22104 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22106 * doc/install.texi: Add links to macOS binary distributions.
22108 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22112 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22114 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22116 2017-05-26 Martin Liska <mliska@suse.cz>
22118 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22120 2017-05-26 Martin Liska <mliska@suse.cz>
22122 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22123 always leading ';; '.
22124 (dump_bb_info): Likewise.
22125 (brief_dump_cfg): Likewise.
22126 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22127 * dumpfile.c: Remove usage of TDF_VERBOSE.
22128 * dumpfile.h (enum dump_kind): Likewise.
22129 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22130 * print-tree.c (debug_verbose): Remove.
22131 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22132 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22133 * tree-diagnostic.c (default_tree_printer): Replace
22134 TDF_DIAGNOSTIC with TDF_SLIM.
22136 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22138 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22139 in parameter loop, rather than loop_vinfo.
22140 (create_intersect_range_checks): Ditto.
22141 (vect_create_cond_for_alias_checks): Update call to above functions.
22143 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22145 PR tree-optimization/80815
22146 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22147 for merging runtime alias checks. Handle negative DR_STEPs.
22149 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22151 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22153 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22154 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22155 out code pruning runtime alias checks.
22156 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22157 factored out from above.
22158 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22160 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22162 (prune_runtime_alias_test_list): New decalaration.
22164 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22166 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22167 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22168 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22169 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22170 (operator==, comp_dr_with_seg_len_pair): Ditto.
22171 (vect_prune_runtime_alias_test_list): Ditto.
22173 2017-05-26 Martin Liska <mliska@suse.cz>
22176 * params.def: Bound partial-inlining-entry-probability param.
22178 2017-05-26 Marek Polacek <polacek@redhat.com>
22181 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22184 2017-05-26 Richard Biener <rguenther@suse.de>
22186 PR tree-optimization/80842
22187 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22190 2017-05-26 Richard Biener <rguenther@suse.de>
22192 PR tree-optimization/80844
22193 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22195 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22197 * doc/md.texi (Machine Constraints): Update x86 family
22198 machine constraints section to match 'config/i386/constraints.md'.
22200 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22202 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22204 2017-05-25 Nathan Sidwell <nathan@acm.org>
22206 * doc/invoke.texi (--enable-languages): Update documentation.
22208 2017-05-25 Martin Liska <mliska@suse.cz>
22210 * dumpfile.c: Add TDF_FOLDING.
22211 * dumpfile.h (enum dump_kind): Likewise.
22212 * genmatch.c (dt_simplify::gen_1): Use it.
22214 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22216 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22218 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22220 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22221 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22223 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22225 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22226 * match.pd (X == C): Rewrite it here.
22227 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22228 with_certain_nonzero_bits2): New predicates.
22229 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22231 2017-05-24 Nathan Sidwell <nathan@acm.org>
22233 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22236 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22239 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22241 * config/powerpcspe: New port. Files are copied from the rs6000
22242 port, with "rs6000" in filenames replaced by "powerpcspe".
22244 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22246 PR rtl-optimization/80754
22247 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22249 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22251 * config/sparc/sparc.md (length): Return the correct value for -mflat
22252 sibcalls to match output_sibcall.
22254 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22258 * config/rs6000/rs6000.c (struct machine_function): Add new field
22260 (rs6000_get_separate_components): Init that field, use it.
22261 (rs6000_components_for_bb): Use the field.
22263 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22265 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22267 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22269 PR middle-end/80823
22270 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22272 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22275 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22276 address against address_operand predicate.
22277 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22279 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22281 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22282 parameters passed indirectly.
22284 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22286 * config/i386/i386.md (*movdi_internal): Remove SSE4
22287 alternative 18 (?r, *v). Update insn attributes.
22288 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22289 Update insn attributes.
22290 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22291 Update insn attributes.
22292 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22293 alternative 1 (r, v). Remove isa attribute.
22294 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22295 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22296 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22298 2017-05-23 Tom de Vries <tom@codesourcery.com>
22300 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22303 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22305 * cgraphunit.c (symbol_table::process_new_functions): Update.
22306 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22307 (inline_generate_summary): Rename to ...
22308 (ipa_fn_summary_generate): ... this one.
22309 (inline_read_summary): Rename to ...
22310 (ipa_fn_summary_read): ... this one.
22311 (inline_write_summary): Rename to ...
22312 (ipa_fn_summary_write): ... this one.
22313 (inline_free_summary): Rename to ...
22314 (ipa_free_fn_summary): ... this one.
22315 (pass_data_local_fn_summary, pass_local_fn_summary,
22316 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22317 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22318 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22319 make_pass_ipa_fn_summary): New.
22320 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22321 inline_write_summary, inline_free_summary): Remove.
22322 (ipa_free_fn_summary) : New.
22323 * ipa-inline.c (ipa_inline): Update.
22324 (pass_ipa_inline): Do not generate summaries.
22325 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22327 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22328 and add pass_ipa_fn_summary.
22329 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22331 (make_pass_inline_parameters): Remove.
22333 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22335 * omp-low.c (struct omp_context): Remove "default_kind" member.
22338 * omp-offload.c (execute_oacc_device_lower): Remove the
22339 parallelism dimensions function attributes for unparallelized
22340 OpenACC kernels constructs.
22342 2017-05-23 Martin Liska <mliska@suse.cz>
22344 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22346 (cgraph_edge::make_speculative): Likewise.
22347 (cgraph_edge::resolve_speculation): Likewise.
22348 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22349 (cgraph_node::dump): Likewise.
22350 * cgraph.h: Likewise.
22351 * cgraphunit.c (analyze_functions): Likewise.
22352 (symbol_table::compile): Likewise.
22353 * ipa-cp.c (print_all_lattices): Likewise.
22354 (determine_versionability): Likewise.
22355 (initialize_node_lattices): Likewise.
22356 (ipcp_verify_propagated_values): Likewise.
22357 (estimate_local_effects): Likewise.
22358 (update_profiling_info): Likewise.
22359 (create_specialized_node): Likewise.
22360 (perhaps_add_new_callers): Likewise.
22361 (decide_about_value): Likewise.
22362 (decide_whether_version_node): Likewise.
22363 (identify_dead_nodes): Likewise.
22364 (ipcp_store_bits_results): Likewise.
22365 * ipa-devirt.c (dump_targets): Likewise.
22366 (ipa_devirt): Likewise.
22367 * ipa-icf.c (sem_item::dump): Likewise.
22368 (sem_function::equals): Likewise.
22369 (sem_variable::equals): Likewise.
22370 (sem_item_optimizer::read_section): Likewise.
22371 (sem_item_optimizer::execute): Likewise.
22372 (congruence_class::dump): Likewise.
22373 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22374 (dump_inline_summary): Likewise.
22375 (estimate_node_size_and_time): Likewise.
22376 (inline_analyze_function): Likewise.
22377 * ipa-inline-transform.c (inline_call): Likewise.
22378 * ipa-inline.c (report_inline_failed_reason): Likewise.
22379 (want_early_inline_function_p): Likewise.
22380 (edge_badness): Likewise.
22381 (update_edge_key): Likewise.
22382 (inline_small_functions): Likewise.
22383 * ipa-profile.c (ipa_profile): Likewise.
22384 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22385 (ipa_make_edge_direct_to_target): Likewise.
22386 (remove_described_reference): Likewise.
22387 (ipa_impossible_devirt_target): Likewise.
22388 (propagate_controlled_uses): Likewise.
22389 (ipa_print_node_params): Likewise.
22390 (ipcp_transform_function): Likewise.
22391 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22392 (propagate_pure_const): Likewise.
22393 * ipa-reference.c (generate_summary): Likewise.
22394 (read_write_all_from_decl): Likewise.
22395 (propagate): Likewise.
22396 (ipa_reference_read_optimization_summary): Likewise.
22397 * ipa-utils.c (ipa_merge_profiles): Likewise.
22398 * ipa.c (walk_polymorphic_call_targets): Likewise.
22399 (symbol_table::remove_unreachable_nodes): Likewise.
22400 (ipa_single_use): Likewise.
22401 * passes.c (execute_todo): Likewise.
22402 * predict.c (drop_profile): Likewise.
22403 * symtab.c (symtab_node::get_dump_name): New function.
22404 (symtab_node::dump_name): Likewise.
22405 (symtab_node::dump_asm_name): Likewise.
22406 (symtab_node::dump_references): Likewise.
22407 (symtab_node::dump_referring): Likewise.
22408 (symtab_node::dump_base): Likewise.
22409 (symtab_node::debug_symtab): Likewise.
22410 * tree-sra.c (convert_callers_for_node): Likewise.
22411 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22412 * value-prof.c (init_node_map): Likewise.
22414 2017-05-23 Martin Liska <mliska@suse.cz>
22416 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22417 and symtab_node::debug_symtab to symbol_table::debug.
22418 * cgraphunit.c (analyze_functions): Use the renamed function.
22419 (symbol_table::compile): Likewise.
22420 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22421 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22422 * passes.c (execute_todo): Likewise.
22423 * symtab.c (symbol_table::dump): New function.
22424 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22426 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22428 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22429 that nonconst implies exec.
22431 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22433 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22434 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22435 (inline_edge_summary_vec): Turn into ...
22436 (ipa_call_summaries): ... this one.
22437 (redirect_to_unreachable, edge_set_predicate,
22438 evaluate_properties_for_edge, inline_summary_alloc,
22439 reset_ipa_call_summary, reset_inline_summary,
22440 inline_summary_t::duplicate): Update.
22441 (inline_edge_duplication_hook): Turn to ...
22442 (ipa_call_summary_t::duplicate): ... this one.
22443 (inline_edge_removal_hook): Turn to ...
22444 (ipa_call_summary_t::remove): ... this one.
22445 (dump_inline_edge_summary): Turn to ...
22446 (dump_ipa_call_summary): ... this one.
22447 (estimate_function_body_sizes): Update.
22448 (inline_update_callee_summaries): Update.
22449 (remap_edge_change_prob): Update.
22450 (remap_edge_summaries): Update.
22451 (inline_merge_summary): Update.
22452 (do_estimate_edge_time): Update.
22453 (inline_generate_summary): Update.
22454 (inline_read_section): Update.
22455 (inline_read_summary): Update.
22456 (inline_free_summary): Update.
22457 * ipa-inline.c (can_inline_edge_p): Update.
22458 (compute_inlined_call_time): Update.
22459 (want_inline_small_function_p): Update.
22460 (edge_badness): Update.
22461 (early_inliner): Update.
22462 * ipa-inline.h (inline_edge_summary): Turn to ...
22463 (ipa_call_summary): ... this one.
22464 (ipa_call_summary_t): New class.
22465 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22466 (ipa_call_summaries): New.
22467 (inline_edge_summary): Remove.
22468 (estimate_edge_growth): Update.
22469 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22470 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22471 * ipa-split.c (execute_split_functions): Update.
22472 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22474 2017-05-23 Tom de Vries <tom@codesourcery.com>
22476 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22477 attributes): Document rdrand effective target.
22479 2017-05-23 Tom de Vries <tom@codesourcery.com>
22481 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22482 attributes): Sort alphabetically.
22484 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22486 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22488 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22491 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22492 V2DF/V2DI splat into two separate patterns, one that handles
22493 registers, and the other that only handles memory. Drop support
22494 for splatting from a GPR on ISA 2.07 and then splitting the
22495 splat into direct move and splat.
22496 (vsx_splat_<mode>_reg): Likewise.
22497 (vsx_splat_<mode>_mem): Likewise.
22499 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22501 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22503 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22505 PR middle-end/80809
22506 * omp-low.c (finish_taskreg_remap): New function.
22507 (finish_taskreg_scan): If unit size of ctx->record_type
22508 is non-constant, unshare the size expression and replace
22509 decls in it with possible outer var refs.
22511 PR middle-end/80809
22512 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22513 GOVD_SHARED rather than GOVD_PRIVATE with it.
22514 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22515 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22517 PR middle-end/80853
22518 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22519 as last argument to build_outer_var_ref for pointer bases of array
22520 section reductions.
22522 2017-05-19 Martin Sebor <msebor@redhat.com>
22524 * print-tree.c (print_node): Print DECL_READ_P flag.
22526 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22528 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22529 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22530 * cgraph.c: Likewise.
22531 * cgraphunit.c: Likewise.
22532 * gengtype.c: Likewise.
22533 * ipa-cp.c: Likewise.
22534 * ipa-devirt.c: Likewise.
22535 * ipa-icf.c: Likewise.
22536 * ipa-predicate.c: Likewise.
22537 * ipa-profile.c: Likewise.
22538 * ipa-prop.c: Likewise.
22539 * ipa-split.c: Likewise.
22541 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22542 edge_predicate_pool, dump_inline_hints,
22543 inline_summary::account_size_time, redirect_to_unreachable,
22544 edge_set_predicate, set_hint_predicate,
22545 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22546 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22547 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22548 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22549 ipa_call_summary_t::remove, initialize_growth_caches,
22550 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22551 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22552 mark_modified, unmodified_parm_1, unmodified_parm,
22553 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22554 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22555 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22556 will_be_nonconstant_predicate, record_modified_bb_info,
22557 get_minimal_bb, record_modified, param_change_prob,
22558 phi_result_unknown_predicate, predicate_for_phi_result,
22559 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22560 estimate_function_body_sizes, compute_inline_parameters,
22561 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22562 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22563 inline_update_callee_summaries, remap_edge_change_prob,
22564 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22565 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22566 inline_analyze_function, inline_summary_t::insert,
22567 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22568 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22569 inline_free_summary): Move to ipa-fnsummary.h
22570 (predicate_t): Remove.
22571 * ipa-fnsummary.c: New file.
22572 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22573 (enum inline_hints_vals, inline_hints, agg_position_info,
22574 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22575 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22576 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22577 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22578 inline_read_summary, inline_write_summary, inline_free_summary,
22579 inline_analyze_function, initialize_inline_failed,
22580 inline_merge_summary, inline_update_overall_summary,
22581 compute_inline_parameters): Move to ipa-fnsummary.h
22582 * ipa-fnsummary.h: New file.
22583 * ipa-inline-transform.h: Include ipa-inline.h.
22584 * ipa-inline.c: LIkewise.
22586 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22588 * ipa-inline.c (edge_badness): Use inlined_time instead of
22589 inline_summaries->get.
22591 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22593 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22595 2017-05-22 Nathan Sidwell <nathan@acm.org>
22597 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22598 (fdump-lang): Document 'raw' option.
22599 * dumpfile.h (TDI_tu): Delete.
22600 * dumpfile.c (dump_files): Remove translation-unit.
22601 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22603 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22605 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22606 command option from $(AWK) call.
22607 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22608 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22609 [FORMAT]: Remove handling of variable.
22610 * config/avr/t-multilib: Regenerate.
22612 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22614 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22616 (dump_inline_summary): Do not print self_time.
22617 (estimate_function_body_sizes): Do not set self_time.
22618 (compute_inline_parameters): Likewise.
22619 (inline_read_section, inline_write_summary): Do not stream self_time.
22620 * ipa-inline.h (inline_summary): Drop self_time.
22622 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22624 * ipa-inline-analysis.c (account_size_time): Rename to ...
22625 (inline_summary::account_size_time): ... this one.
22626 (reset_ipa_call_summary): Turn to ...
22627 (ipa_call_summary::reset): ... this one.
22628 (reset_inline_summary): Turn to ...
22629 (inline_summary::reset): ... this one.
22630 (inline_summary_t::remove): Update.
22631 (inline_summary_t::duplicate): Update.
22632 (ipa_call_summary_t::remove): Update.
22633 (dump_inline_summary): Update.
22634 (estimate_function_body_sizes): Update.
22635 (compute_inline_parameters): Update.
22636 (estimate_node_size_and_time): Update.
22637 (inline_merge_summary): Update.
22638 (inline_update_overall_summary): Update.
22639 (inline_read_section): Update.
22640 (inline_write_summary): Update.
22641 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22642 add account_size_time and reset member functions.
22643 (ipa_call_summary): Add reset function.
22644 * ipa-predicate.h (predicate::operator &): Constify.
22646 2017-05-22 Richard Biener <rguenther@suse.de>
22648 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22650 2017-05-19 Jason Merrill <jason@redhat.com>
22652 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22654 2017-05-19 Marek Polacek <polacek@redhat.com>
22657 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22658 TYPE_OVERFLOW_WRAPS checks.
22660 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22662 * tree-core.h (enum omp_clause_default_kind): Add
22663 "OMP_CLAUSE_DEFAULT_PRESENT".
22664 * tree-pretty-print.c (dump_omp_clause): Handle it.
22665 * gimplify.c (enum gimplify_omp_var_data): Add
22666 "GOVD_MAP_FORCE_PRESENT".
22667 (gimplify_adjust_omp_clauses_1): Map it to
22668 "GOMP_MAP_FORCE_PRESENT".
22669 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22671 * gimplify.c (oacc_default_clause): Clarify.
22673 2017-05-19 Nathan Sidwell <nathan@acm.org>
22675 LANG_HOOK_REGISTER_DUMPS
22676 * toplev.c (general_init): Call register dump lang hook.
22677 * doc/invoke.texi: Document -fdump-lang option family.
22678 * dumpfile.c (dump_files): Remove class dump here.
22679 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22680 * dumpfile.h (tree_dump_index): Remove TDI_class.
22681 * langhooks-def.h (lhd_register_dumps): Declare.
22682 (LANG_HOOKS_REGISTER_DUMPS): Define.
22683 (LANG_HOOKS_INITIALIZER): Add it.
22684 * langhooks.c (lhd_register_dumps): Define.
22685 * langhooks.h (struct lang_hooks): Add register_dumps.
22687 2017-05-19 Nathan Sidwell <nathan@acm.org>
22689 * context.h (context::set_passes): New.
22690 * context.c (context::context): Do not create pass manager.
22691 * toplev.c (general_init): Create pass manager here.
22693 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22695 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22696 use this splitter if two add or or instructions would also work for
22697 the constant we want to generate.
22699 2017-05-19 Richard Biener <rguenther@suse.de>
22702 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22703 predicate evaluation.
22705 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22707 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22709 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22712 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22714 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22715 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22716 (inline_edge_summary_vec): Turn into ...
22717 (ipa_call_summaries): ... this one.
22718 (redirect_to_unreachable, edge_set_predicate,
22719 evaluate_properties_for_edge, inline_summary_alloc,
22720 reset_ipa_call_summary, reset_inline_summary,
22721 inline_summary_t::duplicate): Update.
22722 (inline_edge_duplication_hook): Turn to ...
22723 (ipa_call_summary_t::duplicate): ... this one.
22724 (inline_edge_removal_hook): Turn to ...
22725 (ipa_call_summary_t::remove): ... this one.
22726 (dump_inline_edge_summary): Turn to ...
22727 (dump_ipa_call_summary): ... this one.
22728 (estimate_function_body_sizes): Update.
22729 (inline_update_callee_summaries): Update.
22730 (remap_edge_change_prob): Update.
22731 (remap_edge_summaries): Update.
22732 (inline_merge_summary): Update.
22733 (do_estimate_edge_time): Update.
22734 (inline_generate_summary): Update.
22735 (inline_read_section): Update.
22736 (inline_read_summary): Update.
22737 (inline_free_summary): Update.
22738 * ipa-inline.c (can_inline_edge_p): Update.
22739 (compute_inlined_call_time): Update.
22740 (want_inline_small_function_p): Update.
22741 (edge_badness): Update.
22742 (early_inliner): Update.
22743 * ipa-inline.h (inline_edge_summary): Turn to ...
22744 (ipa_call_summary): ... this one.
22745 (ipa_call_summary_t): New class.
22746 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22747 (ipa_call_summaries): New.
22748 (inline_edge_summary): Remove.
22749 (estimate_edge_growth): Update.
22750 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22751 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22752 * ipa-split.c (execute_split_functions): Update.
22753 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22755 2017-05-19 Richard Biener <rguenther@suse.de>
22757 PR middle-end/80764
22758 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22760 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22762 * config/rs6000/rs6000.c (struct machine_function): Add field
22763 fpr_is_wrapped_separately.
22764 (rs6000_get_separate_components): Use 64 components. Handle the
22765 new FPR components.
22766 (rs6000_components_for_bb): Handle the FPR components.
22767 (rs6000_emit_prologue_components): Handle the FPR components.
22768 (rs6000_emit_epilogue_components): Handle the FPR components.
22769 (rs6000_set_handled_components): Handle the FPR components.
22770 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22771 that are already separately shrink-wrapped.
22772 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22773 that are already separately shrink-wrapped.
22775 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22778 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22781 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22782 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22783 cases where the register allocator uses a move and an offsettable
22784 memory operation to/from a FPR register on ISA 2.06/2.07.
22785 (define_peephole2 for Altivec d-form store): Likewise.
22787 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22790 * config/i386/mmx.md (*mov<mode>_internal): Enable
22791 alternatives 11, 12, 13 and 14 also for 32bit targets.
22792 Remove alternatives 15, 16, 17 and 18.
22793 * config/i386/sse.md (vec_concatv2di): Change
22794 alternative (!x, *y) to (x, ?!*Yn).
22796 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22798 * dumpfile.h (enum dump_kind): Remove stray comma.
22800 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22802 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22803 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22804 predicate::num_conditions
22805 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22806 (CHANGED): turn into predicate::changed.
22807 (agg_position_info): Move to ipa-predicate.h
22808 (add_condition, predicate::add_clause, predicate::operator &=,
22809 predicate::or_with, predicate::evaluate, predicate::probability,
22810 dump_condition, dump_clause, predicate::dump,
22811 predicate::remap_after_duplication, predicate::remap_after_inlining,
22812 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22813 (evaluate_conditions_for_known_args): Update.
22814 (set_cond_stmt_execution_predicate): Update.
22815 * ipa-inline.h: Include ipa-predicate.h
22816 (condition, inline_param_summary, conditions, agg_position_info,
22817 predicate): Move to ipa-predicate.h
22818 * ipa-predicate.c: New file.
22819 * ipa-predicate.h: New file.
22821 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22823 * final.c (leaf_function_p): Check we are not in a sequence.
22825 2017-05-18 Martin Liska <mliska@suse.cz>
22827 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22828 * dumpfile.c (dump_register): Use new enum dump_kind.
22829 (get_dump_file_name): Likewise.
22830 (dump_enable_all): Likewise.
22831 (dump_switch_p_1): Likewise.
22832 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22833 * dumpfile.h (enum dump_kind): New enum type.
22834 (struct dump_file_info): Create constructor and
22835 format fields and comments.
22836 * passes.c (pass_manager::register_one_dump_file):
22838 * statistics.c (statistics_early_init): Likewise.
22839 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22840 TDF_TREE with TDF_SLIM.
22841 (gather_memory_references_ref): Likewise.
22843 2017-05-18 Martin Liska <mliska@suse.cz>
22845 * vec.h (struct vnull): Use it.
22847 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22849 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22850 (true_predicate, false_predicate, true_predicate_p,
22851 false_predicate_p): Remove.
22852 (single_cond_predicate, not_inlined_predicate): Turn to member function
22854 (add_condition): Update.
22855 (add_clause): Turn to...
22856 (predicate::add_clause): ... this one; update; allow passing NULL
22858 (and_predicates): Turn to ...
22859 (predicate::operator &=): ... this one.
22860 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22861 (or_predicates): Turn to ...
22862 (predicate::or_with): ... this one.
22863 (evaluate_predicate): Turn to ...
22864 (predicate::evaluate): ... this one.
22865 (predicate_probability): Turn to ...
22866 (predicate::probability): ... this one.
22867 (dump_condition): Update.
22868 (dump_predicate): Turn to ...
22869 (predicate::dump): ... this one.
22870 (account_size_time): Update.
22871 (edge_set_predicate): Update.
22872 (set_hint_predicate): UPdate.
22873 (evaluate_conditions_for_known_args): Update.
22874 (evaluate_properties_for_edge): Update.
22875 (remap_predicate_after_duplication): Turn to...
22876 (predicate::remap_after_duplication): ... this one.
22877 (remap_hint_predicate_after_duplication): Update.
22878 (inline_summary_t::duplicate): UPdate.
22879 (dump_inline_edge_summary): Update.
22880 (dump_inline_summary): Update.
22881 (set_cond_stmt_execution_predicate): Update.
22882 (set_switch_stmt_execution_predicate): Update.
22883 (compute_bb_predicates): Update.
22884 (will_be_nonconstant_expr_predicate): Update.
22885 (will_be_nonconstant_predicate): Update.
22886 (phi_result_unknown_predicate): Update.
22887 (predicate_for_phi_result): Update.
22888 (array_index_predicate): Update.
22889 (estimate_function_body_sizes): Update.
22890 (estimate_node_size_and_time): Update.
22891 (estimate_ipcp_clone_size_and_time): Update.
22892 (remap_predicate): Rename to ...
22893 (predicate::remap_after_inlining): ... this one.
22894 (remap_hint_predicate): Update.
22895 (inline_merge_summary): Update.
22896 (inline_update_overall_summary): Update.
22897 (estimate_size_after_inlining): Update.
22898 (read_predicate): Rename to ...
22899 (predicate::stream_in): ... this one.
22900 (read_inline_edge_summary): Update.
22901 (write_predicate): Rename to ...
22902 (predicate::stream_out): ... this one.
22903 (write_inline_edge_summary): Update.
22904 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22905 (clause_t): Turn to uint32_t
22906 (predicate): Turn to class; implement constructor and operators
22908 (size_time_entry): Update.
22909 (inline_summary): Update.
22910 (inline_edge_summary): Update.
22912 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22914 * fold-const.c (fold_binary_loc): Move transformation...
22915 * match.pd (C - X CMP X): ... here.
22917 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22919 * config/sparc/sparc.c (sparc_option_override): Set function
22920 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22921 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22923 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22925 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22927 2017-05-18 Marek Polacek <polacek@redhat.com>
22930 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22931 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22932 gimple_assign_load_p.
22934 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22936 PR middle-end/80692
22937 * real.c (do_compare): Give decimal_do_compare preference over
22938 comparing just the signs.
22940 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22942 * doc/md.texi (Canonicalization of Instructions): Describe the
22943 canonical form of instructions that inherently set a condition
22946 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22948 PR middle-end/80775
22949 * tree-cfg.c: Move deletion of unreachable case statements to after
22950 the merging of consecutive case labels.
22952 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22954 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22955 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22956 restoring of callee-saved registers.
22958 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22960 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22961 * config/visium/visium.c (single_set_and_flags): Likewise.
22962 * config/visium/visium.md (Substitutions): Likewise.
22964 2017-05-17 Martin Liska <mliska@suse.cz>
22966 * cfg.c: Introduce dump_flags_t type and
22967 use it instead of int type.
22969 * cfghooks.c: Likewise.
22970 * cfghooks.h (struct cfg_hooks): Likewise.
22971 * cfgrtl.c: Likewise.
22972 * cfgrtl.h: Likewise.
22973 * cgraph.c (cgraph_node::get_body): Likewise.
22974 * coretypes.h: Likewise.
22975 * domwalk.c: Likewise.
22976 * domwalk.h: Likewise.
22977 * dumpfile.c (struct dump_option_value_info): Likewise.
22978 (dump_enable_all): Likewise.
22979 (dump_switch_p_1): Likewise.
22980 (opt_info_switch_p): Likewise.
22981 * dumpfile.h (enum tree_dump_index): Likewise.
22982 (struct dump_file_info): Likewise.
22983 * genemit.c: Likewise.
22984 * generic-match-head.c: Likewise.
22985 * gengtype.c (open_base_files): Likewise.
22986 * gimple-pretty-print.c: Likewise.
22987 * gimple-pretty-print.h: Likewise.
22988 * graph.c (print_graph_cfg): Likewise.
22989 * graphite-scop-detection.c (dot_all_sese): Likewise.
22990 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22991 * loop-unroll.c (report_unroll): Likewise.
22992 * passes.c (pass_manager::register_one_dump_file): Likewise.
22993 * print-tree.c: Likewise.
22994 * statistics.c: Likewise.
22995 * tree-cfg.c: Likewise.
22996 * tree-cfg.h: Likewise.
22997 * tree-dfa.c: Likewise.
22998 * tree-dfa.h: Likewise.
22999 * tree-dump.c (dump_function): Likewise.
23000 * tree-dump.h (struct dump_info): Likewise.
23001 * tree-pretty-print.c: Likewise.
23002 * tree-pretty-print.h: Likewise.
23003 * tree-ssa-live.c: Likewise.
23004 * tree-ssa-live.h: Likewise.
23005 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23006 * tree-vect-loop.c: Likewise.
23007 * tree-vect-slp.c: Likewise.
23009 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23010 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23012 PR tree-optimization/80457
23013 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23014 of all arguments to a statement as scalar_to_vec operations.
23015 (vectorizable_call): Adjust call to vect_model_simple_cost for
23017 (vectorizable_conversion): Likewise.
23018 (vectorizable_assignment): Likewise.
23019 (vectorizable_shift): Likewise.
23020 (vectorizable_operation): Likewise.
23021 (vectorizable_comparison): Likewise.
23022 (vect_is_simple_cond): Record the def types for operands.
23023 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23024 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23025 for statement argument count.
23027 2017-05-16 Carl Love <cel@us.ibm.com>
23029 * config/rs6000/rs6000-c: Add support for built-in functions
23030 vector unsigned long long vec_bperm (vector unsigned long long,
23031 vector unsigned char)
23032 vector signed long long vec_mule (vector signed int,
23034 vector unsigned long long vec_mule (vector unsigned int,
23035 vector unsigned int)
23036 vector signed long long vec_mulo (vector signed int,
23038 vector unsigned long long vec_mulo (vector unsigned int,
23039 vector unsigned int)
23040 vector signed char vec_sldw (vector signed char,
23041 vector signed char,
23043 vector unsigned char vec_sldw (vector unsigned char,
23044 vector unsigned char,
23046 vector signed short vec_sldw (vector signed short,
23047 vector signed short,
23049 vector unsigned short vec_sldw (vector unsigned short,
23050 vector unsigned short,
23052 vector signed int vec_sldw (vector signed int,
23055 vector unsigned int vec_sldw (vector unsigned int,
23056 vector unsigned int,
23058 vector signed long long vec_sldw (vector signed long long,
23059 vector signed long long,
23061 vector unsigned long long vec_sldw (vector unsigned long long,
23062 vector unsigned long long,
23064 * config/rs6000/rs6000-c: Add support for built-in functions
23065 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23066 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23067 * doc/extend.texi: Update the built-in documentation for the
23068 new built-in functions.
23070 2017-05-16 Marek Polacek <polacek@redhat.com>
23074 * tree.c (save_expr): Don't fold the expression.
23076 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23078 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23079 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23080 and (?*y,m). Update insn attributes.
23082 2017-05-16 Martin Liska <mliska@suse.cz>
23084 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23085 flags argument of print_gimple_stmt, print_gimple_expr,
23086 print_generic_stmt and print_generic_expr.
23087 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23088 * coretypes.h: Likewise.
23089 * except.c (dump_eh_tree): Likewise.
23090 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23091 * gimple-pretty-print.h: Likewise.
23092 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23093 (backprop::push_to_worklist): Likewise.
23094 (backprop::pop_from_worklist): Likewise.
23095 (backprop::process_use): Likewise.
23096 (backprop::intersect_uses): Likewise.
23097 (note_replacement): Likewise.
23098 * gimple-ssa-store-merging.c
23099 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23100 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23101 (pass_store_merging::execute): Likewise.
23102 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23103 (ssa_base_cand_dump_callback): Likewise.
23104 (dump_incr_vec): Likewise.
23105 (replace_refs): Likewise.
23106 (replace_mult_candidate): Likewise.
23107 (create_add_on_incoming_edge): Likewise.
23108 (create_phi_basis): Likewise.
23109 (insert_initializers): Likewise.
23110 (all_phi_incrs_profitable): Likewise.
23111 (introduce_cast_before_cand): Likewise.
23112 (replace_one_candidate): Likewise.
23113 * gimplify.c (gimplify_expr): Likewise.
23114 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23115 (set_rename): Likewise.
23116 (rename_uses): Likewise.
23117 (copy_loop_phi_nodes): Likewise.
23118 (add_close_phis_to_merge_points): Likewise.
23119 (copy_loop_close_phi_args): Likewise.
23120 (copy_cond_phi_args): Likewise.
23121 (graphite_copy_stmts_from_block): Likewise.
23122 (translate_pending_phi_nodes): Likewise.
23123 * graphite-poly.c (print_pdr): Likewise.
23124 (dump_gbb_cases): Likewise.
23125 (dump_gbb_conditions): Likewise.
23126 (print_scop_params): Likewise.
23127 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23128 (build_cross_bb_scalars_use): Likewise.
23129 (gather_bbs::before_dom_children): Likewise.
23130 * hsa-dump.c (dump_hsa_immed): Likewise.
23131 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23132 (get_replacement_map): Likewise.
23133 * ipa-inline-analysis.c (dump_condition): Likewise.
23134 (estimate_function_body_sizes): Likewise.
23135 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23136 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23137 * ipa-prop.c (ipa_dump_param): Likewise.
23138 (ipa_print_node_jump_functions_for_edge): Likewise.
23139 (ipa_modify_call_arguments): Likewise.
23140 (ipa_modify_expr): Likewise.
23141 (ipa_dump_param_adjustments): Likewise.
23142 (ipa_dump_agg_replacement_values): Likewise.
23143 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23144 * ipa-pure-const.c (check_stmt): Likewise.
23145 (pass_nothrow::execute): Likewise.
23146 * ipa-split.c (execute_split_functions): Likewise.
23147 * omp-offload.c (dump_oacc_loop_part): Likewise.
23148 (dump_oacc_loop): Likewise.
23149 * trans-mem.c (tm_log_emit): Likewise.
23150 (tm_memopt_accumulate_memops): Likewise.
23151 (dump_tm_memopt_set): Likewise.
23152 (dump_tm_memopt_transform): Likewise.
23153 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23154 (print_loop): Likewise.
23155 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23156 (chkp_gather_checks_info): Likewise.
23157 (chkp_get_check_result): Likewise.
23158 (chkp_remove_check_if_pass): Likewise.
23159 (chkp_use_outer_bounds_if_possible): Likewise.
23160 (chkp_reduce_bounds_lifetime): Likewise.
23161 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23162 (chkp_mark_completed_bounds): Likewise.
23163 (chkp_register_incomplete_bounds): Likewise.
23164 (chkp_mark_invalid_bounds): Likewise.
23165 (chkp_maybe_copy_and_register_bounds): Likewise.
23166 (chkp_build_returned_bound): Likewise.
23167 (chkp_get_bound_for_parm): Likewise.
23168 (chkp_build_bndldx): Likewise.
23169 (chkp_get_bounds_by_definition): Likewise.
23170 (chkp_generate_extern_var_bounds): Likewise.
23171 (chkp_get_bounds_for_decl_addr): Likewise.
23172 * tree-chrec.c (chrec_apply): Likewise.
23173 * tree-data-ref.c (dump_data_reference): Likewise.
23174 (dump_subscript): Likewise.
23175 (dump_data_dependence_relation): Likewise.
23176 (analyze_overlapping_iterations): Likewise.
23177 * tree-inline.c (expand_call_inline): Likewise.
23178 (tree_function_versioning): Likewise.
23179 * tree-into-ssa.c (dump_defs_stack): Likewise.
23180 (dump_currdefs): Likewise.
23181 (dump_names_replaced_by): Likewise.
23182 (dump_update_ssa): Likewise.
23183 (update_ssa): Likewise.
23184 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23185 * tree-parloops.c (build_new_reduction): Likewise.
23186 (try_create_reduction_list): Likewise.
23187 (ref_conflicts_with_region): Likewise.
23188 (oacc_entry_exit_ok_1): Likewise.
23189 (oacc_entry_exit_single_gang): Likewise.
23190 * tree-pretty-print.h: Likewise.
23191 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23192 (get_scalar_evolution): Likewise.
23193 (add_to_evolution): Likewise.
23194 (get_loop_exit_condition): Likewise.
23195 (analyze_evolution_in_loop): Likewise.
23196 (analyze_initial_condition): Likewise.
23197 (analyze_scalar_evolution): Likewise.
23198 (instantiate_scev): Likewise.
23199 (number_of_latch_executions): Likewise.
23200 (gather_chrec_stats): Likewise.
23201 (final_value_replacement_loop): Likewise.
23202 (scev_const_prop): Likewise.
23203 * tree-sra.c (dump_access): Likewise.
23204 (disqualify_candidate): Likewise.
23205 (create_access): Likewise.
23206 (reject): Likewise.
23207 (maybe_add_sra_candidate): Likewise.
23208 (create_access_replacement): Likewise.
23209 (analyze_access_subtree): Likewise.
23210 (analyze_all_variable_accesses): Likewise.
23211 (sra_modify_assign): Likewise.
23212 (initialize_constant_pool_replacements): Likewise.
23213 (find_param_candidates): Likewise.
23214 (decide_one_param_reduction): Likewise.
23215 (replace_removed_params_ssa_names): Likewise.
23216 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23217 * tree-ssa-copy.c (dump_copy_of): Likewise.
23218 (copy_prop_visit_cond_stmt): Likewise.
23219 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23220 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23221 (record_equivalences_from_stmt): Likewise.
23222 * tree-ssa-dse.c (compute_trims): Likewise.
23223 (delete_dead_call): Likewise.
23224 (delete_dead_assignment): Likewise.
23225 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23226 (forward_propagate_into_cond): Likewise.
23227 (pass_forwprop::execute): Likewise.
23228 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23229 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23231 (move_computations_worker): Likewise.
23232 (execute_sm): Likewise.
23233 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23234 (remove_exits_and_undefined_stmts): Likewise.
23235 (remove_redundant_iv_tests): Likewise.
23236 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23237 (adjust_iv_update_pos): Likewise.
23238 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23239 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23240 (value_replacement): Likewise.
23241 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23242 * tree-ssa-pre.c (print_pre_expr): Likewise.
23243 (get_representative_for): Likewise.
23244 (create_expression_by_pieces): Likewise.
23245 (insert_into_preds_of_block): Likewise.
23246 (eliminate_insert): Likewise.
23247 (eliminate_dom_walker::before_dom_children): Likewise.
23248 (eliminate): Likewise.
23249 (remove_dead_inserted_code): Likewise.
23250 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23251 * tree-ssa-reassoc.c (get_rank): Likewise.
23252 (eliminate_duplicate_pair): Likewise.
23253 (eliminate_plus_minus_pair): Likewise.
23254 (eliminate_not_pairs): Likewise.
23255 (undistribute_ops_list): Likewise.
23256 (eliminate_redundant_comparison): Likewise.
23257 (update_range_test): Likewise.
23258 (optimize_range_tests_var_bound): Likewise.
23259 (optimize_vec_cond_expr): Likewise.
23260 (rewrite_expr_tree): Likewise.
23261 (rewrite_expr_tree_parallel): Likewise.
23262 (linearize_expr): Likewise.
23263 (break_up_subtract): Likewise.
23264 (linearize_expr_tree): Likewise.
23265 (attempt_builtin_powi): Likewise.
23266 (attempt_builtin_copysign): Likewise.
23267 (transform_stmt_to_copy): Likewise.
23268 (transform_stmt_to_multiply): Likewise.
23269 (dump_ops_vector): Likewise.
23270 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23271 (print_scc): Likewise.
23272 (set_ssa_val_to): Likewise.
23273 (visit_reference_op_store): Likewise.
23274 (visit_use): Likewise.
23275 (sccvn_dom_walker::before_dom_children): Likewise.
23276 (run_scc_vn): Likewise.
23277 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23279 (expr_hash_elt::print): Likewise.
23280 (const_and_copies::pop_to_marker): Likewise.
23281 (const_and_copies::record_const_or_copy_raw): Likewise.
23282 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23283 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23284 (dump_predicates): Likewise.
23285 (find_uninit_use): Likewise.
23286 (warn_uninitialized_phi): Likewise.
23287 (pass_late_warn_uninitialized::execute): Likewise.
23288 * tree-ssa.c (verify_vssa): Likewise.
23289 (verify_ssa): Likewise.
23290 (maybe_optimize_var): Likewise.
23291 * tree-vrp.c (dump_value_range): Likewise.
23292 (dump_all_value_ranges): Likewise.
23293 (dump_asserts_for): Likewise.
23294 (register_edge_assert_for_2): Likewise.
23295 (vrp_visit_cond_stmt): Likewise.
23296 (vrp_visit_switch_stmt): Likewise.
23297 (vrp_visit_stmt): Likewise.
23298 (vrp_visit_phi_node): Likewise.
23299 (simplify_cond_using_ranges_1): Likewise.
23300 (fold_predicate_in): Likewise.
23301 (evrp_dom_walker::before_dom_children): Likewise.
23302 (evrp_dom_walker::push_value_range): Likewise.
23303 (evrp_dom_walker::pop_value_range): Likewise.
23304 (execute_early_vrp): Likewise.
23306 2017-05-16 Richard Biener <rguenther@suse.de>
23308 * dwarf2out.c (loc_list_from_tree_1): Do not create
23309 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23311 2017-05-16 Richard Biener <rguenther@suse.de>
23313 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23315 (note_variable_value_in_expr): If we resolved the decl ref
23316 do not push to the stack.
23318 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23320 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23321 operations in fast-math mode.
23322 (vaddq_f16): Likewise.
23323 (vmul_f16): Likewise.
23324 (vmulq_f16): Likewise.
23325 (vsub_f16): Likewise.
23326 (vsubq_f16): Likewise.
23327 * config/arm/neon.md (add<mode>3): New.
23329 (fma:<VH:mode>3): New. Also remove outdated comment.
23332 2017-05-16 Martin Liska <mliska@suse.cz>
23336 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23337 (odr_types_equivalent_p): Likewise.
23339 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23341 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23343 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23346 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23347 non-interunit SSE move alternatives with '?'.
23348 (zero-extendsidi peephole2): New peephole to skip intermediate
23349 general register in SSE zero-extend sequence.
23351 2017-05-15 Jeff Law <law@redhat.com>
23353 * reorg.c (relax_delay_slots): Create a new variable to hold
23354 the temporary target rather than clobbering TARGET_LABEL.
23356 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23357 missing argument to extract_bit_field call.
23358 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23360 2017-05-15 Martin Liska <mliska@suse.cz>
23363 * gcc.c (process_command): Do not allow empty argument of -o option.
23365 2017-05-15 Renlin Li <renlin.li@arm.com>
23367 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23368 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23369 * config/aarch64/constraints.md (Usf): Add long call check.
23370 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23371 (call_value): Likewise.
23372 (sibcall): Likewise.
23373 (sibcall_value): Likewise.
23375 (call_value_insn): New.
23376 (sibcall_insn): Update rtx pattern.
23377 (sibcall_value_insn): Likewise.
23378 (call_internal): Remove.
23379 (call_value_internal): Likewise.
23380 (sibcall_internal): Likewise.
23381 (sibcall_value_internal): Likewise.
23382 (call_reg): Likewise.
23383 (call_symbol): Likewise.
23384 (call_value_reg): Likewise.
23385 (call_value_symbol): Likewise.
23387 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23390 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23392 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23394 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23395 compatible with CCGOCmode and with CCZmode.
23397 2017-05-14 Martin Sebor <msebor@redhat.com>
23399 PR middle-end/77671
23400 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23401 (gimple_fold_builtin_snprintf): Same.
23402 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23403 (gimple_fold_builtin_snprintf): Same.
23404 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23405 of character types.
23406 (is_call_safe): New function.
23407 (try_substitute_return_value): Call it.
23408 (try_simplify_call): New function.
23409 (pass_sprintf_length::handle_gimple_call): Call it.
23411 2017-05-14 Martin Sebor <msebor@redhat.com>
23413 PR middle-end/80669
23414 * builtins.c (expand_builtin_stpncpy): Simplify.
23416 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23418 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23419 * config/i386/i386.h
23420 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23421 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23422 (struct machine_function): Add new members call_ms2sysv,
23423 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23424 (struct machine_frame_state): New fields sp_realigned and
23425 sp_realigned_offset.
23426 * config/i386/i386.c
23427 (enum xlogue_stub): New enum.
23428 (enum xlogue_stub_sets): New enum.
23429 (class xlogue_layout): New class.
23430 (struct ix86_frame): New fields stack_realign_allocate_offset,
23431 stack_realign_offset and outlined_save_offset. Modify comments to
23432 detail stack layout when using out-of-line stubs.
23433 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23434 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23435 -mcall-ms2sysv-xlogues.
23436 (stub_managed_regs): New static variable.
23437 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23438 registers managed by out-of-line stub.
23439 (disable_call_ms2sysv_xlogues): New function.
23440 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23441 m->call_ms2sysv when appropriate and compute frame layout for
23443 (sp_valid_at, fp_valid_at): New inline functions.
23444 (choose_basereg): New function.
23445 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23447 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23448 Use align parameter of choose_baseaddr to generated aligned SSE movs
23450 (pro_epilogue_adjust_stack): Modify to track
23451 machine_frame_state::sp_realigned.
23452 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23453 (ix86_nsaved_sseregs): Likewise.
23454 (ix86_emit_save_regs): Likewise.
23455 (ix86_emit_save_regs_using_mov): Likewise.
23456 (ix86_emit_save_sse_regs_using_mov): Likewise.
23457 (get_scratch_register_on_entry): Likewise.
23458 (gen_frame_set): New function.
23459 (gen_frame_load): Likewise.
23460 (gen_frame_store): Likewise.
23461 (emit_outlined_ms2sysv_save): Likewise.
23462 (emit_outlined_ms2sysv_restore): Likewise.
23463 (ix86_expand_prologue): Modify stack re-alignment code and call
23464 emit_outlined_ms2sysv_save when appropriate.
23465 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23466 parameter rtx_insn *insn, which allows the function to be used to only
23467 generate the notes.
23468 (ix86_expand_epilogue): Modify validity checks of frame and stack
23469 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23470 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23471 * config/i386/predicates.md
23472 (save_multiple): New predicate.
23473 (restore_multiple): Likewise.
23474 * config/i386/sse.md
23475 (save_multiple<mode>): New pattern.
23476 (save_multiple_realign<mode>): Likewise.
23477 (restore_multiple<mode>): Likewise.
23478 (restore_multiple_and_return<mode>): Likewise.
23479 (restore_multiple_leave_return<mode>): Likewise.
23480 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23482 2017-05-14 Julia Koval <julia.koval@intel.com>
23484 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23485 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23486 (__builtin_ia32_xsetbv): New builtins.
23487 * config/i386/i386.c (ix86_expand_special_args_builtin):
23489 (ix86_expand_builtin): Special expand for new intrinsics.
23490 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23491 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23492 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23494 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23496 * cfganal.c (inverted_post_order_compute): Change argument type
23498 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23499 * df-core.c (rest_of_handle_df_initialize): Adjust.
23500 (rest_of_handle_df_finish): Likewise.
23501 (df_analyze_1): Likewise.
23502 (df_analyze): Likewise.
23503 (loop_inverted_post_order_compute): Change argument to be a vec *.
23504 (df_analyze_loop): Adjust.
23505 (df_get_n_blocks): Likewise.
23506 (df_get_postorder): Likewise.
23507 * df.h (struct df_d): Change field to be a vec.
23508 * lcm.c (compute_laterin): Adjust.
23509 (compute_available): Likewise.
23510 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23511 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23512 * tree-ssa-pre.c (compute_antic): Likewise.
23514 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23516 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23517 (depth_first_search::depth_first_search): Change structure init
23518 function to this constructor.
23519 (depth_first_search::add_bb): Rename function to this member.
23520 (depth_first_search::execute): Likewise.
23521 (flow_dfs_compute_reverse_finish): Adjust.
23523 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23525 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23526 (longest_simple_path): Likewise.
23527 * shrink-wrap.c (spread_components): Likewise.
23528 (disqualify_problematic_components): Likewise.
23529 (emit_common_heads_for_components): Likewise.
23530 (emit_common_tails_for_components): Likewise.
23531 (insert_prologue_epilogue_for_components): Likewise.
23533 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23535 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23538 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23540 * df-core.c (df_set_blocks): Start using auto_bitmap.
23541 (df_compact_blocks): Likewise.
23542 * df-problems.c (df_rd_confluence_n): Likewise.
23543 * df-scan.c (df_insn_rescan_all): Likewise.
23544 (df_process_deferred_rescans): Likewise.
23545 (df_update_entry_block_defs): Likewise.
23546 (df_update_exit_block_uses): Likewise.
23547 (df_entry_block_bitmap_verify): Likewise.
23548 (df_exit_block_bitmap_verify): Likewise.
23549 (df_scan_verify): Likewise.
23550 * lra-constraints.c (lra_constraints): Likewise.
23551 (undo_optional_reloads): Likewise.
23552 (lra_undo_inheritance): Likewise.
23553 * lra-remat.c (calculate_gen_cands): Likewise.
23554 (do_remat): Likewise.
23555 * lra-spills.c (assign_spill_hard_regs): Likewise.
23556 (spill_pseudos): Likewise.
23557 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23558 (bitmap_set_subtract_values): Likewise.
23560 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23562 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23563 management with auto_bitmap.
23564 (fix_inter_tick): Likewise.
23565 (fix_recovery_deps): Likewise.
23566 * ira.c (add_store_equivs): Likewise.
23567 (find_moveable_pseudos): Likewise.
23568 (split_live_ranges_for_shrink_wrap): Likewise.
23569 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23570 (rtx_reuse_manager::seen_def_p): Likewise.
23571 (rtx_reuse_manager::set_seen_def): Likewise.
23572 * print-rtl.h (class rtx_reuse_manager): Likewise.
23574 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23576 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23578 (migrate_btr_def): Likewise.
23579 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23580 * df-core.c (loop_post_order_compute): Likewise.
23581 (loop_inverted_post_order_compute): Likewise.
23582 * hsa-common.h: Likewise.
23583 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23584 * init-regs.c (initialize_uninitialized_regs): Likewise.
23585 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23586 (inline_small_functions): Likewise.
23587 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23588 * ira.c (combine_and_move_insns): Likewise.
23589 (build_insn_chain): Likewise.
23590 * loop-invariant.c (find_invariants): Likewise.
23591 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23592 * predict.c (tree_predict_by_opcode): Likewise.
23593 (predict_paths_leading_to): Likewise.
23594 (predict_paths_leading_to_edge): Likewise.
23595 (estimate_loops_at_level): Likewise.
23596 (estimate_loops): Likewise.
23597 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23598 (spread_components): Likewise.
23599 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23600 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23601 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23602 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23603 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23604 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23605 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23606 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23607 (mark_threaded_blocks): Likewise.
23608 (thread_through_all_blocks): Likewise.
23609 * tree-ssa.c (verify_ssa): Likewise.
23610 (execute_update_addresses_taken): Likewise.
23611 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23613 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23615 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23617 (post_order_compute): Likewise.
23618 (inverted_post_order_compute): Likewise.
23619 (pre_and_rev_post_order_compute_fn): Likewise.
23621 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23623 * genrecog.c (int_set::int_set): Explicitly construct our
23624 auto_vec base class.
23625 * vec.h (auto_vec::auto_vec): New constructor.
23627 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23629 * bitmap.h (class auto_bitmap): New constructor taking
23630 bitmap_obstack * argument.
23632 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23634 * bitmap.h (class auto_bitmap): Change type of m_bits to
23635 bitmap_head, and adjust ctor / dtor and member operators.
23637 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23639 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23640 when returned register mode doesn't match original mode.
23642 2017-05-12 Jeff Law <law@redhat.com>
23643 Jakub Jelinek <jakub@redhat.com>
23645 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23646 we look for cc setter after the compare-elim changes.
23647 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23648 within the vector to match what compare-elim now expects.
23649 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23650 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23652 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23653 after the compare-elim changes.
23654 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23655 the vector to match what compare-elim now expects.
23656 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23657 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23658 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23659 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23660 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23662 * config/visium/visium.c (single_set_and_flags): Fix where
23663 we look for cc setter after the compare-elim changes.
23664 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23665 with the vector to match what compare-elim now expects.
23666 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23667 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23668 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23669 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23670 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23671 (neg<mode>2_insn_set_overflow): Likewise.
23673 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23675 PR middle-end/79794
23676 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23677 maybe_expand_insn call, set ops[0].target. If still set after call,
23678 set alt_rtl. Add extra arg to recursive calls.
23679 (extract_bit_field): Add alt_rtl argument. Pass to
23681 * expmed.h (extract_bit_field): Fix prototype.
23682 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23683 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23684 to extract_bit_field_calls.
23685 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23686 Pass alt_rtl to extract_bit_field calls.
23687 * calls.c (store_unaligned_arguments_into_psuedos)
23688 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23689 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23691 * optabs.h (struct expand_operand): Add target bitfield.
23693 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23695 * compare-elim.c (try_eliminate_compare): Canonicalize
23696 operation with embedded compare to
23697 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23698 (set (reg) (operation)].
23700 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23702 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23705 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23706 cost of adding a carry flag for ADC instruction.
23707 [case MINUS]: Ignore the cost of subtracting a carry flag
23708 for SBB instruction.
23710 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23712 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23714 * config/rs6000/bmiintrin.h: New file.
23715 * config/rs6000/bmi2intrin.h: New file.
23716 * config/rs6000/x86intrin.h: New file.
23718 2017-05-12 Jeff Law <law@redhat.com>
23720 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23723 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23725 PR middle-end/80707
23726 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23728 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23730 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23731 early expansion of vector divide builtins.
23732 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23733 builtins identified as having unsigned arguments.
23735 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23737 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23738 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23739 expansion of vector logical operations (and, andc, or, xor,
23742 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23744 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23745 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23747 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23749 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23750 early GIMPLE expansion of vector multiplies.
23752 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23754 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23755 TARGET_HAVE_MOVT conditional.
23756 (movt splitter): Likewise.
23758 2017-05-12 Richard Biener <rguenther@suse.de>
23760 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23761 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23762 Fold all stmts not inplace.
23764 2017-05-12 Richard Biener <rguenther@suse.de>
23766 PR tree-optimization/80713
23767 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23768 inserted_exprs bit for not removed stmts.
23770 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23772 PR middle-end/69921
23773 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23774 parallelized" attribute for parallelized OpenACC kernels.
23775 * omp-offload.c (execute_oacc_device_lower): Use it.
23777 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23778 Set "oacc kernels" attribute.
23779 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23780 parameter. Adjust all users.
23781 (oacc_fn_attrib_kernels_p): Remove function.
23782 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23783 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23784 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23785 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23786 assert "oacc kernels" attribute is set.
23788 2017-05-11 Carl Love <cel@us.ibm.com>
23790 * config/rs6000/rs6000-c: Add support for built-in functions
23791 vector unsigned char vec_popcnt (vector signed char)
23792 vector unsigned char vec_popcnt (vector unsigned char)
23793 vector unsigned short vec_popcnt (vector signed short)
23794 vector unsigned short vec_popcnt (vector unsigned short)
23795 vector unsigned int vec_popcnt (vector signed int)
23796 vector unsigned int vec_popcnt (vector unsigned int)
23797 vector unsigned long long vec_popcnt (vector signed long long)
23798 vector unsigned long long vec_popcnt (vector unsigned long long)
23799 vector signed long long vec_slo (vector signed long long,
23800 vector signed char)
23801 vector signed long long vec_slo (vector signed long long,
23802 vector unsigned char)
23803 vector unsigned long long vec_slo (vector unsigned long long,
23804 vector signed char)
23805 vector unsigned long long vec_slo (vector unsigned long long,
23806 vector unsigned char)
23807 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23808 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23809 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23810 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23811 * doc/extend.texi: Update the built-in documentation file for the
23812 new built-in functions.
23814 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23816 * attribs.h (sorted_attr_string): Move machine independent
23817 functions for target clone support from the i386 port to common
23818 code. Rename ix86_function_versions to common_function_versions.
23819 Rename make_name to make_unique_name.
23820 (common_function_versions): Likewise.
23821 (make_unique_name): Likewise.
23822 (make_dispatcher_decl): Likewise.
23823 (is_function_default_version): Likewise.
23824 * attribs.c (attr_strcmp): Likewise.
23825 (sorted_attr_string): Likewise.
23826 (common_function_versions): Likewise.
23827 (make_unique_name): Likewise.
23828 (make_dispatcher_decl): Likewise.
23829 (is_function_default_version): Likewise.
23830 * config/i386/i386.c (attr_strcmp): Likewise.
23831 (sorted_attr_string): Likewise.
23832 (ix86_function_versions): Likewise.
23833 (make_name): Likewise.
23834 (make_dispatcher_decl): Likewise.
23835 (is_function_default_version): Likewise.
23836 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23838 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23841 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23842 Account for direct move costs for vec_construct of integer
23845 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23848 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23849 (UNSPEC_STX_ATOMIC): Ditto.
23850 (loaddi_via_sse): New insn.
23851 (storedi_via_sse): Ditto.
23852 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23853 Update corresponding peephole2 patterns.
23854 (atomic_storedi_fpu): Ditto.
23856 2017-05-11 Julia Koval <julia.koval@intel.com>
23858 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23859 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23861 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23862 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23863 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23865 2017-05-11 Nathan Sidwell <nathan@acm.org>
23867 * graphite-poly.c: Include dumpfile.h.
23869 * dumpfle.h (dump_function): Declare here ...
23870 * tree-dump.h (dump_function): ... not here.
23871 * dumpfile.c: #include tree-cfg.h.
23872 (dump_function): Move here from ...
23873 * tree-dump.c (dump_function): ... here.
23874 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23875 * graphite-poly.c: Don't include tree-dump.h.
23876 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23877 * print-tree.c: Likewise.
23878 * stor-layout.c: Likewise.
23879 * tree-nested.c: Likewise.
23881 * dumpfile.c (dump_start): Use TDF_FLAGS.
23882 (dump_enable_all): Fix TDF_KIND check thinko.
23884 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23886 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23887 array entries to represent two legal parameterizations of the
23888 overloaded __builtin_cmpb function, as represented by the
23889 P6_OV_BUILTIN_CMPB constant.
23890 (altivec_resolve_overloaded_builtin): Add special case handling
23891 for the __builtin_cmpb function, as represented by the
23892 P6_OV_BUILTIN_CMPB constant.
23893 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23894 (BU_P6_64BIT_2): New macro.
23895 (BU_P6_OVERLOAD_2): New macro
23896 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23897 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23898 (CMPB): Add overload support to represent both 32-bit and 64-bit
23899 compare-bytes function.
23900 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23901 support for TARGET_CMPB.
23902 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23903 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23904 documentation of the __builtin_cmpb overloaded built-in function.
23906 2017-05-11 Richard Biener <rguenther@suse.de>
23908 PR tree-optimization/80705
23909 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23910 bases are not vectorizable.
23912 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23914 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23915 when counting register pressure.
23917 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23919 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23920 (iv_ca_more_deps): Renamed to ...
23921 (iv_ca_compare_deps): ... this.
23922 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23924 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23926 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23928 (determine_group_iv_costs): ... here.
23929 (find_inv_vars_cb): Record inv var if it's not recorded before.
23931 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23933 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23934 (get_shiftadd_cost): Ditto.
23936 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23938 * tree-ssa-address.c: Include header file.
23939 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23941 (add_to_parts): Refactor.
23942 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23943 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23946 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23948 PR tree-optimization/53090
23949 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23951 (extract_cond_operands): Detect condition with IV on both sides
23952 and return COMP_IV_EXPR_2.
23953 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23954 (rewrite_use_compare): Simplify by removing call to function
23955 extract_cond_operands.
23957 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23959 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23960 (extract_cond_operands): Detect condition comparing against non-
23961 invariant bound and return appropriate enum value.
23962 (find_interesting_uses_cond): Update use of extract_cond_operands.
23963 Handle its return value accordingly.
23964 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23966 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23968 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23969 nonlinear iv_use computation in loop invariant sensitive way.
23971 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23973 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23974 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23976 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23978 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23979 (dump_cand): Support iv_cand.inv_exprs.
23980 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23982 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23985 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23987 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23989 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23990 as local function. Include necessary header files.
23991 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23993 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23995 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23997 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23999 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24000 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24001 RSHIFT_EXPR and BIT_NOT_EXPR.
24003 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24005 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24006 (adjust_setup_cost): New parameter supporting round up adjustment.
24007 (struct address_cost_data): Delete.
24008 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24009 (split_address_cost, ptr_difference_cost): Delete.
24010 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24011 (struct ainc_cost_data): New struct.
24012 (get_address_cost_ainc): New function.
24013 (get_address_cost, get_computation_cost): Reimplement.
24014 (determine_group_iv_cost_address): Record inv_expr for all uses of
24016 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24017 (iv_ca_has_deps): Reimplemented to ...
24018 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24020 (iv_ca_extend): Call iv_ca_more_deps.
24022 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24024 * tree-ssa-address.c (struct mem_address): Move to header file.
24025 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24026 * tree-ssa-address.h (struct mem_address): Move from C file.
24027 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24029 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24031 * tree-affine.h (aff_combination_type): New interface.
24032 (aff_combination_zero_p): Remove static.
24033 (aff_combination_const_p): New interface.
24034 (aff_combination_singleton_var_p): New interfaces.
24036 2017-05-11 Richard Biener <rguenther@suse.de>
24038 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24039 Skip unreachable blocks and destinations.
24040 (eliminate): Move stmt removal and fixup ...
24041 (fini_eliminate): ... here. Skip inserted exprs.
24042 (pass_pre::execute): Move fini_pre after fini_eliminate.
24043 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24044 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24045 PRE to get rid of dead code that has invalid SSA form and
24046 split critical edges again.
24048 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24050 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24052 2017-05-11 Richard Biener <rguenther@suse.de>
24054 * passes.c (execute_function_todo): Verify loops if they are
24055 said to be up-to-date.
24056 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24057 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24059 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24062 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24063 handle calling assemble_external ourself.
24066 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24067 modes with zero size. Enhance comment.
24069 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24071 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24072 built-ins for vec_xl and vec_xst with short and char pointer
24075 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24077 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24078 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24079 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24080 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24081 (_mm_maskz_min_round_ss): New intrinsics.
24082 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24083 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24084 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24085 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24086 (__builtin_ia32_minss_mask_round): New builtins.
24087 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24088 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24089 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24091 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24092 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24094 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24097 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24099 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24100 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24101 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24102 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24103 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24104 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24105 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24106 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24107 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24108 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24109 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24110 (__builtin_ia32_mulss_mask_round): New builtins.
24111 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24112 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24113 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24115 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24116 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24118 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24121 2017-05-10 Julia Koval <julia.koval@intel.com>
24123 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24124 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24125 (_mm256_setr_m128i): New intrinsics.
24127 2017-05-10 Julia Koval <julia.koval@intel.com>
24129 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24130 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24131 (_mm_maskz_rcp14_ss): New intrinsics.
24132 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24133 (__builtin_ia32_rcp14ss_mask): New builtins.
24134 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24136 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24138 PR tree-optimization/51513
24139 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24140 (assert_unreachable_fallthru_edge_p): Use it.
24141 (group_case_labels_stmt): Likewise.
24142 * tree-cfg.h: Prototype it.
24143 * stmt.c: Include cfghooks.h and tree-cfg.h.
24144 (emit_case_dispatch_table) <gap_label>: New local variable.
24145 Use it to fill dispatch table gaps.
24146 Test for default_label before updating probabilities.
24147 (expand_case) <default_label>: Remove unneeded initialization.
24148 Test for unreachable default case statement and remove its edge.
24149 Set default_label accordingly.
24150 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24152 2017-05-10 Carl Love <cel@us.ibm.com>
24154 * config/rs6000/rs6000-c: Add support for built-in functions
24155 vector signed char vec_neg (vector signed char)
24156 vector signed short int vec_neg (vector short int)
24157 vector signed int vec_neg (vector signed int)
24158 vector signed long long vec_neg (vector signed long long)
24159 vector float vec_neg (vector float)
24160 vector double vec_neg (vector double)
24161 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24163 * config/rs6000/altivec.h: Add define for vec_neg
24164 * doc/extend.texi: Update the built-in documentation for the
24165 new built-in functions.
24167 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24169 PR tree-optimization/77644
24170 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24172 2017-05-10 Nathan Sidwell <nathan@acm.org>
24174 * dumpfile.h (TDI_lang_all): New.
24175 (TDF_KIND): New. Renumber others
24176 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24178 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24180 (get_dump_file_name): Adjust suffix generation.
24181 (dump_enable_all): Use TDF_KIND.
24182 * doc/invoke.texi (-fdump-lang-all): Document.
24184 * dumpfile.h: Tabify.
24186 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24189 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24190 Move member access before delete.
24192 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24194 * tree-inline.c (expand_call_inline): Split block at stmt
24197 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24200 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24201 are now unused after splitting mov{sf,sd}_hardfloat.
24202 (f32_lr2): Likewise.
24203 (f32_lm): Likewise.
24204 (f32_lm2): Likewise.
24205 (f32_li): Likewise.
24206 (f32_li2): Likewise.
24207 (f32_lv): Likewise.
24208 (f32_sr): Likewise.
24209 (f32_sr2): Likewise.
24210 (f32_sm): Likewise.
24211 (f32_sm2): Likewise.
24212 (f32_si): Likewise.
24213 (f32_si2): Likewise.
24214 (f32_sv): Likewise.
24215 (f32_dm): Likewise.
24216 (f32_vsx): Likewise.
24217 (f32_av): Likewise.
24218 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24219 For movsf, order stores so the VSX stores occur before the GPR
24220 store which encourages the register allocator to use a traditional
24221 FPR instead of a GPR. For movsd, order the stores so that the GPR
24222 store comes before the VSX stores to allow the power6 to work.
24223 This is due to the power6 not having a 32-bit integer store
24224 instruction from a FPR.
24225 (movsf_hardfloat): Likewise.
24226 (movsd_hardfloat): Likewise.
24228 2017-05-09 Martin Sebor <msebor@redhat.com>
24230 PR translation/80280
24231 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24234 PR translation/80280
24235 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24236 data member added in r247778.
24237 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24239 2017-05-09 Nathan Sidwell <nathan@acm.org>
24241 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24243 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24246 2017-05-09 Marek Polacek <polacek@redhat.com>
24248 * doc/invoke.texi: Fix typo.
24250 2017-05-09 Richard Biener <rguenther@suse.de>
24252 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24253 (vrp_val_is_min): Likewise.
24254 (set_value_range_to_value): Likewise.
24255 (set_value_range_to_nonnegative): Likewise.
24256 (gimple_assign_nonzero_p): Likewise.
24257 (gimple_stmt_nonzero_p): Likewise.
24258 (vrp_int_const_binop): Likewise. Remove unreachable case.
24259 (adjust_range_with_scev): Adjust comments.
24260 (compare_range_with_value): Likewise.
24261 (extract_range_from_phi_node): Likewise.
24262 (test_for_singularity): Likewise.
24264 2017-05-09 Richard Biener <rguenther@suse.de>
24266 * tree-vrp.c (get_single_symbol): Add assert that we don't
24267 get overflowed constants as invariant part.
24268 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24269 checks. Use wi::cmp instead of recursing for integer constants.
24270 (compare_values): Just ignore whether we assumed undefined
24271 overflow instead of failing the compare.
24272 (extract_range_for_var_from_comparison_expr): Add comment before the
24273 TREE_NO_WARNING sets.
24274 (test_for_singularity): Likewise.
24275 (extract_range_from_comparison): Do not disable optimization
24276 when we assumed undefined overflow.
24277 (extract_range_basic): Remove init of unused var.
24279 2017-05-09 Richard Biener <rguenther@suse.de>
24281 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24282 (extract_range_from_multiplicative_op_1): Adjust.
24283 (extract_range_from_binary_expr_1): Use int_const_binop.
24285 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24288 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24289 rs6000_store_data_bypass_p in seven define_bypass directives and
24290 in several comments.
24291 * config/rs6000/rs6000-protos.h: Add prototype for
24292 rs6000_store_data_bypass_p function.
24293 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24294 function implements slightly different (rs6000-specific) semantics
24295 than store_data_bypass_p, returning false rather than aborting
24296 with assertion error when arguments do not satisfy the
24297 requirements of store data bypass.
24298 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24299 rs6000_store_data_bypass_p.
24301 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24303 * config/xtensa/xtensa-protos.h
24304 (xtensa_initial_elimination_offset): New declaration.
24305 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24306 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24307 macro definition, add case for FRAME_POINTER_REGNUM when
24308 FRAME_GROWS_DOWNWARD.
24309 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24310 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24311 xtensa_initial_elimination_offset.
24313 2017-05-08 Nathan Sidwell <nathan@acm.org>
24315 * doc/invoke.texi: Alphabetize -fdump options.
24317 2017-05-08 Martin Sebor <msebor@redhat.com>
24319 PR translation/80280
24320 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24322 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24324 * target.def (compute_frame_layout): New optional target hook.
24325 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24326 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24327 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24329 * reload1.c (verify_initial_elim_offsets): Likewise.
24330 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24331 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24332 (arm_get_frame_offsets): Split up into this ...
24333 (arm_compute_frame_layout): ... and this function.
24335 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24337 * config/aarch64/constraints.md (Usa): New constraint.
24338 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24340 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24342 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24343 with_multilib_list after it has been checked.
24345 2017-05-08 Richard Biener <rguenther@suse.de>
24347 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24348 (bitmap_set_subtract_values): Likewise.
24350 2017-05-08 Richard Biener <rguenther@suse.de>
24352 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24353 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24355 (gimple_stmt_nonzero_warnv_p): Rename to ...
24356 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24358 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24359 (extract_range_basic): Adjust, do not disable propagation on
24360 strict overflow sensitive simplification.
24361 (vrp_visit_cond_stmt): Likewise.
24363 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24365 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24366 body size unconditionally.
24368 2017-05-07 Jeff Law <law@redhat.com>
24371 2017-05-06 Jeff Law <law@redhat.com>
24372 PR tree-optimization/78496
24373 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24376 PR tree-optimization/78496
24377 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24378 (simplify_stmt_using_ranges): Call it.
24379 (vrp_dom_walker::before_dom_children): Extract equivalences
24380 from an ASSERT_EXPR with an equality comparison against a
24383 2017-05-06 Jeff Law <law@redhat.com>
24385 PR tree-optimization/78496
24386 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24389 PR tree-optimization/78496
24390 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24391 (simplify_stmt_using_ranges): Call it.
24392 (vrp_dom_walker::before_dom_children): Extract equivalences
24393 from an ASSERT_EXPR with an equality comparison against a
24396 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24398 * lra-constraints.c (lra_copy_reg_equiv): New function.
24399 (split_reg): Use it to copy equivalence information from the
24400 original register to the spill register.
24402 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24404 PR rtl-optimization/75964
24405 * simplify-rtx.c (simplify_const_relational_operation): Remove
24406 invalid handling of comparisons of integer ABS.
24408 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24410 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24411 initialize to zero.
24412 (init_regs): Remove declaration.
24413 (function_arg_advance_32): Initialize error_p as boolean variable.
24415 2017-05-05 Nathan Sidwell <nathan@acm.org>
24417 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24418 lines. Use for (;;).
24420 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24422 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24423 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24424 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24425 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24426 VF=2 that require versioning.
24428 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24430 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24433 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24435 * diagnostic.h (diagnostic_override_option_index): Convert from
24436 macro to inline function.
24438 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24440 * diagnostic.c (last_module_changed_p): New function.
24441 (set_last_module): New function.
24442 (diagnostic_report_current_module): Convert macro usage to
24443 the above functions.
24444 * diagnostic.h (diagnostic_context::last_module): Strengthen
24445 from const line_map * to const line_map_ordinary *.
24446 (diagnostic_last_module_changed): Delete macro.
24447 (diagnostic_set_last_module): Delete macro.
24449 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24451 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24452 with diagnostic_report_diagnostic.
24453 (diagnostic_n_impl_richloc): Likewise.
24454 * diagnostic.h (report_diagnostic): Delete macro.
24455 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24456 with diagnostic_report_diagnostic.
24457 * substring-locations.c (format_warning_va): Likewise.
24459 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24461 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24462 save/restor of format_spec. Move option-printing code to...
24463 (print_option_information): ...this new function, and
24464 reimplement by simply printing to the pretty_printer,
24465 rather than appending to the format string.
24467 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24469 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24470 handling logic into...
24471 (update_effective_level_from_pragmas): ...this new function.
24473 2017-05-04 Andrew Waterman <andrew@sifive.com>
24475 * config/riscv/riscv.opt (mstrict-align): New option.
24476 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24477 (SLOW_UNALIGNED_ACCESS): Define.
24478 (riscv_slow_unaligned_access): Declare.
24479 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24481 (riscv_slow_unaligned_access): New variable.
24482 (rocket_tune_info): Set slow_unaligned_access to true.
24483 (optimize_size_tune_info): Set slow_unaligned_access to false.
24484 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24485 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24486 (riscv_option_override): Set riscv_slow_unaligned_access.
24487 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24489 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24491 * config/riscv/riscv.md: Unify indentation.
24493 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24498 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24500 (extendsi<mode>2): Add support for doing sign extension via
24501 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24502 don't have ISA 3.0 instructions.
24503 (extendsi<mode>2 splitter): Likewise.
24504 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24505 generate the normal insns since SImode can now go in vector
24506 registers. Disallow the special UNSPECs needed for previous
24507 machines to hide SImode being used. Add new insns
24508 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24509 (fix_trunc<mode>si2_stfiwx): Likewise.
24510 (fix_trunc<mode>si2_internal): Likewise.
24511 (fixuns_trunc<mode>si2): Likewise.
24512 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24513 (fctiw<u>z_<mode>_smallint): Likewise.
24514 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24515 of floating point to 32-bit integer from doing a direct move to
24516 the GPR registers to do a store.
24517 (fctiwz_<mode>): Break long line.
24519 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24521 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24522 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24523 (addr_list, addr_offset_valid_p): New.
24524 (split_address_groups): Check offset validity with above function.
24525 (gt-tree-ssa-loop-ivopts.h): Include header file.
24527 2017-05-05 Nathan Sidwell <nathan@acm.org>
24529 * config.gcc (arm*-*-*): Add missing 'fi'.
24531 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24533 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24534 included in -fopt-info does not matter.
24535 * doc/optinfo.texi (-fopt-info): Fix description of default
24536 behavour. Explicitly say order of options included in -fopt-info
24539 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24541 * config.gcc: Allow combinations of aprofile and rmprofile values for
24542 --with-multilib-list.
24543 * config/arm/t-multilib: New file.
24544 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24545 variables. Remove setting of ISA and floating-point ABI in
24546 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24547 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24548 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24549 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24551 * config/arm/t-rmprofile: Likewise except for the matches changes.
24552 * doc/install.texi (--with-multilib-list): Document the combination of
24553 aprofile and rmprofile values and warn about pitfalls in doing that.
24555 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24557 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24558 (movdi_aarch64): Likewise.
24560 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24562 PR tree-optimization/80632
24563 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24565 (build_arrays): Initialize it for virtual phis.
24566 (fix_phi_nodes): Use it for virtual phis.
24568 PR tree-optimization/80558
24569 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24570 [x, y] op z into [x op, y op z] for op & or | if conditions
24573 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24574 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24577 * config/arm/arm.md (use_literal_pool): Remove.
24578 (64-bit immediate split): No longer takes cost into consideration
24579 if arm_disable_literal_pool is enabled.
24580 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24581 used when arm_disable_literal_pool is enabled.
24582 (arm_max_const_double_inline_cost): Remove use of
24583 arm_disable_literal_pool.
24584 (push_minipool_fix): Add assert.
24585 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24586 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24587 (no_literal_pool_sf_immediate): New.
24589 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24591 PR tree-optimization/80613
24592 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24593 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24595 2017-05-05 Richard Biener <rguenther@suse.de>
24597 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24599 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24601 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24602 of this flag from insn conditions due to removal from r247495.
24604 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24606 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24608 (arm_early_store_addr_dep_ptr): Likewise.
24609 * config/arm/aarch-common-protos.h
24610 (arm_early_load_addr_dep_ptr): Add prototype.
24611 (arm_early_store_addr_dep_ptr): Likewise.
24612 * config/arm/cortex-a53.md: Add new bypasses.
24614 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24616 * tree.c (next_type_uid): Change type to unsigned.
24617 (type_hash_canon): Decrement back next_type_uid if
24618 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24619 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24622 2017-05-04 Martin Sebor <msebor@redhat.com>
24624 * builtins.c: Fix a trivial typo in a comment.
24626 PR middle-end/79234
24627 * builtins.c (check_sizes): Adjust to handle reading past the end.
24628 Avoid printing excessive upper bound of ranges. Use %E to print
24629 tree nodes instead of converting them to %wu.
24630 (expand_builtin_memchr): New function.
24631 (compute_dest_size): Rename...
24632 (compute_objsize): ...to this.
24633 (expand_builtin_memcpy): Adjust.
24634 (expand_builtin_mempcpy): Adjust.
24635 (expand_builtin_strcat): Adjust.
24636 (expand_builtin_strcpy): Adjust.
24637 (check_strncat_sizes): Adjust.
24638 (expand_builtin_strncat): Adjust.
24639 (expand_builtin_strncpy): Adjust and simplify.
24640 (expand_builtin_memset): Adjust.
24641 (expand_builtin_bzero): Adjust.
24642 (expand_builtin_memcmp): Adjust.
24643 (expand_builtin): Handle memcmp.
24644 (maybe_emit_chk_warning): Check strncat just once.
24646 2017-05-04 Martin Sebor <msebor@redhat.com>
24648 PR preprocessor/79214
24649 PR middle-end/79222
24650 PR middle-end/79223
24651 * builtins.c (check_sizes): Add inlining context and issue
24652 warnings even when -Wno-system-headers is set.
24653 (check_strncat_sizes): Same.
24654 (expand_builtin_strncat): Same.
24655 (expand_builtin_memmove): New function.
24656 (expand_builtin_stpncpy): Same.
24657 (expand_builtin): Handle memmove and stpncpy.
24659 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24661 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24662 which is not used any more.
24664 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24666 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24668 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24670 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24671 (cortexa53_tunings): Likewise.
24672 (cortexa57_tunings): Likewise.
24673 (cortexa72_tunings): Likewise.
24674 (cortexa73_tunings): Likewise.
24676 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24678 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24679 Set loop alignment to 8.
24681 2017-05-04 Martin Sebor <msebor@redhat.com>
24683 PR translation/80280
24684 * builtins.c (expand_builtin_object_size): Add missing quoting to
24685 %D and like directives.
24686 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24687 (hsa_type_for_tree_type): Same.
24688 (verify_function_arguments): Same.
24689 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24690 * varasm.c (get_section): Same.
24693 2017-05-04 Martin Sebor <msebor@redhat.com>
24695 PR translation/80280
24696 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24698 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24700 * config/aarch64/aarch64.c (generic_addrcost_table):
24701 Change HI/TI mode setting.
24703 2017-05-04 Martin Jambor <mjambor@suse.cz>
24705 PR tree-optimization/80622
24706 * tree-sra.c (comes_initialized_p): New function.
24707 (build_accesses_from_assign): Only set write lazily when
24708 comes_initialized_p is false.
24709 (analyze_access_subtree): Use comes_initialized_p.
24710 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24711 instead of testing for PARM_DECL.
24713 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24715 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24716 constraint on operand 0 to allow more general addressing modes.
24717 Adjust output template.
24718 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24720 * config/aarch64/aarch64-protos.h
24721 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24722 * config/aarch64/constraints.md (Dp): New address constraint.
24723 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24726 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24728 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24729 update use of estimate_ipcp_clone_size_and_time.
24730 (estimate_local_effects): Update use of
24731 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24732 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24733 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24734 Return nonspecialized time.
24736 2017-05-04 Richard Biener <rguenther@suse.de>
24738 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24739 for the last VUSE which def dominates the PHI. Directly call
24741 (get_continuation_for_phi_1): Remove.
24743 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24745 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24746 to explain the use of truncating division. Cap the number of
24747 iterations to the maximum given by nb_iterations_upper_bound,
24750 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24752 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24753 * configure: Regenerate.
24754 * config.in: Regenerate.
24755 * config/i386/driver-mingw32.c: new file.
24756 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24757 * config.host: Link driver-mingw32.o on MinGW host.
24758 * doc/install.texi: Document new --enable-mingw-wildcard configure
24761 2017-05-04 Marek Polacek <polacek@redhat.com>
24763 PR tree-optimization/80612
24764 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24766 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24767 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24769 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24770 (movt splitter): Likewise.
24771 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24772 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24773 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24774 block for Thumb-1 with MOVT.
24775 (thumb2_legitimate_address_p): Move code block ...
24776 (can_avoid_literal_pool_for_label_p): ... into this new function.
24777 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24779 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24780 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24781 "M-profile targets with the MOVT instruction".
24783 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24785 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24786 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24787 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24789 2017-05-04 Martin Liska <mliska@suse.cz>
24791 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24792 variable cond_code.
24794 2017-05-04 Richard Biener <rguenther@suse.de>
24796 * tree.c (array_at_struct_end_p): Handle arrays at struct
24797 end with flexarrays more conservatively. Refactor and treat
24798 arrays of arrays or aggregates more strict. Fix
24799 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24800 * tree.c (array_at_struct_end_p): Adjust prototype.
24801 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24802 * gimple-fold.c (get_range_strlen): Likewise.
24803 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24805 2017-05-04 Richard Biener <rguenther@suse.de>
24807 PR tree-optimization/31130
24808 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24810 (supports_overflow_infinity): Likewise.
24811 (is_negative_overflow_infinity): Likewise.
24812 (is_positive_overflow_infinity): Likewise.
24813 (is_overflow_infinity): Likewise.
24814 (stmt_overflow_infinity): Likewise.
24815 (overflow_infinity_range_p): Likewise.
24816 (usable_range_p): Remove as always returning true.
24817 (make_overflow_infinity): Remove.
24818 (negative_overflow_infinity): Likewise.
24819 (positive_overflow_infinity): Likewise.
24820 (avoid_overflow_infinity): Likewise.
24821 (set_value_range): Adjust accordingly.
24822 (set_value_range_to_nonnegative): Likewise, remove now unused
24823 overflow_infinity arg.
24824 (vrp_operand_equal_p): Adjust.
24825 (update_value_range): Likewise.
24826 (range_int_cst_singleton_p): Likewise.
24827 (operand_less_p): Likewise.
24828 (compare_values_warnv): Likewise.
24829 (extract_range_for_var_from_comparison_expr): Likewise.
24830 (vrp_int_const_binop): Likewise.
24831 (zero_nonzero_bits_from_vr): Likewise.
24832 (extract_range_from_multiplicative_op_1): Likewise.
24833 (extract_range_from_binary_expr_1): Likewise.
24834 (extract_range_from_unary_expr): Likewise.
24835 (extract_range_from_comparison): Likewise.
24836 (extract_range_basic): Likewise.
24837 (adjust_range_with_scev): Likewise.
24838 (compare_ranges): Likewise.
24839 (compare_range_with_value): Likewise.
24840 (dump_value_range): Likewise.
24841 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24843 (simplify_cond_using_ranges): Adjust.
24845 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24847 * brig-builtins.def: Added a builtin for class_f64.
24848 * builtin-types.def: Added a builtin type needed by class_f64.
24850 2017-05-03 Jason Merrill <jason@redhat.com>
24852 * timevar.def: Add TV_CONSTEXPR.
24854 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24856 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24858 2017-05-03 Martin Jambor <mjambor@suse.cz>
24860 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24861 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24862 * ipa-cp.c (ipcp_propagate_stage): Do not call
24863 ipa_update_after_lto_read.
24864 * ipa-inline.c (ipa_inline): Likewise.
24866 2017-05-03 Martin Jambor <mjambor@suse.cz>
24868 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24869 tag. Added a default constructor and a destructor.
24870 (ipa_edge_args_sum_t): New class;
24871 (ipa_edge_args_sum): Declare.
24872 (ipa_edge_args_vector): Remove declaration.
24873 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24874 (ipa_free_edge_args_substructures): Remove declaration.
24875 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24876 (ipa_edge_args_info_available_for_edge_p): Likewise.
24877 * ipa-prop.c (ipa_edge_args_vector): Removed.
24878 (edge_removal_hook_holder): Likewise.
24879 (edge_duplication_hook_holder): Likewise.
24880 (ipa_edge_args_sum): New variable.
24881 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24882 ipa_edge_args_vector.
24883 (ipa_free_edge_args_substructures): Likewise.
24884 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24885 ipa_edge_args_vector.
24886 (ipa_edge_removal_hook): Turned into method
24887 ipa_edge_args_sum_t::remove.
24888 (ipa_edge_duplication_hook): Turned into method
24889 ipa_edge_args_sum_t::duplicate.
24890 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24891 registering edge hooks.
24892 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24893 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24894 ipa_edge_args_sum instead of ipa_edge_args_vector.
24895 * ipa-profile.c (ipa_profile): Likewise.
24897 2017-05-03 Martin Jambor <mjambor@suse.cz>
24899 * symbol-summary.h (function_summary): New method exists.
24900 (function_summary::symtab_removal): Deallocate through release.
24901 (call_summary): New class.
24902 (gt_ggc_mx): New overload.
24903 (gt_pch_nx): Likewise.
24904 (gt_pch_nx): Likewise.
24906 2017-05-03 Jeff Law <law@redhat.com>
24908 PR tree-optimization/78496
24909 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24910 from simplify_cond_using_ranges. Split off code to walk
24911 backwards through casts into ...
24912 (simplify_cond_using_ranges_2): New function.
24913 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24914 (execute_vrp): After identifying jump threads, call
24915 simplify_cond_using_ranges_2.
24917 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24920 * ipa-inline.h (inline_summary): Add ctor.
24921 (create_ggc): Do not use ggc_cleared_alloc.
24923 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24924 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24926 * gcc.c (handle_braces): Support escaping in switch matching
24928 * doc/invoke.texi (Spec Files): Document it.
24929 Remove superfluous @code markup in items.
24931 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24933 * diagnostic-show-locus.c (struct column_range): New struct.
24934 (get_affected_columns): New function.
24935 (get_printed_columns): New function.
24936 (struct correction): New struct.
24937 (correction::ensure_capacity): New function.
24938 (correction::ensure_terminated): New function.
24939 (struct line_corrections): New struct.
24940 (line_corrections::~line_corrections): New dtor.
24941 (line_corrections::add_hint): New function.
24942 (layout::print_trailing_fixits): Reimplement in terms of the new
24944 (selftest::test_overlapped_fixit_printing): New function.
24945 (selftest::diagnostic_show_locus_c_tests): Call it.
24947 2017-05-03 Nathan Sidwell <nathan@acm.org>
24949 Canonicalize canonical type hashing
24950 * tree.h (type_hash_canon_hash): Declare.
24951 * tree.c (type_hash_list, attribute_hash_list): Move into
24952 type_hash_canon_hash.
24953 (build_type_attribute_qual_variant): Break out hash code calc into
24954 type_hash_canon_hash.
24955 (type_hash_canon_hash): New. Generic type hash computation.
24956 (build_range_type_1, build_array_type_1, build_function_type,
24957 build_method_type_directly, build_offset_type, build_complex_type,
24958 make_vector_type): Call it.
24960 2017-05-03 Richard Biener <rguenther@suse.de>
24962 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24963 When all DRs have unknown misaligned do not always peel
24964 when there is a store but apply the same costing model as if
24965 there were only loads.
24967 2017-05-03 Richard Biener <rguenther@suse.de>
24970 PR tree-optimization/80492
24971 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24972 compare_base_decls returning dont-know properly.
24974 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24976 * config/arm/iterators.md (CCSI): New mode iterator.
24977 (arch): New mode attribute.
24978 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24979 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24980 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24981 code iterator for success result mode.
24982 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24983 the corresponding new insn generators.
24985 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24988 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24989 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24991 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24993 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24994 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24995 (DDR_A): Wrap DDR argument in brackets.
24996 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24997 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24998 (DDR_REVERSED_P): Likewise.
25000 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25002 PR tree-optimization/79472
25003 * tree-switch-conversion.c (struct switch_conv_info): Add
25004 contiguous_range and default_case_nonstandard fields.
25005 (collect_switch_conv_info): Compute contiguous_range and
25006 default_case_nonstandard fields, don't clear final_bb if
25007 contiguous_range and only the default case doesn't have the required
25009 (check_all_empty_except_final): Set default_case_nonstandard instead
25010 of failing if contiguous_range and the default case doesn't have empty
25012 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25013 and only the default case doesn't have the required constants. Skip
25015 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25016 if default_case_nonstandard.
25017 (build_constructors): Build constant 1 just once. Assert that default
25018 values aren't inserted in between cases if contiguous_range. Skip
25020 (build_arrays): Skip virtual phis.
25021 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25022 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25023 Handle virtual phis.
25024 (gen_inbound_check): Handle default_case_nonstandard case.
25025 (process_switch): Adjust check_final_bb caller. Call
25026 gather_default_values with the first non-default case instead of
25027 default case if default_case_nonstandard.
25029 2017-05-02 Nathan Sidwell <nathan@acm.org>
25031 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25032 check. Fix formatting.
25034 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25036 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25037 errors when comparing specialized and unspecialized times.
25039 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25041 * diagnostic-show-locus.c
25042 (layout::should_print_annotation_line_p): Make private.
25043 (layout::print_annotation_line): Make private.
25044 (layout::annotation_line_showed_range_p): Make private.
25045 (layout::show_ruler): Make private.
25046 (layout::print_source_line): Make private. Pass in line and
25047 line_width, rather than calling location_get_source_line. Drop
25049 (layout::print_leading_fixits): New method.
25050 (layout::print_any_fixits): Rename to...
25051 (layout::print_trailing_fixits): ...this, and make private.
25052 Don't print newline fixits.
25053 (diagnostic_show_locus): Move logic for printing one row into...
25054 (layout::print_line): ...this new function. Move the
25055 location_get_source_line call and error-handling from
25056 print_source_line to here. Call print_leading_fixits, and rename
25057 print_any_fixits to print_trailing_fixits.
25058 (selftest::test_fixit_insert_containing_newline): Update now that
25059 newlines are partially supported.
25060 (selftest::test_fixit_insert_containing_newline_2): New test.
25061 (selftest::test_fixit_replace_containing_newline): Update comments.
25062 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25063 * edit-context.c (class added_line): New class.
25064 (class edited_line): Describe newline handling in comment.
25065 (edited_line::actually_edited_p): New method.
25066 (edited_line::print_content): Delete redundant decl.
25067 (edited_line::m_predecessors): New field.
25068 (edited_file::print_content): Call edited_line::print_content.
25069 (edited_file::print_diff): Update to support newlines.
25070 (edited_file::print_diff_hunk): Likewise.
25071 (edited_file::print_run_of_changed_lines): New function.
25072 (edited_file::print_diff_line): Convert to...
25073 (print_diff_line): ...this.
25074 (edited_file::get_effective_line_count): New function.
25075 (edited_line::edited_line): Initialize new field m_predecessors.
25076 (edited_line::~edited_line): Clean up m_predecessors.
25077 (edited_line::apply_fixit): Handle newlines.
25078 (edited_line::get_effective_line_count): New function.
25079 (edited_line::print_content): New function.
25080 (edited_line::print_diff_lines): New function.
25081 (selftest::test_applying_fixits_insert_containing_newline): New
25083 (selftest::test_applying_fixits_replace_containing_newline): New
25085 (selftest::insert_line): New function.
25086 (selftest::test_applying_fixits_multiple_lines): Add example of
25088 (selftest::edit_context_c_tests): Call the new tests.
25090 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25092 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25093 parameter cand. Update dump information.
25094 (get_computation_cost): Update uses.
25096 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25098 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25099 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25100 (get_computation_at, rewrite_use_address): Update use of
25101 get_computation_aff.
25103 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25105 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25106 (get_computation): Delete.
25107 (get_computation_cost): Implement like get_computation_cost_at.
25108 Use get_computation_at.
25109 (get_computation_cost_at): Delete.
25110 (rewrite_use_nonlinear_expr): Use get_computation_at.
25111 (rewrite_use_compare, remove_unused_ivs): Ditto.
25113 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25115 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25117 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25119 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25120 (ivopts_global_cost_for_size): Rename parameter and update uses.
25121 (iv_ca_recount_cost): Update uses.
25122 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25123 candidates seperately in n_invs and n_cands.
25124 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25126 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25128 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25129 (find_inv_vars_cb): New.
25130 (find_depends): Renamed to ...
25131 (find_inv_vars): ... this.
25132 (add_candidate_1, force_var_cost): Call find_inv_vars.
25133 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25135 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25137 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25138 inv_vars. Add inv_exprs.
25139 (struct iv_cand): Rename depends_on to inv_vars.
25140 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25141 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25142 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25143 (dump_cand): Dump inv_vars.
25144 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25145 (record_invariant, find_depends, add_candidate_1): Ditto.
25146 (set_group_iv_cost, force_var_cost): Ditto.
25147 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25148 (get_computation_cost_at, get_computation_cost): Ditto.
25149 (determine_group_iv_cost_generic): Ditto.
25150 (determine_group_iv_cost_address): Ditto.
25151 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25152 (determine_group_iv_costs): Ditto.
25153 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25154 (iv_ca_set_remove_invariants): Renamed to ...
25155 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25156 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25157 (iv_ca_set_add_invariants): Renamed to ...
25158 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25159 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25160 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25161 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25162 (create_new_ivs): Remove useless dump.
25164 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25166 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25168 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25169 (iv_ca_set_no_cp, create_new_iv): Ditto.
25171 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25173 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25175 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25177 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25178 function tree_check2.
25180 2017-05-02 Martin Liska <mliska@suse.cz>
25182 * doc/gcov.texi: Add missing preposition.
25183 * gcov.c (function_info::function_info): Properly fill up
25184 all member variables.
25186 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25188 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25190 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25192 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25194 2017-05-02 Martin Liska <mliska@suse.cz>
25197 * lto-streamer-in.c (lto_read_tree_1): Remove
25198 LTO_STREAMER_DEBUG.
25199 * lto-streamer.c (struct tree_hash_entry): Likewise.
25200 (struct tree_entry_hasher): Likewise.
25201 (tree_entry_hasher::hash): Likewise.
25202 (tree_entry_hasher::equal): Likewise.
25203 (lto_streamer_init): Likewise.
25204 (lto_orig_address_map): Likewise.
25205 (lto_orig_address_get): Likewise.
25206 (lto_orig_address_remove): Likewise.
25207 * lto-streamer.h: Likewise.
25208 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25209 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25211 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25213 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25214 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25215 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25216 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25217 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25218 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25219 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25220 (mm_maskz_sub_ss): New intrinsics.
25221 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25222 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25223 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25224 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25225 (__builtin_ia32_subss_mask_round): New builtins.
25226 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25227 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25228 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25230 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25231 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25233 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25236 2017-05-02 Martin Jambor <mjambor@suse.cz>
25238 PR tree-optimization/78687
25239 * tree-sra.c (access): New field parent.
25240 (process_subtree_disqualification): New function.
25241 (disqualify_candidate): Call it.
25242 (build_accesses_from_assign): Reset write flag if creating an
25244 (build_access_subtree): Fill in parent field and also prpagate
25245 down grp_write flag.
25246 (create_artificial_child_access): New parameter set_grp_write, set
25247 grp_write to its value.
25248 (propagate_subaccesses_across_link): Also propagate grp_write flag
25250 (propagate_all_subaccesses): Push the closest parent back to work
25251 queue if add_access_to_work_queue returned true.
25253 2017-05-02 Richard Biener <rguenther@suse.de>
25255 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25256 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25257 -fstrict-overflow documentation.
25258 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25259 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25260 flag_strict_overflow.
25261 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25262 * lto-opts.c (lto_write_options): Do not stream it.
25263 * lto-wrapper.c (merge_and_complain): Do not handle it.
25264 * opts.c (default_options_table): Do not set -fstrict-overflow.
25265 (finish_options): Likewise do not clear it when sanitizing.
25266 * simplify-rtx.c (simplify_const_relational_operation): Do not
25267 test flag_strict_overflow.
25269 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25271 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25272 using enabled attribute.
25273 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25274 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25275 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25276 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25277 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25278 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25279 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25280 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25281 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25282 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25284 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25286 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25288 2017-05-02 Richard Biener <rguenther@suse.de>
25290 PR tree-optimization/80591
25292 2017-04-10 Richard Biener <rguenther@suse.de>
25294 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25297 2017-05-02 Richard Biener <rguenther@suse.de>
25299 PR tree-optimization/80549
25300 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25301 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25302 headers if they do not have a preheader.
25304 2017-05-02 Martin Liska <mliska@suse.cz>
25307 * common.opt: Fix typo.
25308 * doc/invoke.texi: Likewise.
25310 2017-05-01 Jan Beulich <jbeulich@suse.com>
25312 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25313 swapping, add (x,x,m,x,n) alternative.
25315 2017-05-01 Nathan Sidwell <nathan@acm.org>
25317 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25318 unnecessary unadjusted_alignment check.
25320 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25323 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25325 * gimplify.c (gimplify_cilk_detach): New function.
25326 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25327 * tree-core.h: Document EXPR_CILK_SPAWN.
25328 * tree.h (EXPR_CILK_SPAWN): Define.
25330 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25332 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25333 to use new fixit_hint representation, using the "replace" logic.
25334 (get_line_span_for_fixit_hint): Likewise.
25335 (layout::print_any_fixits): Likewise.
25336 (selftest::test_one_liner_many_fixits): Rename to...
25337 (selftest::test_one_liner_many_fixits_1): ...this, and update
25338 comment and expected output to reflect that the multiple fix-it
25339 hints are now consolidated into one insertion.
25340 (selftest::test_one_liner_many_fixits_2): New test.
25341 (selftest::test_diagnostic_show_locus_one_liner): Update for
25343 (selftest::test_fixit_consolidation): Update for fix-it API
25345 * diagnostic.c (print_parseable_fixits): Likewise.
25346 * edit-context.c (edited_line::m_line_events): Convert from
25347 auto_vec <line_event *> to auto_vec <line_event>.
25348 (class line_event): Convert from abstract base class to a concrete
25349 class, taking over the role of replace_event.
25350 (class insert_event): Delete.
25351 (class replace_event): Rename to class line_event. Convert to
25353 (edit_context::add_fixits): Reimplement.
25354 (edit_context::apply_insert): Delete.
25355 (edit_context::apply_replace): Rename to...
25356 (edit_context::apply_fixit): ...this. Convert to half-open range.
25357 (edited_file::apply_insert): Delete.
25358 (edited_file::apply_replace): Rename to...
25359 (edited_file::apply_fixit): ...this.
25360 (edited_line::~edited_line): Drop deletion of events.
25361 (edited_line::apply_insert): Delete.
25362 (edited_line::apply_replace): Rename to...
25363 (edited_line::apply_fixit): ...this. Convert to half-open range.
25364 Update for change to type of m_line_events.
25365 * edit-context.h (edit_context::apply_insert): Delete.
25366 (edit_context::apply_replace): Rename to...
25367 (edit_context::apply_fixit): ...this.
25369 2017-05-01 Martin Sebor <msebor@redhat.com>
25371 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25374 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25377 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25378 __get_cpuid_max returns 0.
25379 (__get_cpuid_count): Ditto.
25381 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25383 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25384 replacement expression is another instance of one of its arguments.
25386 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25389 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25390 check for stack push/pop autoinc.
25391 * config/i386/i386.c (ix86_agi_dependent): Return false
25392 if the only reason why modified_in_p returned true is that
25393 addr is SP based and set_insn is a push or pop.
25395 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25397 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25400 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25403 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25404 (account_size_time): Use two predicates - exec_pred and
25406 (evaluate_conditions_for_known_args): Compute both clause and
25408 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25409 (inline_summary_t::duplicate): Update.
25410 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25412 (compute_inline_parameters): Likewise.
25413 (estimate_edge_size_and_time): Update caluclation of time.
25414 (estimate_node_size_and_time): Compute both time and nonspecialized
25416 (estimate_ipcp_clone_size_and_time): Update.
25417 (inline_merge_summary): Update.
25418 (do_estimate_edge_time): Update.
25419 (do_estimate_edge_size): Update.
25420 (do_estimate_edge_hints): Update.
25421 (inline_read_section, inline_write_summary): Stream both new predicates.
25422 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25424 (compute_inlined_call_time): Cleanup.
25425 (big_speedup_p): Update.
25426 (edge_badness): Update.
25427 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25428 (size_time_entry): Replace predicate by exec_predicate and
25429 nonconst_predicate.
25430 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25431 (estimate_edge_time): Return also nonspec_time.
25432 (reset_edge_growth_cache): Update.
25434 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25436 PR rtl-optimization/80491
25437 * ifcvt.c (noce_process_if_block): When looking for x setter
25438 with missing else_bb, don't check only the insn right before
25439 cond_earliest, but look for the last insn that x is modified in
25440 within the same bb.
25442 PR rtl-optimization/80491
25443 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25445 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25447 PR tree-optimization/80487
25448 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25450 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25452 PR tree-optimization/79697
25453 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25454 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25455 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25457 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25458 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25460 2017-04-28 Martin Sebor <msebor@redhat.com>
25462 PR tree-optimization/80523
25463 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25464 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25466 (maybe_warn, format_directive, parse_directive): Use new functions.
25467 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25469 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25471 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25473 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25475 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25476 target_header_dir): Set correctly.
25477 * configure: Regenerated.
25478 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25479 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25480 instead of SYSTEM_HEADER_DIR.
25482 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25484 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25485 (estimate_local_effects): Likewise.
25486 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25487 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25488 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25489 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25490 do_estimate_edge_time, estimate_edge_time): Likewise.
25491 * ipa-inline-analysis.c (estimate_node_size_and_time,
25492 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25493 (estimate_time_after_inlining): Remove.
25495 2017-04-28 Martin Liska <mliska@suse.cz>
25497 * doc/gcov.texi: Enhance documentation of gcov.
25499 2017-04-28 Martin Liska <mliska@suse.cz>
25501 * doc/gcov.texi: Sort options in alphabetic order.
25502 * doc/gcov-dump.texi: Likewise.
25503 * doc/gcov-tool.texi: Likewise.
25504 * gcov.c (print_usage): Likewise.
25505 * gcov-dump.c (print_usage): Likewise.
25506 * gcov-tool.c (print_merge_usage_message): Likewise.
25507 (print_rewrite_usage_message): Likewise.
25508 (print_overlap_usage_message): Likewise.
25510 2017-04-28 Martin Liska <mliska@suse.cz>
25512 PR gcov-profile/53915
25513 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25515 2017-04-28 Martin Liska <mliska@suse.cz>
25517 PR gcov-profile/79891
25518 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25519 is marked by compiler as living on a line.
25520 (get_cycles_count): Remove usage of the union.
25521 (output_intermediate_file): Likewise.
25522 (find_source): Fix GNU coding style.
25523 (accumulate_line_counts): Remove old non-all block mode.
25524 (output_lines): Remove usage of the union.
25525 * profile.c (output_location): Include all BBs, even if
25526 belonging to a same line (and file) as a previous BB.
25528 2017-04-28 Martin Liska <mliska@suse.cz>
25530 * gcov.c (process_args): Handle new argument 'w'.
25531 (read_graph_file): Assign ID to BBs.
25532 (output_branch_count): Display BB # if verbose flag is set.
25533 (output_lines): Likewise for arcs.
25534 (print_usage): Add '--verbose' option help.
25535 * doc/gcov.texi: Document --verbose (-w) option.
25537 2017-04-28 Martin Liska <mliska@suse.cz>
25539 * gcov.c (struct block_location_info): New struct.
25540 (process_file): Fill up the new structure.
25541 (read_graph_file): Replace usage of encoding by the newly added
25543 (add_line_counts): Likewise.
25544 (accumulate_line_counts): Remove usage of the union.
25545 (function_info::function_info): New function.
25546 (function_info::~function_info): Likewise.
25547 (process_file): Call delete instead of release_function.
25548 (release_function): Release the function.
25549 (release_structures): Call delete instead of release_function.
25550 (solve_flow_graph): Replace usage of num_blocks.
25551 (find_exception_blocks): Likewise.
25552 (output_lines): Fix GNU coding style.
25554 2017-04-28 Martin Liska <mliska@suse.cz>
25557 * coverage.c (coverage_remove_note_file): New function.
25558 * coverage.h: Declare the function.
25559 * toplev.c (finalize): Clean if an error has been seen.
25561 2017-04-28 Martin Liska <mliska@suse.cz>
25563 PR gcov-profile/80031
25564 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25565 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25566 * gcov.c (read_graph_file): Read just number of blocks.
25567 * profile.c (branch_prob): Do not stream 0 flags per a basic
25570 2017-04-28 Martin Liska <mliska@suse.cz>
25572 * gcov-dump.c (tag_*): Add new argument to declarations.
25573 (dump_gcov_file): Likewise.
25574 (tag_blocks): Add and use new argument depth.
25575 (tag_arcs): Likewise.
25576 (tag_lines): Likewise.
25577 (tag_counters): Likewise.
25578 (tag_summary): Likewise.
25579 (dump_working_sets): Use depth to do a proper indentation.
25581 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25584 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25585 * symtab.c (symtab_node::debug_symtab): Move definition here.
25587 2017-04-28 Richard Biener <rguenther@suse.de>
25589 * lto-streamer.h (LTO_major_version): Bump to 7.
25591 2017-04-28 Richard Biener <rguenther@suse.de>
25593 * tree-vrp.c (assert_info): New struct.
25594 (add_assert_info): New helper.
25595 (register_edge_assert_for_2): Refactor to add asserts to a vector
25597 (register_edge_assert_for_1): Likewise.
25598 (register_edge_assert_for): Likewise.
25599 (finish_register_edge_assert_for): New helper actually registering
25600 asserts where live on edge.
25601 (find_conditional_asserts): Adjust.
25602 (find_switch_asserts): Likewise.
25603 (evrp_dom_walker::try_find_new_range): Generalize.
25604 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25606 2017-04-27 Marek Polacek <polacek@redhat.com>
25609 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25610 arg10 and arg11 to itype.
25612 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25614 * doc/extend.texi (Object Size Checking): Improve grammar.
25616 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25619 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25620 that the logic for permitting reciprocal estimates matches that
25623 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25626 * tree.c (type_cache_hasher::equal): Only compare
25627 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25628 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25629 non-aggregate element types.
25630 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25631 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25633 2017-04-27 Richard Biener <rguenther@suse.de>
25635 PR middle-end/80533
25636 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25637 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25638 keeping a reference to a trailing array.
25640 2017-04-27 Richard Biener <rguenther@suse.de>
25642 PR middle-end/80539
25643 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25644 being in loop-closed SSA form conservatively.
25645 (chrec_fold_multiply_poly_poly): Likewise.
25647 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25649 PR middle-end/79665
25650 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25651 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25653 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25656 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25657 (aarch64_function_arg_alignment): Return unsigned int again, but still
25658 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25659 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25660 Don't emit -Wpsabi note.
25661 (aarch64_function_arg_boundary): Likewise.
25662 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25665 2017-04-26 Nathan Sidwell <nathan@acm.org>
25667 * tree.h (crc32_unsigned_n): Declare.
25668 (crc32_unsigned, crc32_unsigned): Make inline.
25669 * tree.c (crc32_unsigned_bits): Replace with ...
25670 (crc32_unsigned_n): ... this.
25671 (crc32_unsigned, crc32_byte): Remove.
25672 (crc32_string): Remove unnecessary braces.
25674 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25676 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25677 * ipa-inline-analysis.c (MAX_TIME): Remove.
25678 (account_size_time): Use sreal for time.
25679 (dump_inline_summary): Update.
25680 (estimate_function_body_sizes): Update.
25681 (estimate_edge_size_and_time): Update.
25682 (estimate_calls_size_and_time): Update.
25683 (estimate_node_size_and_time): Update.
25684 (inline_merge_summary): Update.
25685 (inline_update_overall_summary): Update.
25686 (estimate_time_after_inlining): Update.
25687 (inline_read_section): Update.
25688 (inline_write_summary): Update.
25689 * ipa-inline.c (compute_uninlined_call_time): Update.
25690 (compute_inlined_call_time): Update.
25691 (recursive_inlining): Update.
25692 (inline_small_functions): Update.
25693 (dump_overall_stats): Update.
25694 * ipa-inline.h: Include sreal.h.
25695 (size_time_entry): Turn time to sreal.
25696 (inline_summary): Turn self_time nad time to sreal.
25698 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25700 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25702 (sreal::stream_out, sreal::stream_in): New.
25703 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25705 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25707 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25710 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25713 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25715 (dimode_scalar_chain::compute_convert_gain): Ditto.
25716 (dimode_scalar_chain::make_vector_copies): Ditto.
25717 (dimode_scalar_chain::convert_reg): Ditto.
25718 (dimode_scalar_chain::convert_insn): Ditto.
25719 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25720 (VI248_AVX512BW_1): New mode iterator.
25721 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25722 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25725 2017-04-25 Martin Sebor <msebor@redhat.com>
25727 PR tree-optimization/80497
25728 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25729 constants are representable in HOST_WIDE_INT.
25730 (parse_directive): Ditto.
25732 2017-04-25 Martin Sebor <msebor@redhat.com>
25735 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25736 (new_zero_array): Adjust signature.
25737 (dom_info::dom_init): Used unsigned rather that size_t.
25738 (dom_info::dom_info): Same.
25740 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25741 Jakub Jelinek <jakub@redhat.com>
25744 * config/arm/arm.c: Include gimple.h.
25745 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25746 returns negative, increment ncrn only if it returned positive.
25747 (arm_needs_doubleword_align): Return int instead of bool,
25748 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25749 members, but if there is any such non-FIELD_DECL
25750 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25751 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25752 returns negative, increment nregs only if it returned positive.
25753 (arm_setup_incoming_varargs): Likewise.
25754 (arm_function_arg_boundary): Emit -Wpsabi note if
25755 arm_needs_doubleword_align returns negative, return
25756 DOUBLEWORD_ALIGNMENT only if it returned positive.
25758 2017-04-25 Marek Polacek <polacek@redhat.com>
25761 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25762 first argument to type.
25764 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25767 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25768 type checks to test for compatibility instead of equality.
25770 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25771 Jakub Jelinek <jakub@redhat.com>
25774 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25776 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25777 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25778 the alignment computation, but return their maximum in warn_alignment.
25779 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25780 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25782 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25783 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25786 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25788 * config/arc/simdext.md (dmpyh): Fix typo.
25790 2017-04-25 Richard Biener <rguenther@suse.de>
25792 PR tree-optimization/80492
25793 * alias.c (compare_base_decls): Handle registers with asm
25794 specification conservatively.
25795 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25796 compare_base_decls returning dont-know properly.
25798 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25800 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25801 (legitimate_offset_address_p): New function.
25802 (arc_legitimate_address_p): Use above function.
25804 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25806 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25808 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25810 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25811 ACCH registers whenever they are available.
25813 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25815 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25816 double regs fix when not used.
25818 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25820 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25822 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25823 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25825 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25827 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25828 when emitting short ADD instructions.
25830 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25832 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25834 (cmpsi_cc_c_insn): Likewise.
25835 (cbranchsi4_scratch): Compute proper instruction length using
25836 compact_hreg_operand.
25837 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25839 2017-04-25 Richard Biener <rguenther@suse.de>
25841 PR middle-end/80509
25842 * passes.c (pass_manager::pass_manager): Initialize
25843 m_name_to_pass_map.
25845 2017-04-25 Richard Biener <rguenther@suse.de>
25847 PR tree-optimization/79201
25848 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25850 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25853 * config/s390/vector.md: Split MEM->GPR vector moves for
25854 non-s_operand addresses.
25856 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25859 * config/s390/predicates.md (reload_const_wide_int_operand): New
25861 * config/s390/s390.md ("movti"): Remove d/P alternative.
25862 ("movti_bigconst"): New pattern definition.
25864 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25867 * s390-protos.h (s390_expand_cs_hqi): Removed.
25868 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25869 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25870 modes as well as CCZ1mode and CCZmode.
25871 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25872 signature of s390_emit_compare_and_swap.
25873 (s390_expand_cs_hqi): Likewise, make static.
25874 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25875 compare-and-swap, in some cases.
25876 (s390_expand_cs): Wrapper function.
25877 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25879 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25880 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25881 patterns for small and large integers. Forbid symref memory operands.
25882 Move expander to s390.c. Require cc register.
25883 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25884 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25885 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25886 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25887 symref memory operands. Remove CC mode and call s390_match_ccmode
25889 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25891 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25893 * config/s390/s390.md (define_peephole2): New peephole to help
25894 combining the load-and-test pattern with volatile memory.
25896 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25898 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25899 with CCZmode for TARGET_Z196.
25901 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25903 PR rtl-optimization/80501
25904 * combine.c (make_compound_operation_int): Set subreg_code to SET
25905 even for AND with mask of the sign bit of mode.
25907 PR rtl-optimization/80500
25908 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25909 sum's initial value.
25911 2017-04-25 Julian Brown <julian@codesourcery.com>
25912 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25914 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25916 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25918 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25920 2017-04-25 Julian Brown <julian@codesourcery.com>
25921 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25923 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25924 (thunderx2t99_sha): New Reservation.
25926 2017-04-25 Julian Brown <julian@codesourcery.com>
25927 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25929 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25930 type for 1-element load.
25932 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25934 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25936 2017-04-24 Martin Jambor <mjambor@suse.cz>
25938 PR tree-optimization/80293
25939 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25940 char arrays not totally scalarizable if it is false.
25941 (analyze_all_variable_accesses): Pass correct value in the new
25942 parameter. Add a statistics counter.
25944 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25946 PR middle-end/79931
25947 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25949 2017-04-24 Richard Biener <rguenther@suse.de>
25951 PR tree-optimization/80494
25952 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25953 out for complex types.
25955 2017-04-24 Richard Biener <rguenther@suse.de>
25957 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25958 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25959 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25960 (DFS): Adjust and never fail.
25961 (sccvn_dom_walker::fail): Remove.
25962 (sccvn_dom_walker::before_dom_children): Adjust.
25963 (run_scc_vn): Likewise and never fail.
25964 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25965 (pass_fre::execute): Likewise.
25967 2017-04-24 Richard Biener <rguenther@suse.de>
25969 PR tree-optimization/79725
25970 * tree-ssa-sink.c (statement_sink_location): Return whether
25971 failure reason was zero uses. Move that check later.
25972 (sink_code_in_bb): Deal with zero uses by removing the stmt
25975 2017-04-24 Richard Biener <rguenther@suse.de>
25978 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25979 pointer-based references.
25981 2017-04-24 Richard Biener <rguenther@suse.de>
25984 * pass_manager.h (pass_manager::operator new): Remove.
25985 (pass_manager::operator delete): Likewise.
25986 * passes.c (pass_manager::operator new): Remove.
25987 (pass_manager::operator delete): Likewise.
25988 (pass_manager::pass_manager): Zero individual pass members.
25990 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25993 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25994 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25995 Check "XEXP (src, 1)" operand here.
25996 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25997 Check "XEXP (src, 1)" operand here.
25998 (dimode_scalar_chain::make_vector_copies): Detect count register
25999 of a shift instruction. Zero extend count register from QImode
26000 to DImode to satisfy vector shift pattern count operand predicate.
26001 Substitute vector shift count operand with a DImode copy.
26002 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26005 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26007 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26008 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26009 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26010 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26011 (UNSPEC_NOREX_MEM): Remove definition.
26013 2017-04-21 Richard Biener <rguenther@suse.de>
26015 PR tree-optimization/79547
26016 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26017 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26018 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26019 without any constraints.
26021 2017-04-21 Richard Biener <rguenther@suse.de>
26023 PR tree-optimization/78847
26024 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26026 2017-04-21 Richard Biener <rguenther@suse.de>
26028 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26029 (build_distinct_type_copy): Likewise.
26030 (build_variant_type_copy): Likewise.
26031 * tree.c (build_qualified_type): Pass down mem-stat info.
26032 (build_distinct_type_copy): Likewise.
26033 (build_variant_type_copy): Likewise.
26035 2017-04-21 Richard Biener <rguenther@suse.de>
26037 PR tree-optimization/80237
26038 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26040 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26041 for a simplified result.
26043 2016-04-21 Richard Biener <rguenther@suse.de>
26045 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26046 sth as strict as a simple_iv but a chrec without symbols and an
26047 operand defined in the loop we are peeling (and not some subloop).
26048 (propagate_constants_for_unrolling): Propagate all constants.
26050 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26053 * config/i386/i386.c (print_reg): Remove assert for disalowed
26054 regno values, call output_operand_lossage instead.
26056 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26059 * config/i386/constraints.md (Yc): New register constraint.
26060 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26061 Use Yc constraint for alternative 2 of operand 0. Remove
26062 preferred_for_speed attribute.
26064 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26066 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26067 lastprivate clauses in SIMT case.
26069 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26071 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26073 2017-04-20 Richard Biener <rguenther@suse.de>
26075 PR tree-optimization/57796
26076 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26077 as N scalar stores.
26078 (vect_model_load_cost): Cost gathers as N scalar loads.
26080 2017-04-20 Richard Biener <rguenther@suse.de>
26082 * ggc-page.c (ggc_allocated_p): Rename to ...
26083 (safe_lookup_page_table_entry): ... this and return the lookup
26085 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26087 2017-04-20 Richard Biener <rguenther@suse.de>
26089 PR tree-optimization/80453
26090 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26091 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26092 from the conditions.
26093 (vn_phi_eq): Pass them down.
26094 (vn_phi_lookup): Record them.
26095 (vn_phi_insert): Likewise.
26097 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26099 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26100 uninitialized variable warning to avoid buffer overrun.
26102 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26105 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26106 is suppressed for '{ 0 }' in C.
26108 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26110 * BASE-VER: Set to 8.0.0.
26112 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26114 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26115 priority .init_array and .fini_array section with SECTION_NOTYPE
26118 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26120 PR middle-end/80423
26121 * tree.h (build_array_type): Add typeless_storage default argument.
26122 * tree.c (type_cache_hasher::equal): Also compare
26123 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26124 (build_array_type): Add typeless_storage argument, set
26125 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26127 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26128 (build_array_type): Likewise. Add typeless_storage argument.
26130 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26131 Jakub Jelinek <jakub@redhat.com>
26133 PR tree-optimization/80426
26134 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26135 operation on symbolic operands, also compute the overflow for the
26136 invariant part when the operation degenerates into a negation.
26138 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26141 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26142 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26145 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26147 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26150 * config/avr/avr.c (tree.h): Include it.
26151 (cgraph.h): Include it.
26152 (avr_encode_section_info): Don't warn for uninitialized progmem
26153 variable if it's just an alias.
26155 2017-04-19 Richard Biener <rguenther@suse.de>
26158 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26159 when needed by AutoPGO.
26161 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26164 * doc/lto.texi: Remove an extra 'that'.
26166 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26168 PR rtl-optimization/80429
26169 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26170 are only used in debug insns.
26172 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26173 Vladimir Makarov <vmakarov@redhat.com>
26175 * config/sparc/predicates.md (input_operand): Add comment. Return
26176 true for any memory operand when LRA is in progress.
26177 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26179 2017-04-18 Jeff Law <law@redhat.com>
26182 * mips.md ({return,simple_return}_internal): Do not overwrite
26185 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26187 PR tree-optimization/80443
26188 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26189 instead of adding 1, subtract -1 and similarly instead of subtracting
26192 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26194 PR rtl-optimization/80357
26195 * haifa-sched.c (tmp_bitmap): New variable.
26196 (model_recompute): Handle duplicate use records.
26197 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26198 (free_global_sched_pressure_data): Free it.
26200 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26203 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26204 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26205 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26206 instead of SYSTEM_HEADER_DIR.
26208 2017-04-18 Jeff Law <law@redhat.com>
26210 PR middle-end/80422
26211 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26212 predecessors after walking up the insn chain.
26214 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26217 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26218 sizetype type into debug info.
26220 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26223 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26224 unneeded test for TARGET_UPPER_REGS_SF.
26225 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26227 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26230 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26231 instead of gsi_after_labels.
26233 2017-04-18 Jeff Law <law@redhat.com>
26235 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26239 2017-04-13 Jeff Law <law@redhat.com>
26240 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26241 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26243 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26246 * config/avr/avr.c (intl.h): Include it.
26247 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26249 2017-04-18 Martin Liska <mliska@suse.cz>
26251 PR gcov-profile/78783
26252 * gcov-tool.c (gcov_output_files): Validate that destination
26253 file is either removed by the tool or by a user.
26255 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26256 Guy Benyei <guybe@mellanox.com>
26258 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26259 block, and do not negate it, the stored id is already negative.
26261 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26263 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26265 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26268 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26269 masks of options that should be turned off if the VSX vector
26270 options are turned off.
26271 (OTHER_P8_VECTOR_MASKS): Likewise.
26272 (OTHER_VSX_VECTOR_MASKS): Likewise.
26273 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26274 rs6000_disable_incompatible_switches to validate no type switches
26276 (rs6000_incompatible_switch): New function to disallow turning on
26277 other vector options if -mno-vsx, -mno-power8-vector, or
26278 -mno-power9-vector are specified.
26280 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26282 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26284 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26286 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26287 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26288 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26289 (ARG_POINTER_CFA_OFFSET): Likewise.
26291 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26293 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26294 conditions to take advantage of various optimizations.
26296 2017-04-13 Jeff Law <law@redhat.com>
26298 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26299 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26300 (zero_extendsidi2_dext): Likewise.
26302 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26305 * fold-const.c (fold_ternary_loc): Revert
26306 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26309 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26311 PR rtl-optimization/80343
26312 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26313 new scratch pseudo.
26315 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26318 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26319 to ubsan_encode_value.
26321 2017-04-13 Jeff Law <law@redhat.com>
26323 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26324 appearing in DEBUG_INSNs.
26326 2017-04-13 Martin Liska <mliska@suse.cz>
26328 PR gcov-profile/80413
26329 * gcov-io.c (gcov_write_string): Copy to buffer just when
26330 allocated size is greater than zero.
26332 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26335 * dwarf2out.c (decls_for_scope): Ignore declarations of
26336 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26338 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26341 * ipa-visibility.c (non_local_p): Fix typos.
26342 (localize_node): When localizing symbol in same comdat group,
26343 dissolve the group only when we know external symbols are going
26345 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26347 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26349 PR tree-optimization/79390
26350 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26351 order does not result in usable sequence, retry with reversed operand
26357 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26358 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26359 op0 instead of fold_convert_loc (loc, type, arg0).
26361 2017-04-12 Jeff Law <law@redhat.com>
26363 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26364 has a delay slot in the generated code.
26366 * config/cris/cris.md (cris_preferred_reload_class): Return
26367 GENNONACR_REGS rather than GENERAL_REGS.
26369 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26372 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26373 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26374 signedness of the result type.
26376 2017-04-12 Richard Biener <rguenther@suse.de>
26377 Jeff Law <law@redhat.com>
26379 PR tree-optimization/80359
26380 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26381 trim stores to TARGET_MEM_REFs.
26383 2017-04-12 Richard Biener <rguenther@suse.de>
26385 PR tree-optimization/79390
26386 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26387 threading case even more.
26389 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26392 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26393 for quad_address_p for TImode, instead of just not indexed_address.
26395 2017-04-12 Richard Biener <rguenther@suse.de>
26396 Bernd Edlinger <bernd.edlinger@hotmail.de>
26398 PR middle-end/79671
26399 * alias.c (component_uses_parent_alias_set_from): Handle
26400 TYPE_TYPELESS_STORAGE.
26401 (get_alias_set): Likewise.
26402 * tree-core.h (tree_type_common): Add typeless_storage flag.
26403 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26404 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26405 for types containing members with TYPE_TYPELESS_STORAGE.
26406 (place_field): Likewise.
26407 (layout_type): Likewise for ARRAY_TYPE.
26408 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26409 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26410 TYPE_TYPELESS_STORAGE.
26411 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26413 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26416 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26417 first argument to type.
26419 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26423 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26424 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26425 (rs6000_expand_binop_builtin): Likewise.
26426 (rs6000_expand_ternop_builtin): Likewise; also add missing
26427 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26428 vshasigma built-ins.
26429 * doc/extend.texi: Document that vec_xxpermdi's third argument
26430 must be a constant.
26432 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26434 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26435 Use shift_const cost parameter when calculating gain of STV shifts.
26437 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26439 PR rtl-optimization/70478
26440 * lra-constraints.c (process_alt_operands): Check memory for
26441 disfavoring memory insn operand.
26443 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26445 PR middle-end/80100
26446 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26447 left shift in unsigned HOST_WIDE_INT type.
26449 PR rtl-optimization/80385
26450 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26451 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26454 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26455 if they have any depend clauses.
26457 2017-04-11 Martin Liska <mliska@suse.cz>
26460 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26461 * ipa-split.c (split_function): Create a local comdat symbol
26462 if caller is in a comdat group.
26464 2017-04-11 Martin Liska <mliska@suse.cz>
26467 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26470 2017-04-11 Martin Sebor <msebor@redhat.com>
26472 PR middle-end/80364
26473 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26474 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26476 (directive::set_width, directive::set_precision, format_character):
26478 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26481 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26484 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26485 conflict, set target->arch_name instead of target->cpu_name.
26487 2017-04-11 Richard Biener <rguenther@suse.de>
26489 PR tree-optimization/80374
26490 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26491 build_zero_cst, remove fold_convertible_p check again.
26493 2017-04-11 Martin Liska <mliska@suse.cz>
26496 * ubsan.c (instrument_object_size): Do not instrument register
26499 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26502 * config/i386/i386-builtin-types.def
26503 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26504 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26505 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26506 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26507 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26508 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26509 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26510 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26511 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26512 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26513 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26514 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26515 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26516 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26517 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26518 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26519 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26520 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26521 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26522 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26523 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26524 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26525 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26526 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26527 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26528 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26529 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26530 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26531 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26532 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26533 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26534 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26535 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26536 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26537 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26538 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26539 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26540 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26541 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26542 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26543 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26544 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26545 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26546 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26547 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26548 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26550 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26551 flag to second_arg_count, handle 4 argument function type _COUNT
26552 aliases, handle second_arg_count on second argument rather than last.
26554 2017-04-10 Jeff Law <law@redhat.com>
26556 PR tree-optimization/80374
26557 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26558 record anything if we can not convert integer_zero_node to the
26561 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26564 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26565 Enhance special handling given to the TARGET_P9_MINMAX option in
26566 relation to certain other options.
26568 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26570 PR tree-optimization/80153
26571 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26572 remove POINTER_PLUS_EXPR's base part directly, rather than through
26575 2017-04-10 Richard Biener <rguenther@suse.de>
26576 Bin Cheng <bin.cheng@arm.com>
26578 PR tree-optimization/80153
26579 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26580 the first element of pointer type aff_tree. Build result expr in
26582 (add_elt_to_tree): Convert to type unconditionally. Remove other
26583 fold_convert calls.
26584 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26585 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26587 2017-04-10 Richard Biener <rguenther@suse.de>
26589 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26592 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26594 PR rtl-optimization/70478
26595 * lra-constraints.c (curr_small_class_check): New.
26596 (update_and_check_small_class_inputs): New.
26597 (process_alt_operands): Update curr_small_class_check. Disfavor
26598 alternative insn memory operands. Check available regs for small
26601 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26604 * config/mips/mips.opt (-mvirt): Update description.
26605 * doc/invoke.texi (-mvirt): Likewise.
26607 2017-04-10 Richard Biener <rguenther@suse.de>
26609 PR middle-end/80362
26610 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26611 looking for NEGATE_EXPR in -A / -B to A / B folding.
26613 2017-04-10 Martin Liska <mliska@suse.cz>
26615 PR gcov-profile/80224
26616 * gcov.c (print_usage): Fix usage string.
26617 (get_gcov_intermediate_filename): Remove.
26618 (output_gcov_file): Use both for normal and intermediate format.
26619 (generate_results): Do not initialize special file for
26620 intermediate format.
26622 2017-04-10 Richard Biener <rguenther@suse.de>
26624 PR tree-optimization/80304
26625 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26628 2017-04-10 Nathan Sidwell <nathan@acm.org>
26631 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26632 (rs6000_init_builtins): Use it.
26634 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26636 * config/arm/arm.md (<mrc>): Add mode to SET source.
26637 (<mrrc>): Likewise.
26639 2017-04-10 Richard Biener <rguenther@suse.de>
26641 PR middle-end/80344
26642 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26644 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26647 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26648 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26649 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26650 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26651 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26652 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26653 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26654 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26655 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26656 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26657 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26658 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26659 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26660 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26661 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26662 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26663 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26664 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26665 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26666 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26667 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26668 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26669 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26671 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26673 PR rtl-optimization/70478
26674 * lra-constraints.c: Reverse the last patch.
26676 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26678 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26679 Add comment for WCHAR_T.
26681 2017-04-08 Martin Liska <mliska@suse.cz>
26684 2017-04-07 Martin Liska <mliska@suse.cz>
26687 * ipa-split.c (split_function): Add function part to a same comdat
26690 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26693 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26695 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26697 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26698 (vec_pairing): ...to this.
26699 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26700 (rs6000_sched_init): Adjust for name change.
26701 (struct rs6000_sched_context): Likewise.
26702 (rs6000_init_sched_context): Likewise.
26703 (rs6000_set_sched_context): Likewise.
26705 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26711 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26713 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26714 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26715 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26717 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26719 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26721 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26723 PR rtl-optimization/70703
26724 * ira-color.c (update_conflict_hard_regno_costs): Use
26725 int64_t instead of HOST_WIDE_INT.
26727 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26729 PR rtl-optimization/70478
26730 * lra-constraints.c (process_alt_operands): Disfavor alternative
26731 insn memory operands.
26733 2017-04-07 Jeff Law <law@redhat.com>
26735 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26736 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26738 2017-04-07 Martin Liska <mliska@suse.cz>
26741 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26742 Show error message instead of an ICE.
26744 2017-04-07 Martin Liska <mliska@suse.cz>
26747 * ipa-split.c (split_function): Add function part to a same comdat
26750 2017-04-07 Richard Biener <rguenther@suse.de>
26752 PR middle-end/80341
26753 * tree.c (get_unwidened): Also handle ! for_type case for
26755 * convert.c (do_narrow): Split out from ...
26756 (convert_to_integer_1): ... here. Do not pass final truncation
26757 type to get_unwidened for TRUNC_DIV_EXPR.
26759 2017-04-07 Richard Biener <rguenther@suse.de>
26761 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26763 (aff_combination_const): Adjust.
26764 (aff_combination_scale): Likewise.
26765 (aff_combination_add_elt): Likewise.
26766 (aff_combination_add_cst): Likewise.
26767 (aff_combination_convert): Likewise.
26768 (add_elt_to_tree): Likewise. Remove unused argument.
26769 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26771 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26773 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26775 * config/arm/arm.c (arm_default_short_enums): Use
26776 ARM_DEFAULT_SHORT_ENUMS.
26777 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26779 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26782 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26783 members with redundant out-of-class redeclaration.
26785 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26788 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26789 * config/i386/i386.md (*zero_extendsidi2):
26790 Add (?*x,*x) and (?*v,*v) alternatives.
26792 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26795 * config/i386/i386.c (ix86_expand_builtin)
26796 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26797 mode from insn data. Convert operands to insn operand mode.
26798 Copy operands that don't satisfy insn predicate to a register.
26800 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26802 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26805 2017-04-06 Richard Biener <rguenther@suse.de>
26807 PR tree-optimization/80334
26808 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26809 preserve alignment of accesses.
26811 2017-04-06 Richard Biener <rguenther@suse.de>
26813 PR tree-optimization/80262
26814 * tree-sra.c (build_ref_for_offset): Preserve address-space
26816 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26817 Drop useless address-space information on MEM_REF offsets.
26819 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26821 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26823 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26825 PR rtl-optimization/70703
26826 * ira-color.c (update_conflict_hard_regno_costs): Use
26827 HOST_WIDE_INT instead of long.
26829 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26832 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26833 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26835 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26836 for x86_64 target. Handle -m3dnowa option.
26838 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26840 PR rtl-optimization/70703
26841 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26842 (update_conflict_hard_regno_costs): Use long instead of unsigned
26843 arithmetic for cost calculation.
26845 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26846 Bernd Edlinger <bernd.edlinger@hotmail.de>
26849 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26852 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26855 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26856 ptr_mode with Pmode throughout.
26857 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26858 into probe_stack_range and use DImode.
26860 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26863 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26864 call_eh_return is true.
26866 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26868 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26869 Initialize last_match_fntype_index.
26871 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26874 * tree-nvr.c: Include internal-fn.h.
26875 (pass_return_slot::execute): Ignore internal calls without
26878 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26879 Richard Biener <rguenther@suse.de>
26882 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26883 captures used multiple times, except for the last use.
26884 * generic-match-head.c: Include gimplify.h.
26886 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26888 PR tree-optimization/79390
26889 * target.h (struct noce_if_info): Declare.
26890 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26891 * target.def (noce_conversion_profitable_p): New target hook.
26892 * ifcvt.h (struct noce_if_info): New type, moved from ...
26893 * ifcvt.c (struct noce_if_info): ... here.
26894 (noce_conversion_profitable_p): Renamed to ...
26895 (default_noce_conversion_profitable_p): ... this. No longer
26897 (noce_try_store_flag_constants, noce_try_addcc,
26898 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26899 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26900 instead of noce_conversion_profitable_p.
26901 * config/i386/i386.c: Include ifcvt.h.
26902 (ix86_option_override_internal): Don't override
26903 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26904 (ix86_noce_conversion_profitable_p): New function.
26905 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26906 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26907 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26908 * doc/tm.texi: Regenerated.
26910 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26912 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26915 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26918 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26919 instructions for small multiply cores.
26921 2017-04-04 Jeff Law <law@redhat.com>
26923 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26925 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26926 that are not set by the loop over the elements.
26928 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26931 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26932 int mode, convert_modes it to mode as unsigned, otherwise use
26933 lowpart_subreg to mode rather than SImode.
26934 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26935 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26936 Use DImode instead of SImode for the shift count operand.
26937 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26940 2017-04-04 Richard Biener <rguenther@suse.de>
26942 PR middle-end/80281
26943 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26944 arithmetic done for the negate or the plus. Simplify.
26945 (A - (-B) -> A + B): Likewise.
26946 * fold-const.c (split_tree): Make sure to not negate pointers.
26948 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26950 PR rtl-optimization/60818
26951 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26952 a compare of comparisons with the thing compared if this results
26953 in a different machine mode.
26955 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26957 * alias.c (base_alias_check): Fix typo in comment.
26958 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26959 * cgraphunit.c (symbol_table::compile): Likewise.
26960 * collect2.c (maybe_run_lto_and_relink): Likewise.
26961 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26962 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26963 * config/avr/avr.c (avr_map_op_t): Likewise.
26964 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26965 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26966 * config/epiphany/epiphany.md (movcc): Likewise.
26967 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26968 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26970 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26971 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26972 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26973 * config/sh/sh.c (sh_rtx_costs): Likewise.
26974 * fold-const.c (fold_truth_andor): Likewise.
26975 * genautomata.c (collapse_flag): Likewise.
26976 * gengtype.h (struct type::u::s): Likewise.
26977 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26978 * input.c (FORMAT_AMOUNT): Likewise.
26979 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26980 (known_aggs_to_agg_replacement_list): Likewise.
26981 * ipa-inline-analysis.c: Likewise.
26982 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26983 * ipa-polymorphic-call.c
26984 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26985 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26986 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26988 * modulo-sched.c (apply_reg_moves): Likewise.
26989 * omp-expand.c (build_omp_regions_1): Likewise.
26990 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26991 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26992 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26993 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26994 * value-prof.c: Likewise.
26995 * var-tracking.c (val_reset): Likewise.
26997 2017-04-03 Richard Biener <rguenther@suse.de>
26999 PR tree-optimization/80275
27000 * fold-const.c (split_address_to_core_and_offset): Handle
27003 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27005 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27006 descriptors is at least equal to that of functions.
27008 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27010 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27012 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27015 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27016 (mov<IMOD4:mode>): New expander.
27017 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27019 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27021 PR rtl-optimization/79405
27022 * fwprop.c (propagations_left): New variable.
27023 (forward_propagate_into): Decrement it.
27024 (fwprop_init): Initialize it.
27025 (fw_prop): If the variable has reached zero, stop propagating.
27026 (fwprop_addr): Ditto.
27028 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27031 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27032 a FUNCTION_DECL, pass it as decl instead of origin to
27035 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27037 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27040 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27043 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27044 TARGET_VSX_SMALL_INTEGER.
27046 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27048 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27049 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27051 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27053 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27054 extraction from odd-numbered MSA register.
27056 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27058 PR middle-end/80173
27059 * expmed.c (store_bit_field_1): Don't attempt to create
27060 a word subreg out of hard registers wider than word if they
27061 have HARD_REGNO_NREGS of 1 for their mode.
27063 PR middle-end/80163
27064 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27065 conversions to integer types wider than word and pointer.
27068 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27069 (rtx_equal_for_cselib_p): Pass 0 to it.
27070 * cselib.c (cselib_hasher::equal): Likewise.
27071 (rtx_equal_for_cselib_1): Add depth argument. If depth
27072 is 128, don't look up VALUE locs and punt. Increment
27073 depth in recursive calls when walking VALUE locs.
27075 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27077 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27078 (make_gcov_file_name): Use the canonical path name for generating
27080 (read_line): Fix handling of files with ascii null bytes.
27082 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27084 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27085 to initialise a vector register instead
27086 of using a const_int.
27088 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27090 PR translation/80189
27091 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27092 diagnostic messages.
27094 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27097 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27098 (dfp_diex_<mode>): Update mode of operand 1.
27099 * doc/extend.texi (dxex, dxexq): Document change to return type.
27100 (diex, diexq): Document change to argument type.
27102 2017-03-30 Martin Jambor <mjambor@suse.cz>
27105 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27106 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27107 it reflects the signature changes performed at the callee side.
27108 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27109 to cgraph_build_function_type_skip_args.
27110 (build_function_decl_skip_args): Adjust call to the above function.
27112 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27115 * config/i386/sse.md
27116 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27117 register as dest whenever it is a MEM not rtx_equal_p to the
27118 corresponding dup operand, and when forcing into reg move the
27119 reg into the memory afterwards.
27120 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27121 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27122 for the force_reg mode.
27123 (avx512vl_vextractf128<mode>): Use register as dest either
27124 always when a MEM, or when it is a MEM not rtx_equal_p to the
27125 corresponding dup operand, or even not when it is a CONST_VECTOR
27126 depending on the mode and lo vs. hi.
27127 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27129 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27130 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27131 Likewise. Require that operands[2] is even.
27132 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27133 Remove extraneous parens. Require that operands[2] is a multiple
27135 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27136 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27137 disallow memory then.
27139 2017-03-30 Richard Biener <rguenther@suse.de>
27141 PR tree-optimization/77498
27142 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27143 to non-constants over backedges.
27145 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27147 PR rtl-optimization/80233
27148 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27149 as last_combined_insn. Do not test for BARRIER_P separately.
27151 2017-03-29 Andreas Schwab <schwab@suse.de>
27154 * calls.c (prepare_call_address): Convert funexp to Pmode before
27155 copying to temp reg.
27157 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27159 PR tree-optimization/80158
27160 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27161 Handle possible future case of more than one alternate
27163 (replace_rhs_if_not_dup): Likewise.
27164 (replace_one_candidate): Likewise.
27166 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27168 PR rtl-optimization/80193
27169 * ira.c (ira): Do not check allocation for LRA.
27171 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27173 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27174 (nvptx_output_simt_exit): Declare.
27175 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27176 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27177 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27178 (nvptx_declare_function_name): ...here. Emit declaration of local
27179 memory space buffer for omp_simt_enter insn.
27180 (nvptx_output_unisimt_switch): New.
27181 (nvptx_output_softstack_switch): New.
27182 (nvptx_output_simt_enter): New.
27183 (nvptx_output_simt_exit): New.
27184 * config/nvptx/nvptx.h (struct machine_function): New fields
27185 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27186 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27187 (UNSPECV_SIMT_EXIT): Ditto.
27188 (omp_simt_enter_insn): New insn.
27189 (omp_simt_enter): New expansion.
27190 (omp_simt_exit): New insn.
27191 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27193 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27194 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27195 (expand_GOMP_SIMT_EXIT): New.
27196 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27197 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27198 (GOMP_SIMT_EXIT): Ditto.
27199 * target-insns.def (omp_simt_enter): New insn.
27200 (omp_simt_exit): Ditto.
27201 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27203 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27204 (lower_rec_input_clauses): Likewise.
27205 (lower_lastprivate_clauses): Handle SIMT privatization.
27207 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27208 (ompdevlow_adjust_simt_enter): New.
27209 (find_simtpriv_var_op): New.
27210 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27211 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27213 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27214 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27215 (copy_decl_for_dup_finish): Ditto.
27217 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27219 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27222 * config/i386/i386.c (ix86_option_override_internal): Always
27223 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27225 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27227 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27229 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27231 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27232 mark new edge's irreducible flag accordign to it.
27233 (vect_do_peeling): Check loop preheader edge's irreducible flag
27234 and pass it to function slpeel_add_loop_guard.
27236 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27238 PR tree-optimization/80218
27239 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27240 Update block frequencies and counts.
27242 2017-03-28 Richard Biener <rguenther@suse.de>
27244 PR tree-optimization/78644
27245 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27246 of a simplification result we may not use it at all.
27248 2017-03-28 Richard Biener <rguenther@suse.de>
27251 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27252 without arguments, generate default definition of a SSA name.
27254 2017-03-28 Richard Biener <rguenther@suse.de>
27256 PR middle-end/80222
27257 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27258 TYPE_REF_CAN_ALIAS_ALL references.
27259 * fold-const.c (fold_indirect_ref_1): Likewise.
27261 2017-03-28 Martin Liska <mliska@suse.cz>
27264 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27265 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27267 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27268 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27270 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27271 (EXTRA_SPECS): Define.
27272 (SUBTARGET_EXTRA_SPECS): Likewise.
27273 (SUBTARGET_CPP_SPEC): Likewise.
27274 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27275 SUBTARGET_EXTRA_SPECS.
27276 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27278 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27280 * config/arc/simdext.md (vst64_insn): Update pattern.
27281 (vld32wh_insn): Likewise.
27282 (vld32wl_insn): Likewise.
27283 (vld64_insn): Likewise.
27284 (vld32_insn): Likewise.
27286 2017-03-28 Marek Polacek <polacek@redhat.com>
27289 * fold-const.c (fold_comparison): Use protected_set_expr_location
27290 instead of SET_EXPR_LOCATION.
27292 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27294 * tree.c (add_expr): Avoid name lookup warning.
27296 2017-03-27 Jeff Law <law@redhat.com>
27298 PR tree-optimization/80216
27299 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27300 function name. Limit recursion depth.
27301 (record_temporary_equivalences): Corresponding changes.
27303 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27305 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27308 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27311 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27313 * cfgcleanup.c (reg_note_cfa_p): New array.
27314 (insns_have_identical_cfa_notes): New function.
27315 (old_insns_match_p): Don't cross-jump in between /f
27316 and non-/f instructions. If both i1 and i2 are frame related,
27317 verify all CFA notes, their order and content.
27319 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27322 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27323 HImode and SImode with zero extend to DImode to one insn.
27324 (bswap<mode>2_extenddi): Likewise.
27325 (bswapsi2_extenddi): Likewise.
27326 (bswaphi2_extendsi): Likewise.
27327 (bswaphi2): Combine bswap HImode and SImode into one insn.
27328 Separate memory insns from swapping register.
27329 (bswapsi2): Likewise.
27330 (bswap<mode>2): Likewise.
27331 (bswaphi2_internal): Delete, no longer used.
27332 (bswapsi2_internal): Likewise.
27333 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27334 store, and gpr<-gpr swap insns.
27335 (bswap<mode>2_store): Likewise.
27336 (bswaphi2_reg): Register only splitter, combine with the splitter.
27337 (bswaphi2 splitter): Likewise.
27338 (bswapsi2_reg): Likewise.
27339 (bswapsi2 splitter): Likewise.
27340 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27341 the insns into load, store, and register/register insns.
27342 (bswapdi2_ldbrx): Likewise.
27343 (bswapdi2_load): Likewise.
27344 (bswapdi2_store): Likewise.
27345 (bswapdi2_reg): Likewise.
27347 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27349 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27350 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27352 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27355 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27357 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27358 special handling for target option conflicts between dform
27359 options (-mpower9-dform, -mpower9-dform-vector,
27360 -mpower9-dform-scalar) and -mno-direct-move.
27362 2017-03-27 Richard Biener <rguenther@suse.de>
27364 PR tree-optimization/80181
27365 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27367 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27369 * config/arc/predicates.md (move_double_src_operand): Replace the
27370 call to move_double_src_operand with a call to address_operand.
27372 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27374 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27375 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27376 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27378 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27380 * config/arc/predicates.md (long_immediate_loadstore_operand):
27381 Consider scaled addresses cases.
27383 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27385 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27386 restored when in interrupt.
27387 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27388 doesn't have delay slot.
27390 2017-03-27 Richard Biener <rguenther@suse.de>
27393 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27394 inlined thunk clones.
27396 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27399 * asan.c (instrument_derefs): Copy over last operand from
27400 original COMPONENT_REF to the new COMPONENT_REF with
27401 DECL_BIT_FIELD_REPRESENTATIVE.
27402 * ubsan.c (instrument_object_size): Likewise.
27404 2017-03-27 Richard Biener <rguenther@suse.de>
27406 PR tree-optimization/80170
27407 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27408 sure DR/SCEV didnt fold in constants we do not see when looking
27409 at the reference base alignment.
27411 2017-03-27 Richard Biener <rguenther@suse.de>
27413 PR middle-end/80171
27414 * gimple-fold.c (fold_ctor_reference): Properly guard against
27415 NULL return value from canonicalize_constructor_val.
27417 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27420 * config/i386/i386.c (ix86_expand_builtin)
27421 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27422 flags reg setting and flags reg using instructions.
27423 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27424 clobbering instructions to zero extend op2.
27426 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27428 * doc/install.texi (Configuration) <--with-aix-soname>:
27429 Update link to AIX ld.
27431 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27433 PR rtl-optimization/80160
27434 PR rtl-optimization/80159
27435 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27436 reg_alternate_class into account.
27438 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27441 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27442 to consider in curr_insn_transform.
27444 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27446 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27447 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27448 and emit_mode_inner.
27450 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27452 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27453 argument to the overloaded builtin variants. Use the new flag to
27454 deprecate certain builtin variants.
27455 * config/s390/s390-builtin-types.def: Add new builtin types.
27456 * config/s390/s390-builtins.h: Support new flags field for
27457 overloaded builtins.
27458 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27459 (s390_macro_to_expand): Enable vector float data type.
27460 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27461 builtins by incrementing the __VEC__ version number.
27462 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27464 (s390_resolve_overloaded_builtin): Emit error messages depending
27465 on the builtin flags.
27466 * config/s390/s390.c (s390_expand_builtin): Support additional
27467 flags argument. Change error message to match the messages
27468 emitted in s390-c.c.
27469 * config/s390/s390.md: New UNSPEC_* constants.
27470 (op_type): Add new instruction types.
27471 * config/s390/vecintrin.h: Add new builtins and test data class
27473 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27474 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27475 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27476 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27477 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27478 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27480 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27481 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27482 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27483 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27485 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27486 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27487 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27488 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27490 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27491 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27492 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27493 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27494 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27495 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27496 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27498 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27499 ("vec_scatter_element<V_HW_4:mode>_DI")
27500 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27501 ("vec_fpint<mode>", "vflls")
27502 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27503 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27504 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27505 ("*vec_cmphe<mode>_cc"): ... these.
27507 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27508 mode constant instead of magic value.
27510 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27512 * config/s390/s390.c (s390_expand_vec_compare): Support other
27513 vector floating point modes than just V2DF.
27514 (s390_expand_vcond): Likewise.
27515 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27516 (s390_cannot_change_mode_class): Prevent mode changes between TF
27517 and V1TF in vector registers.
27518 * config/s390/s390.md (DF, SF): New mode attributes.
27519 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27520 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27521 SFmode support for VRs.
27522 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27524 (VFT, VF_HW): New mode iterators.
27525 (vw, sdx): New mode attributes.
27526 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27527 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27528 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27529 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27530 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27531 also the new vector floating point modes. Renaming to ...
27533 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27534 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27535 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27536 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27537 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27538 ("vec_unordered<mode>"): ... these.
27540 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27541 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27542 ("*vec_extendv2df"): New insn definitions.
27544 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27546 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27547 ("mulditi3_2", "*muldi3_sign"): New patterns.
27548 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27549 rename the pattern definition.
27551 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27553 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27555 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27557 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27559 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27560 instruction if possible.
27561 * config/s390/vector.md (vec_halfnumelts): New mode
27563 ("*vec_vllezlf<mode>"): New pattern.
27565 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27567 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27568 ("popcountv4si2", "popcountv2di2"): Rename to ...
27569 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27570 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27572 ("popcount<mode>2_vxe"): New pattern.
27574 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27576 * common/config/s390/s390-common.c (processor_flags_table): Add
27578 * config.gcc: Add arch12.
27579 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27580 Default to arch12 for unknown CPU model numbers.
27581 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27582 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27583 PROCESSOR_max sanity check.
27584 * config/s390/s390-opts.h (enum processor_type): Add
27586 * config/s390/s390.c (processor_table): Add arch12.
27587 (s390_expand_builtin): Add check for B_VXE flag.
27588 (s390_issue_rate): Add PROCESSOR_ARCH12.
27589 (s390_get_sched_attrmask): Likewise.
27590 (s390_get_unit_mask): Likewise.
27591 (s390_sched_score): Enable z13 scheduling for arch12.
27592 (s390_sched_reorder): Likewise.
27593 (s390_sched_variable_issue): Likewise.
27594 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27596 (s390_tune_attr): Use z13 scheduling also for arch12.
27597 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27598 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27599 (TARGET_VXE_P): New macros.
27600 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27601 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27602 * config/s390/s390.opt: Add arch12 as processor_type.
27604 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27606 * config/s390/s390.md
27607 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27608 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27609 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27611 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27612 Rename expanders to ...
27614 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27615 ("fixuns_truncdddi2_emu"): ... these.
27617 ("fixuns_trunc<mode>si2_emu"): New expander.
27619 ("*fixuns_truncdfdi2_z13"): Rename to ...
27620 ("*fixuns_truncdfdi2_vx"): ... this.
27622 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27624 * config/s390/2964.md: Remove the single element vector compare
27625 instructions which are no longer used.
27626 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27628 (s390_canonicalize_comparison): Remove handling of DFmode
27630 (s390_expand_vec_compare_scalar): Remove function.
27631 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27632 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27634 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27636 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27638 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27640 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27641 will anyway by matched by mov<mode>_64dfp.
27643 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27645 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27646 vlef/vstef. Add missing operand to vleif.
27648 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27650 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27651 pair for all vector types with 64 bit elements.
27652 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27653 * config/s390/vector.md (V_HW_64): ... here.
27654 (V_128_NOSINGLE): New mode iterator.
27655 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27656 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27657 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27658 ("*vec_load_pairv2di"): Change to ...
27659 ("*vec_load_pair<mode>"): ... this one.
27661 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27663 * config/s390/constraints.md: Add comments.
27664 (jKK): Reject element sizes > 8 bytes.
27665 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27667 * config/s390/s390.md: Add the s_operand checks formerly in
27668 s390_split_ok_p to various splitters where they are still
27670 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27671 for 128 bit vectors. Plus two splitters.
27673 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27675 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27678 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27681 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27682 error if the boundary argument is not constant.
27684 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27686 PR rtl-optimization/80112
27687 * loop-doloop.c (doloop_condition_get): Don't check condition
27688 if cmp isn't SET with IF_THEN_ELSE src.
27690 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27692 PR tree-optimization/80158
27693 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27694 replacing a candidate statement, also replace it for the
27695 candidate's alternate interpretation.
27696 (replace_rhs_if_not_dup): Likewise.
27697 (replace_one_candidate): Likewise.
27699 2017-03-24 Richard Biener <rguenther@suse.de>
27701 PR tree-optimization/80167
27702 * graphite-isl-ast-to-gimple.c
27703 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27705 (translate_isl_ast_to_gimple::get_rename): Likewise.
27707 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27709 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27710 handling of certain combinations of target options, including the
27711 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27712 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27714 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27717 * config/arm/arm.md (*load_multiple): Add reload_completed to
27718 matching condition.
27720 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27721 Richard Biener <rguenth@suse.de>
27723 PR tree-optimization/79908
27724 PR tree-optimization/80136
27725 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27726 been cast away, gimplify_and_add suffices.
27728 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27730 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27732 2017-03-23 Richard Biener <rguenther@suse.de>
27734 PR tree-optimization/80032
27735 * gimplify.c (gimple_push_cleanup): Forced unconditional
27736 cleanups still have to go to the conditional_cleanups
27739 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27741 PR tree-optimization/80072
27742 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27744 (next_operand_entry_id): Change type to unsigned int.
27745 (sort_by_operand_rank): Make sure to return the right return value
27746 even if unsigned fields are bigger than INT_MAX.
27747 (struct oecount): Change cnt and id type to unsigned int.
27748 (oecount_hasher::equal): Formatting fix.
27749 (oecount_cmp): Make sure to return the right return value
27750 even if unsigned fields are bigger than INT_MAX.
27751 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27754 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27755 TREE_READONLY on result if writing it more than once.
27758 * doc/invoke.texi (-fsanitize=thread): Document that with
27759 -fnon-call-exceptions atomics are not able to throw
27763 * tsan.c: Include tree-eh.h.
27764 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27765 maybe_clean_or_replace_eh_stmt where needed.
27766 (instrument_memory_accesses): Add cfg_changed argument.
27767 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27768 if it returned true.
27769 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27771 PR rtl-optimization/63191
27772 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27773 wrapper function, moved the whole old content into ...
27774 (ix86_delegitimize_address_1): ... this. New inline function.
27775 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27776 true as last argument instead of ix86_delegitimize_address.
27778 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27780 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27781 cortexa57_branch_cost.
27783 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27785 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27787 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27790 * doc/md.texi (Constraints): Document wA constraint.
27791 * config/rs6000/constraints.md (wA): New.
27792 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27793 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27794 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27795 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27797 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27800 * gimplify.c (is_oacc_declared): New function.
27801 (oacc_default_clause): Use it to set default flags for acc declared
27802 variables inside parallel regions.
27803 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27804 declared variables.
27805 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27806 declare attribute to any decl as necessary.
27808 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27811 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27812 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27813 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27814 (arm_arch_lpae): This.
27815 * config/arm/arm.c (arm_arch7ve): Rename into ...
27816 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27817 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27820 2017-03-22 Martin Liska <mliska@suse.cz>
27823 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27824 error message instead of an ICE.
27826 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27828 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27830 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27832 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27834 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27837 2017-03-21 Martin Sebor <msebor@redhat.com>
27839 * doc/extend.texi: Use "cannot" instead of "can't."
27840 * doc/hostconfig.texi: Same.
27841 * doc/install.texi: Same.
27842 * doc/invoke.texi: Same.
27843 * doc/loop.texi: Same.
27844 * doc/md.texi: Same.
27845 * doc/objc.texi: Same.
27846 * doc/rtl.texi: Same.
27847 * doc/tm.texi: Same.
27848 * doc/tm.texi.in: Same.
27849 * doc/trouble.texi: Same.
27851 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27854 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27855 (collect_checksum_attributes): Set it.
27856 (die_checksum_ordered): Use it.
27858 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27860 PR tree-optimization/79908
27861 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27862 change: For a VA_ARG whose LHS has been cast away, use
27863 force_gimple_operand to construct the side effects.
27865 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27867 PR translation/80001
27868 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27869 more amenable to translation.
27870 (oacc_loop_auto_partitions): Likewise.
27872 2017-03-21 Marek Polacek <polacek@redhat.com>
27873 Martin Sebor <msebor@redhat.com>
27875 PR tree-optimization/80109
27876 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27877 on INTEGRAL_TYPE_P.
27879 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27880 Segher Boessenkool <segher@kernel.crashing.org>
27883 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27884 check reg_used_between_p between insn and one of succ or succ2
27885 depending on if succ is artificial insn not inserted into insn
27888 2017-03-21 Martin Liska <mliska@suse.cz>
27890 PR gcov-profile/80081
27891 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27892 * doc/gcc.texi: Include gcov-dump stuff.
27893 * doc/gcov-dump.texi: New file.
27895 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27897 PR rtl-optimization/79150
27898 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27899 conditional jump, if the jump is the last insn of the loop.
27901 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27902 Richard Biener <rguenth@suse.de>
27904 PR tree-optimization/79908
27905 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27906 been cast away, use force_gimple_operand to construct the side
27909 2017-03-21 Martin Liska <mliska@suse.cz>
27911 PR libfortran/79956
27912 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27915 2017-03-21 Brad Spengler <spender@grsecurity.net>
27918 * plugin.c (htab_hash_plugin): New function.
27919 (add_new_plugin): Use it and adjust.
27920 (parse_plugin_arg_opt): Adjust.
27921 (init_one_plugin): Likewise.
27923 2017-03-21 Richard Biener <rguenther@suse.de>
27925 PR tree-optimization/80032
27926 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27927 if set force the cleanup to happen unconditionally.
27928 (gimplify_target_expr): Push inserted clobbers with force_uncond
27929 to avoid them being removed by control-dependent DCE.
27931 2017-03-21 Richard Biener <rguenther@suse.de>
27933 PR tree-optimization/80122
27934 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27935 va_arg_pack_len when the inlined call stmt requires pack
27937 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27939 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27942 * tsan.c (instrument_builtin_call): If the memory model argument
27943 is not a constant, assume it is valid.
27946 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27949 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27951 PR rtl-optimization/79910
27952 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27953 if its dest is used by an insn before I2 (other than the combined
27954 insns themselves, which are properly handled already).
27956 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27959 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27961 * combine.c (record_used_regs): New static function.
27962 (try_combine): Handle situations where there is an additional
27963 instruction between I2 and I3 which needs to have a LOG_LINK
27967 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27969 * combine.c (try_combine): Delete redundant i1 test. Call
27970 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27972 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27975 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27976 alternatives 13/14.
27978 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27980 PR tree-optimization/80054
27981 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27982 the optimization if a PHI or any of its arguments is not dominated
27983 by the candidate's basis. Use gphi* rather than gimple* as
27985 (replace_profitable_candidates): Clean up a gimple* variable that
27986 should be a gphi* variable.
27988 2017-03-20 Martin Sebor <msebor@redhat.com>
27991 * doc/extend.texi (attribute constructor): Document present limitation.
27993 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27996 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27997 __POWER9_VECTOR__ #ifdef control, change template definition to
27998 use Power9-specific built-in function.
27999 (vec_any_eq): Likewise.
28000 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28001 to control outcomes from this test.
28002 (vector_ae_<mode>p): For VEC_F modes, likewise.
28004 2017-03-20 Ian Lance Taylor <iant@google.com>
28006 * config/i386/i386.c (ix86_function_regparm): Save an extra
28007 register for -fsplit-stack with DECL_STATIC_CHAIN.
28009 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28012 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28013 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28015 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28017 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28019 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28022 2017-03-20 Marek Polacek <polacek@redhat.com>
28025 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28027 2017-03-20 Richard Biener <rguenther@suse.de>
28029 PR tree-optimization/80113
28030 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28031 allocate extra SSA name for PHI def.
28032 (add_close_phis_to_outer_loops): Likewise.
28033 (add_close_phis_to_merge_points): Likewise.
28034 (copy_loop_close_phi_args): Likewise.
28035 (copy_cond_phi_nodes): Likewise.
28037 2017-03-20 Martin Liska <mliska@suse.cz>
28039 PR middle-end/79753
28040 * tree-chkp.c (chkp_build_returned_bound): Do not build
28041 returned bounds for a LHS that's not a BOUNDED_P type.
28043 2017-03-20 Martin Liska <mliska@suse.cz>
28047 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28048 COMPLEX_CST and VECTOR_CST.
28050 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28053 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28054 target operand. A new splitter adds the clobber statement in case
28055 the target operand is dead anyway.
28057 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28059 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28060 to age-old versions of binutils and glibc.
28062 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28064 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28066 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28068 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28070 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28072 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28073 requirement for binutils 2.13.
28075 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28077 * combine.c (try_combine): Delete redundant i1 test. Call
28078 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28080 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28082 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28083 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28085 <riscv64-*-elf>: Re-arrange section
28086 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28087 <riscv32-*-linux>: Likewise.
28088 <riscv64-*-elf>: Likewise
28089 <riscv64-*-linux>: Likewise.
28091 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28094 * aarch64.opt(verbose-cost-dump): Fix typo.
28096 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28099 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28100 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28102 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28104 * reload.c (find_reloads): When reloading a nonoffsettable address,
28105 use RELOAD_OTHER for it and its address reloads.
28107 PR rtl-optimization/79910
28108 * combine.c (record_used_regs): New static function.
28109 (try_combine): Handle situations where there is an additional
28110 instruction between I2 and I3 which needs to have a LOG_LINK
28113 2017-03-17 Jeff Law <law@redhat.com>
28115 PR tree-optimization/71437
28116 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28117 conditional in the hash table first.
28118 (vrp_dom_walker::before_dom_children): Extract condition from
28119 ASSERT_EXPR. Record condition, its inverion and any implied
28120 conditions as well.
28122 2017-03-17 Marek Polacek <polacek@redhat.com>
28123 Markus Trippelsdorf <markus@trippelsdorf.de>
28125 PR tree-optimization/80079
28126 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28129 2017-03-17 Richard Biener <rguenther@suse.de>
28131 PR middle-end/80075
28132 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28133 Properly verify the LHS before the RHS possibly claims to be
28135 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28138 2017-03-17 Martin Jambor <mjambor@suse.cz>
28140 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28141 (List of -O2 options): Likewise.
28142 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28145 2017-03-17 Tom de Vries <tom@codesourcery.com>
28147 * gcov-dump.c (print_usage): Print bug_report_url.
28149 2017-03-17 Richard Biener <rguenther@suse.de>
28151 PR middle-end/80050
28152 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28153 (parser::peek): Likewise.
28155 2017-03-17 Richard Biener <rguenther@suse.de>
28157 PR tree-optimization/80048
28158 * sese.c (free_sese_info): Properly release rename_map and
28159 copied_bb_map elements.
28161 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28163 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28164 Add linked-list forward and backlinks. Insert on
28165 construction, remove on destruction.
28166 (class pass_store_merging): Add m_stores_head field.
28167 (pass_store_merging::terminate_and_process_all_chains):
28168 Iterate over m_stores_head list.
28169 (pass_store_merging::terminate_all_aliasing_chains):
28171 (pass_store_merging::execute): Check for debug stmts first.
28172 Push new chains onto the m_stores_head stack.
28174 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28177 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28178 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28179 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28181 2017-03-16 Jeff Law <law@redhat.com>
28183 PR tree-optimization/71437
28184 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28185 member function. Implementation moved into after_dom_children
28186 member function and into the threader's thread_outgoing_edges
28188 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28189 some code into new thread_outgoing_edges.
28190 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28191 definition. Simplify marker handling (do it here). Assume we always
28192 have the available expression and the const/copies tables.
28193 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28195 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28196 * tree-vrp.c (equiv_stack): No longer file scoped.
28197 (vrp_dom_walker): New class.
28198 (vrp_dom_walker::before_dom_children): New member function.
28199 (vrp_dom_walker::after_dom_children): Likewise.
28200 (identify_jump_threads): Setup domwalker. Use it rather than
28201 walking edges in a random order by hand. Simplify setup/finalization.
28202 (finalize_jump_threads): Remove.
28203 (vrp_finalize): Do not call identify_jump_threads here.
28204 (execute_vrp): Do it here instead and call thread_through_all_blocks
28207 PR tree-optimization/71437
28208 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28210 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28212 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28213 (dom_opt_dom_walker::thread_across_edge): Remove
28214 handle_dominating_asserts argument. All callers changed.
28215 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28216 changes. Remove calls to lhs_of_dominating_assert. Other
28217 uses of handle_dominating_asserts turn into unconditional code
28218 (simplify_control_stmt_condition_1): Likewise.
28219 (simplify_control_stmt_condition): Likewise.
28220 (thread_through_normal_block, thread_across_edge): Likewise.
28221 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28222 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28223 object if it is not an SSA_NAME.
28224 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28225 before calling into the VRP specific simplifiers.
28226 (identify_jump_threads): Remove handle_dominating_asserts
28229 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28232 * tree-diagnostic.c (default_tree_printer): No longer static.
28233 * tree-diagnostic.h (default_tree_printer): New prototype.
28235 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28237 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28238 Change ins into fmov.
28240 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28242 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28243 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28244 Use h_con constraint for operand 1.
28245 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28246 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28248 2017-03-15 Jeff Law <law@redhat.com>
28250 PR tree-optimization/71437
28251 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28252 (record_temporary_equivalences): Use it.
28254 PR tree-optimization/71437
28255 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28256 tree-ssa-scopedtables.
28257 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28258 (record_conditions, record_cond, vuse_eq): Likewise.
28259 (record_edge_info): Adjust to API tweak of record_conditions.
28260 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28261 (record_temporary_equivalences, optimize_stmt): Likewise.
28262 (eliminate_redundant_computations): Likewise.
28263 (record_equivalences_from_stmt): Likewise.
28264 * tree-ssa-scopedtables.c: Include options.h and params.h.
28265 (vuse_eq): New function, moved from tree-ssa-dom.c
28266 (build_and_record_new_cond): Likewise.
28267 (record_conditions): Likewise. Accept vector of conditions rather
28268 than edge_equivalence structure for first argument.
28269 for the first argument.
28270 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28271 from tree-ssa-dom.c.
28272 (avail_exprs_stack::record_cond): Likewise.
28273 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28274 from tree-ssa-dom.c.
28275 (avail_exprs_stack): Add new member functions lookup_avail_expr
28277 (record_conditions): Declare.
28279 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28282 * lra-constraints.c (process_alt_operands): Increase reject for
28283 reloading an input/output operand.
28285 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28288 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28289 insns to convert from signed/unsigned char/short to IEEE 128-bit
28291 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28293 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28296 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28297 subreg of inner mode for values already in registers.
28299 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28301 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28302 iteration reg is used after the loop.
28304 2017-03-14 Martin Sebor <msebor@redhat.com>
28306 PR tree-optimization/79800
28307 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28308 precision in negative-positive range.
28309 (format_floating): Call non-const overload with adjusted precision.
28311 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28314 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28317 2017-03-14 Martin Sebor <msebor@redhat.com>
28319 PR middle-end/80020
28320 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28321 * builtins.def (aligned_alloc): Use it.
28324 * Makefile.in (GTFILES): Add calls.c.
28325 * calls.c: Include "gt-calls.h".
28327 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28329 PR rtl-optimization/79728
28330 * regs.h (struct target_regs): New field
28331 x_contains_allocatable_regs_of_mode.
28332 (contains_allocatable_regs_of_mode): New macro.
28333 * reginfo.c (init_reg_sets_1): Initialize it, and change
28334 contains_reg_of_mode so it includes global regs as well.
28335 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28336 rather than contains_regs_of_mode.
28338 2017-03-14 Martin Liska <mliska@suse.cz>
28340 * doc/invoke.texi: Document options that can't be combined with
28341 -fcheck-pointer-bounds.
28343 2017-03-14 Martin Liska <mliska@suse.cz>
28345 PR middle-end/79831
28346 * doc/invoke.texi (-Wchkp): Document the option.
28348 2017-03-14 Martin Liska <mliska@suse.cz>
28350 * Makefile.in: Install gcov-dump.
28352 2017-03-14 Martin Liska <mliska@suse.cz>
28354 * multiple_target.c (expand_target_clones): Bail out for
28355 an invalid attribute.
28357 2017-03-14 Richard Biener <rguenther@suse.de>
28359 * alias.c (struct alias_set_entry): Pack properly.
28360 * cfgloop.h (struct loop): Likewise.
28361 * cse.c (struct set): Likewise.
28362 * ipa-utils.c (struct searchc_env): Likewise.
28363 * loop-invariant.c (struct invariant): Likewise.
28364 * lra-remat.c (struct cand): Likewise.
28365 * recog.c (struct change_t): Likewise.
28366 * rtl.h (struct address_info): Likewise.
28367 * symbol-summary.h (function_summary): Likewise.
28368 * tree-loop-distribution.c (struct partition): Likewise.
28369 * tree-object-size.c (struct object_size_info): Likewise.
28370 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28371 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28372 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28373 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28374 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28375 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28376 (struct _stmt_vec_info): Likewise.
28378 2017-03-14 Martin Liska <mliska@suse.cz>
28381 * multiple_target.c (create_dispatcher_calls): Check that
28382 a target can create a function dispatcher.
28384 2017-03-14 Martin Liska <mliska@suse.cz>
28387 * multiple_target.c (expand_target_clones): Drop local.local
28388 flag for default implementation.
28390 2017-03-14 Richard Biener <rguenther@suse.de>
28392 PR tree-optimization/80030
28393 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28395 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28397 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28398 gcc_fallthrough() instead of __attribute__((fallthrough));
28400 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28402 * doc/gcc.texi: Remove "up" link to (DIR).
28403 * doc/gccint.texi: Ditto.
28405 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28407 * doc/install.texi (Specific) <avr>: Remove reference to
28410 2017-03-13 Jeff Law <law@redhat.com>
28412 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28413 attribute rather than comments.
28415 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28416 match_scratch operand is highest.
28418 2017-03-13 Martin Liska <mliska@suse.cz>
28420 PR middle-end/78339
28421 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28422 is a CHKP clone, use original declaration.
28424 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28426 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28427 (arc_conditional_register_usage): Use a different allocation order
28428 when optimizing for size.
28429 * common/config/arc/arc-common.c (arc_option_optimization_table):
28430 Section anchors default on when optimizing for size.
28432 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28434 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28436 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28438 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28439 * config/arc/arc.md (cpu_facility): Add cd variant.
28440 (*movqi_insn): Add code density variant.
28441 (*movhi_insn): Likewise.
28442 (*movqi_insn): Likewise.
28443 (*addsi3_mixed): Likewise.
28444 (subsi3_insn): Likewise.
28446 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28448 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28450 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28452 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28453 expressions with MINUS and UNARY ops.
28455 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28458 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28460 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28461 between vec_select and vector argument.
28462 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28463 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28464 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28465 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28466 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28467 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28469 2017-03-13 Richard Biener <rguenther@suse.de>
28472 * params.def (vect-max-peeling-for-alignment): Fix typo.
28474 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28476 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28477 issue that only occurred with binutils below 2.18.
28479 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28481 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28482 refer to binutils 2.11/2.12 minimum.
28484 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28486 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28487 ftp.kernel.org and simplify binutils requirement.
28489 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28491 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28493 (Optimize Options): Ditto. Also remove redundancy.
28495 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28497 PR translation/79848
28498 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28500 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28501 to G_ to avoid double translation.
28503 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28505 PR translation/79923
28506 * auto-profile.c (get_combined_location): Convert leading
28507 character of diagnostics to lower case and remove trailing period.
28508 (read_profile): Likewise for various diagnostics.
28509 * config/arm/arm.c (arm_option_override): Remove trailing period
28510 from various diagnostics.
28511 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28512 (msp430_expand_delay_cycles): Likewise.
28514 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28517 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28518 full command-line argument, rather than just "str".
28519 (aarch64_validate_march): Likewise.
28520 (aarch64_validate_mtune): Likewise.
28522 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28524 PR rtl-optimization/78911
28525 * lra-assigns.c (must_not_spill_p): New function.
28526 (spill_for): Use it.
28528 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28530 PR tree-optimization/79981
28531 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28532 ATOMIC_COMPARE_EXCHANGE ifn result.
28533 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28534 IFN_ATOMIC_COMPARE_EXCHANGE.
28536 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28539 * opts.c (parse_sanitizer_options): Add missing question mark to
28540 "did you mean" message.
28542 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28544 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28546 (VMULEUH_UNS): Likewise.
28547 (VMULOUB_UNS): Likewise.
28548 (VMULOUH_UNS): Likewise.
28549 * config/rs6000/rs6000.c (builtin_function_type): Remove
28550 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28552 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28555 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28556 x with result of extra_parsing_for_operand_code_0.
28557 (function_reader::extra_parsing_for_operand_code_0): Convert
28558 return type from void to rtx, returning x. When reading
28559 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28560 larger size containing struct block_symbol.
28562 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28564 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28565 -mfloat128-hardware without -m64.
28567 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28570 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28571 entries to the case statement that marks unsigned arguments to
28572 overloaded functions.
28574 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28576 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28577 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28579 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28582 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28583 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28585 2017-03-10 Martin Liska <mliska@suse.cz>
28589 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28590 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28593 2017-03-10 Olivier Hainque <hainque@adacore.com>
28595 * tree-switch-conversion (array_value_type): Start by resetting
28596 candidate type to it's main variant.
28598 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28600 PR rtl-optimization/79909
28601 * combine.c (try_combine): Use simplify_replace_rtx on individual
28602 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28603 of the whole CALL_INSN_FUNCTION_USAGE.
28605 PR tree-optimization/79972
28606 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28607 get_range_info on SSA_NAMEs. Formatting fixes.
28609 2017-03-10 Richard Biener <rguenther@suse.de>
28610 Jakub Jelinek <jakub@redhat.com>
28612 PR tree-optimization/77975
28613 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28614 edge to be constant.
28615 (get_val_for): For constant x return it. Formatting fix.
28616 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28617 would use the same bases as the current one.
28619 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28621 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28622 instead of vec_select for V1TImode.
28623 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28625 (VSX_LE_128): Add V1TI to this mode iterator.
28626 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28627 (*vsx_le_perm_store_<mode>): Likewise.
28628 (pre-reload splitter for VSX stores): Likewise.
28629 (post-reload splitter for VSX stores): Likewise.
28630 (*vsx_xxpermdi2_le_<mode>): Likewise.
28631 (*vsx_lxvd2x2_le_<mode>): Likewise.
28632 (*vsx_stxvd2x2_le_<mode>): Likewise.
28634 2017-03-09 Michael Eager <eager@eagercon.com>
28636 Correct failures with --enable-checking=yes,rtl.
28638 * config/microblaze/microblaze.c (microblaze_expand_shift):
28639 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28640 test for const0_rtx.
28641 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28642 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28644 2017-03-09 Richard Biener <rguenther@suse.de>
28646 PR tree-optimization/79977
28647 * graphite-scop-detection.c (scop_detection::merge_sese):
28648 Handle the case of extra exits to blocks dominating the entry.
28650 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28652 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28655 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28657 PR rtl-optimization/79949
28658 * lra-constraints.c (process_alt_operands): Check memory when
28659 trying to predict a cycle. Print about the overall increase.
28661 2017-03-09 Richard Biener <rguenther@suse.de>
28663 PR middle-end/79971
28664 * gimple-expr.c (useless_type_conversion_p): Preserve
28665 TYPE_SATURATING for fixed-point types.
28667 2017-03-09 Richard Biener <rguenther@suse.de>
28670 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28671 alignment of BLKmode params.
28673 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28676 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28677 (VALL_NO_V2Q): Likewise.
28679 * config/aarch64/aarch64-simd.md
28680 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28682 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28683 VALL_NO_V2Q mode iterator.
28684 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28686 2017-03-09 Martin Liska <mliska@suse.cz>
28688 PR tree-optimization/79631
28689 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28690 tree_int_cst_sign_bit just for INTEGER constants.
28692 2017-03-09 Martin Liska <mliska@suse.cz>
28696 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28699 2017-03-09 Marek Polacek <polacek@redhat.com>
28702 * tree.c (inchash::add_expr): Handle TREE_VEC.
28704 2017-03-09 Martin Liska <mliska@suse.cz>
28707 (chkp_narrow_size_and_offset): New function.
28708 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28709 (void chkp_parse_bit_field_ref): New function.
28710 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28711 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28713 2017-03-09 Martin Liska <mliska@suse.cz>
28716 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28717 (chkp_find_bounds_1): Remove gcc_unreachable.
28719 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28722 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28723 BUILT_IN_SYNC*, determine the access type from the size suffix and
28724 always build a MEM_REF with that type. Handle forgotten
28725 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28728 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28729 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28730 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28731 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28732 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28733 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28734 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28735 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28736 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28737 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28738 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28739 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28740 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28741 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28742 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28743 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28744 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28745 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28746 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28747 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28748 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28749 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28750 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28751 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28752 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28753 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28754 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28755 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28756 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28757 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28758 definitions outside of __OPTIMIZE__ guarded section.
28761 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28762 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28763 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28764 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28767 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28769 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28770 ("vfenez<mode>"): Add missing constraints.
28772 2017-03-08 Martin Sebor <msebor@redhat.com>
28775 * config/nds32/nds32.c (nds32_option_override):
28776 Fix misspelled diagnostic.
28778 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28781 * gimplify.c (gimplify_omp_for): Replace index var in outer
28782 taskloop statement with an artificial variable and add
28783 OMP_CLAUSE_PRIVATE clause for it.
28785 2017-03-08 Richard Biener <rguenther@suse.de>
28787 PR tree-optimization/79955
28788 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28789 for accesses that are completely outside of the variable.
28791 2017-03-08 Andrew Haley <aph@redhat.com>
28793 PR tree-optimization/79943
28794 * tree-ssa-loop-split.c (compute_new_first_bound): When
28795 calculating the new upper bound, (END-BEG) should be added, not
28798 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28800 * config/avr/avr.md (setmemhi): Make sure match_dup
28801 operand number comes before match_scratch.
28803 2017-03-08 Richard Biener <rguenther@suse.de>
28805 PR tree-optimization/79920
28806 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28807 with ncopies == 1 to ...
28808 (vect_transform_slp_perm_load): ... here. Properly compute
28809 all element loads by iterating VF times over the group. Do
28810 not handle ncopies (computed in a broken way) in
28811 vect_create_mask_and_perm.
28813 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28816 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28817 is a uniform vector, use uniform_vector_p return value instead of
28818 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28820 2017-03-07 Marek Polacek <polacek@redhat.com>
28822 PR middle-end/79809
28823 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28824 (alloca_call_type): Likewise.
28826 2017-03-07 Martin Liska <mliska@suse.cz>
28828 * gcov.c (process_args): Put comment to correct location.
28830 2017-03-07 Martin Liska <mliska@suse.cz>
28832 PR middle-end/68270
28833 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28834 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28835 (chkp_narrow_bounds_for_field): Likewise.
28836 (chkp_parse_array_and_component_ref): Pass one more argument to
28839 2017-03-07 Richard Biener <rguenther@suse.de>
28841 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28844 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28846 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28847 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28849 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28852 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28853 to end of description.
28854 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28856 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28858 PR rtl-optimization/79901
28859 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28861 (*avx512f_<code><mode>3<mask_name>): ... this.
28862 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28863 iterator instead of VI8_AVX2_AVX512BW.
28865 PR rtl-optimization/79901
28866 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28867 min/max expander, expand it using expand_vec_cond_expr.
28870 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28873 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28876 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28877 to void * for PCH reasons.
28878 * dwarf2out.c (output_loc_operands, output_die): Cast
28879 v.val_vec.array to unsigned char *.
28881 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28884 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28887 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28889 PR rtl-optimization/79571
28890 * lra-constraints.c (process_alt_operands): Calculate static
28891 reject and subtract it from overall when only addresses will be
28894 2017-03-06 Julia Koval <julia.koval@intel.com>
28897 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28898 incoming stack boundary to 128 for 64-bit targets.
28900 2017-03-06 Richard Biener <rguenther@suse.de>
28902 PR tree-optimization/79894
28903 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28904 to NULL after folding it.
28906 2017-03-06 Richard Biener <rguenther@suse.de>
28908 PR tree-optimization/79824
28909 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28910 check disabling peeling for gaps.
28912 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28914 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28915 attributes): Document gettimeofday.
28917 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28919 * config/s390/s390.c (s390_option_override_internal): Set
28920 PARAM_MIN_VECT_LOOP_BOUND
28922 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28924 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28925 * config/s390/s390.md: Likewise.
28927 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28930 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28931 (<avx2_avx512>_perm<mode>): Rename to ...
28932 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28934 (<avx512>_perm<mode>_mask): Rename to ...
28935 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28937 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28938 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28939 instead of VI8F_256_512.
28940 (avx512f_perm<mode>): New define_expand.
28941 (avx512f_perm<mode>_mask): Likewise.
28942 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28943 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28945 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28947 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28948 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28950 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28952 2017-03-06 Martin Liska <mliska@suse.cz>
28955 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28956 when having a SSA NAME w/o VAR_DECL assigned to it.
28958 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28960 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28961 msa_dpsub_<su>_d): Fix MODE for vec_select.
28963 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28965 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28967 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28969 2017-03-06 Richard Biener <rguenther@suse.de>
28971 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28972 * plugin.c (register_plugin_info): Likewise.
28973 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28975 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28977 * config/i386/sse.md (sse_storehps, sse_storelps,
28978 avx_<castmode><avxsizesuffix>_<castmode>,
28979 avx512f_<castmode><avxsizesuffix>_<castmode>,
28980 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28981 in condition that at least one operand is not a MEM.
28983 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28985 PR middle-end/79805
28986 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28987 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28989 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28990 gimple_call_nothrow_p flag based on whether original builtin can throw.
28991 If it can, emit following stmts on the fallthrough edge.
28992 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28993 don't create new bb if inserting just debug stmts on the edge, try to
28994 insert them on the fallthru bb or just reset debug stmts.
28996 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28999 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29000 restore recog_data (including the operand rtxes inside it) around
29001 the call to get_insn_template.
29003 2017-03-03 Martin Sebor <msebor@redhat.com>
29005 PR tree-optimization/79699
29006 * context.c (context::~context): Free MPFR caches to avoid
29007 a memory leak on program exit.
29009 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29011 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29012 Use wide_int::ulow () instead of .elt (0).
29014 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29016 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29017 (*pushxf): Limit oF constraint to 32bit targets and add oC
29018 constraint for 64bit targets.
29019 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29020 (*pushdf): Change rmF constraint to rmC.
29022 2017-03-03 Martin Liska <mliska@suse.cz>
29024 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29025 Remove unused variable.
29027 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29030 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29031 is a memory operand, increase num_memory.
29032 (ix86_expand_args_builtin): Likewise.
29034 2017-03-03 Jan Hubicka <jh@suse.cz>
29037 * ipa-devirt.c (maybe_record_node): Properly handle
29038 __cxa_pure_virtual visibility.
29040 2017-03-03 Martin Liska <mliska@suse.cz>
29042 PR tree-optimization/79803
29043 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29045 (pass_loop_prefetch::execute): Disabled optimization if an
29046 assumption about L1 cache size is not met.
29048 2017-03-03 Martin Liska <mliska@suse.cz>
29050 PR rtl-optimization/79574
29051 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29052 (hash_scan_set): Likewise.
29053 (dump_hash_table): Likewise.
29054 (hoist_code): Likewise.
29056 2017-03-03 Richard Biener <rguenther@suse.de>
29058 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29059 in place of uhigh ().
29060 (fixed_convert_from_real): Likewise.
29062 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29065 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29067 2017-03-03 Richard Biener <rguenther@suse.de>
29069 PR middle-end/79818
29070 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29071 TYPE_OVERFLOW_UNDEFINED check.
29073 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29075 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29077 (vector_ae_<mode>_p): Likewise.
29078 (vector_nez_<mode>_p): Likewise.
29079 (vector_ne_v2di_p): Likewise.
29080 (vector_ae_v2di_p): Likewise.
29081 (vector_ne_<mode>_p): Likewise.
29082 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29084 (vsx_tsqrt<mode>2_fe): Likewise.
29086 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29089 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29091 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29093 PR rtl-optimization/79780
29094 * cprop.c (one_cprop_pass): When second and further conditional trap
29095 in a single basic block is turned into an unconditional trap, turn it
29096 into a deleted note to avoid RTL verification failures.
29098 2017-03-02 Richard Biener <rguenther@suse.de>
29100 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29102 2017-03-02 Richard Biener <rguenther@suse.de>
29104 PR tree-optimization/79345
29106 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29107 param and abort the walk, returning -1 if it is hit.
29108 (walk_aliased_vdefs): Take a limit param and pass it on.
29109 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29110 defaulting to 0 and return a signed int.
29111 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29112 (check_defs): New helper.
29113 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29114 about uninitialized memory.
29115 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29116 bogus uninitialized warning.
29117 (fixed_convert_from_real): Likewise.
29119 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29121 PR tree-optimization/66768
29122 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29123 iv_use if base object can't be determined.
29125 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29127 PR tree-optimization/79345
29128 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29129 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29130 (get_pattern_stats): Initialize it.
29131 * genemit.c (gen_expand): Verify match_scratch numbers come after
29132 match_operand/match_dup numbers.
29133 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29134 match_scratch numbers.
29135 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29137 * config/s390/s390.md (trunctdsd2): Likewise.
29139 2017-03-02 Richard Biener <rguenther@suse.de>
29141 * wide-int.h (wide_int_storage::operator=): Implement in terms
29144 2017-03-02 Richard Biener <rguenther@suse.de>
29146 PR tree-optimization/79777
29147 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29148 the to insert expression to sth existing.
29150 2017-03-01 Martin Sebor <msebor@redhat.com>
29152 PR middle-end/79692
29153 * gimple-ssa-sprintf.c
29154 (directive::known_width_and_precision): New function.
29155 (format_integer): Use it.
29156 (get_mpfr_format_length): Consider the full range of precision
29157 when computing %g output with the # flag. Set the likely byte
29158 count to 3 rather than 1 when precision is indeterminate.
29159 (format_floating): Correct the lower bound of precision.
29161 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29163 * doc/invoke.texi: Document default code model for 64-bit Linux.
29165 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29168 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29169 udiv rather than div since input pattern is unsigned.
29171 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29173 * config/i386/i386.c (print_reg): Warn for values of
29174 unsupported size in integer register.
29176 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29179 * config/rs6000/predicates.md (current_file_function_operand): Do
29180 not allow self calls to be local if the function is replaceable.
29182 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29185 * config/rs6000/altivec.h (vec_ctz and others): Change the
29186 preprocessor macro that controls conditional compilation from
29187 _ARCH_PWR9 to __POWER9_VECTOR__.
29188 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29189 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29190 control (instead of _ARCH_PWR9 control) so that template
29191 definition uses power9-specific function.
29192 (vec_any_eq): Likewise.
29193 (vec_all_ne): Change macro definition to use a power9-specific
29194 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29195 _ARCH_PWR9 control).
29196 (vec_any_eq) Likewise.
29197 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29198 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29199 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29200 support for xvcmpnedp instruction.
29201 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29202 macro expansion so that Power9 implementation of vec_all_ne does
29203 not use the AltiVec predicate framework.
29204 (VCMPNEH_P): Likewise.
29205 (VCMPNEW_P): Likewise.
29206 (VCMPNED_P): Likewise.
29207 (VCMPNEFP_P): Likewise.
29208 (VCMPNEDP_P): Likewise.
29209 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29210 implementation of vec_any_eq to not use AltiVec predicate
29212 (VCMPAEH_P): Likewise.
29213 (VCMPAEW_P): Likewise.
29214 (VCMPAED_P): Likewise.
29215 (VCMPAEFP_P): Likewise.
29216 (VCMPAEDP_P): Likewise.
29217 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29218 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29219 not use the AltiVec predicate framework.
29220 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29221 of vec_any_eq to not use AltiVec predicate framework.
29222 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29223 support for predefined __POWER9_VECTOR__ macro to indicate that
29224 Power9 instruction selection is enabled.
29225 (altivec_overloaded_builtins): Remove extraneous
29226 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29227 function argument types RS6000_BTI_bool_V16QI and
29228 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29229 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29230 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29231 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29232 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29233 Power9 for implementations of vec_cmpne. Change the signature for
29234 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29235 (representing vec_all_ne) to remove the previously described first
29236 argument of type RS6000_BTI_INTSI, as this was an artifact of
29237 reliance on the AltiVec predicate framework, which is no longer
29238 used in the implementation of these functions. Add
29239 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29240 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29241 since, unlike the AltiVec predicate framework implementation, we
29242 do not share function descriptors between vec_alle and vec_anyeq.
29243 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29244 set of modes that receive special treatment even when
29245 TARGET_P9_VECTOR is true. The special treatment emits code that
29246 does not depend on Power9 instructions.
29247 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29248 define_expand to not rely on AltiVec predicate framework.
29249 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29251 (vector_ne_v2di_p): Change this define_expand to not rely on
29252 AltiVec predicate framework.
29253 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29255 (vector_ne_<mode>_p): Change this define_expand to not rely on
29256 AltiVec predicate framework.
29257 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29259 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29260 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29261 define_insn pattern.
29262 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29263 define_insn pattern because the xvcmpne<VSs>. instruction is not
29265 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29266 instruction is not supported.
29268 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29270 * config/nvptx/nvptx.c: Include intl.h.
29272 2017-03-01 Martin Jambor <mjambor@suse.cz>
29275 * ipa-prop.h (ipa_bits): Removed field known.
29276 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29277 to pointers. Adjusted their comments to warn about their sharing.
29278 (ipcp_transformation_summary): Change bits to a vector of pointers.
29279 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29280 (ipa_get_ipa_bits_for_value): Declare.
29281 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29282 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29283 (ipa_bits_hash_table): Likewise.
29284 (ipa_vr_ggc_hash_traits): Likewise.
29285 (ipa_vr_hash_table): Likewise.
29286 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29287 being pointers and vr_known being removed.
29288 (ipa_set_jf_unknown): Likewise.
29289 (ipa_get_ipa_bits_for_value): New function.
29290 (ipa_set_jfunc_bits): Likewise.
29291 (ipa_get_value_range): New overloaded functions.
29292 (ipa_set_jfunc_vr): Likewise.
29293 (ipa_compute_jump_functions_for_edge): Use the above functions to
29294 construct bits and vr parts of jump functions.
29295 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29296 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29298 (ipcp_grow_transformations_if_necessary): Also allocate
29299 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29301 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29302 them. Fix too long lines.
29303 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29304 vr_known being removed.
29305 (ipa_read_jump_function): Use new setter functions to construct bits
29306 and vr parts of jump functions or set them to NULL.
29307 (write_ipcp_transformation_info): Adjust for bits being pointers.
29308 (read_ipcp_transformation_info): Likewise.
29309 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29311 Include gt-ipa-prop.h.
29312 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29314 (ipcp_store_bits_results): Likewise.
29315 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29316 Do not write to existing jump functions but use a temporary instead.
29318 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29321 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29322 attempt to use its first operand as BIT_FIELD_REF base.
29324 2017-03-01 Richard Biener <rguenther@suse.de>
29326 PR middle-end/79721
29327 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29328 interpolating formula in wrapping arithmetic.
29329 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29331 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29333 PR tree-optimization/79734
29334 * tree-vect-generic.c (expand_vector_condition): Optimize
29335 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29336 Handle VEC_COND_EXPR where comparison has different inner width from
29337 type's inner width.
29339 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29341 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29342 markup, and similar issues. Remove @opindex entries for things
29343 that aren't options. Add missing -mmpy-option entries.
29345 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29347 PR tree-optimization/79737
29348 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29349 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29350 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29351 instead of byte_size. Formatting fix.
29352 (shift_bytes_in_array_right): Formatting fix.
29354 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29357 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29358 condition on optimize for the leaf function test.
29360 2017-02-28 Martin Liska <mliska@suse.cz>
29363 * read-rtl-function.c (function_reader::handle_unknown_directive):
29364 Bail out when one uses -flto.
29366 2017-02-28 Martin Liska <mliska@suse.cz>
29368 * common.opt: Replace space with tabular for options of <number>
29370 * config/i386/i386.opt: Show <number> value for
29371 -mlarge-data-threshold.
29372 * opts.c (print_filtered_help): Do not display number in hexadecimal
29375 2017-02-28 Martin Liska <mliska@suse.cz>
29377 * common.opt: Fix --help=option -Q for options which are of
29380 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29382 * config/i386/i386.c (print_reg): Error out for values
29383 of 8-bit size in invalid integer register.
29385 2017-02-28 Martin Sebor <msebor@redhat.com>
29387 PR tree-optimization/79691
29388 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29390 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29393 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29394 gcc_unreachable with output_operand_lossage.
29396 2017-02-28 Richard Biener <rguenther@suse.de>
29398 PR tree-optimization/79740
29399 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29401 (visit_nary_op): Insert the nary into the hashtable if we
29402 pattern-matched sth.
29403 * tree-ssa-pre.c (eliminate_insert): Robustify.
29405 2017-02-28 Richard Biener <rguenther@suse.de>
29407 PR middle-end/79731
29408 * fold-const.c (decode_field_reference): Reject out-of-bound
29411 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29413 * config/i386/i386.c: Include intl.h.
29414 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29415 instead of just cond ? "..." : "...".
29416 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29417 * coverage.c (read_counts_file): Likewise.
29418 * omp-offload.c: Include intl.h.
29419 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29420 of just cond ? "..." : "...".
29421 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29422 of just cond ? "..." : "...".
29424 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29427 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29429 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29430 'tune for' CPU name.
29431 * config/arm/arm-cpu-data.h: Regenerated.
29433 2017-02-28 Richard Biener <rguenther@suse.de>
29435 PR tree-optimization/79732
29436 * tree-inline.c (expand_call_inline): Do not shadow var.
29438 2017-02-28 Richard Biener <rguenther@suse.de>
29440 PR tree-optimization/79723
29441 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29442 address-space properly.
29444 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29446 * doc/optinfo.texi (Optimization groups): Fix option used for
29448 * doc/invoke.texi (-fopt-info): Document "omp".
29449 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29450 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29451 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29452 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29453 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29455 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29457 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29458 OPTGROUP_OMP with "omp".
29460 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29463 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29464 for arithmetic shift of unsigned V2DI.
29466 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29468 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29469 arc/linux.h headers.
29470 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29471 (LINK_SPEC): Likewise.
29472 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29473 (EXTRA_SPECS): Likewise.
29474 (STARTFILE_SPEC): Likewise.
29475 (ENDFILE_SPEC): Likewise.
29476 (LIB_SPEC): Likewise.
29477 (TARGET_SDATA_DEFAULT): Likewise.
29478 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29479 (MULTILIB_DEFAULTS): Likewise.
29480 (DWARF2_UNWIND_INFO): Likewise.
29481 * config/arc/big.h: New file.
29482 * config/arc/elf.h: Likewise.
29483 * config/arc/linux.h: Likewise.
29484 * config/arc/t-uClibc: Remove.
29486 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29488 PR tree-optimization/77536
29489 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29490 (tree_transform_and_unroll_loop): Use above function to compute the
29491 estimated niter of unrolled loop and use it when scaling profile.
29492 Also use count info rather than frequency if it's non-zero.
29493 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29494 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29495 (vect_transform_loop): Call above function.
29497 2017-02-27 Richard Biener <rguenther@suse.de>
29499 PR tree-optimization/45397
29500 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29501 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29502 (visit_nary_op): Add pattern matching for CSEing sign-changed
29503 or truncated operations with wider ones.
29505 2017-02-27 Richard Biener <rguenther@suse.de>
29507 PR tree-optimization/79690
29508 * tree-vect-stmts.c (vectorizable_store): Use vector type
29509 built from the DR with address-space.
29511 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29513 * doc/invoke.texi (Optimize Options): Refine the description
29514 of asan-use-after-return.
29516 2017-02-25 Alan Modra <amodra@gmail.com>
29518 PR rtl-optimization/79584
29519 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29520 base, not ad->base_term, the reg within base. Remove assertion
29521 that ad->base == ad->base_term. Replace gen_int_mode using
29522 bogus mode with const0_rtx.
29524 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29526 PR middle-end/79396
29527 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29528 FMA_EXPR like tcc_binary or tcc_unary.
29530 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29533 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29535 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29536 (output_loc_operands): Handle DW_OP_call_ref and
29537 DW_OP_GNU_variable_value.
29538 (struct variable_value_struct): New type.
29539 (struct variable_value_hasher): Likewise.
29540 (variable_value_hash): New variable.
29541 (string_types): Remove.
29542 (copy_loc_descr): New function.
29543 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29544 (prepend_loc_descr_to_each): New function.
29545 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29546 instead of add_loc_descr_to_each if the first argument is single
29547 location list and the second has multiple.
29548 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29549 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29550 when looking for variable value which doesn't have other location info.
29551 (loc_list_from_tree): Formatting fix.
29552 (gen_array_type_die): Simplify DW_AT_string_length handling.
29553 (adjust_string_types): Remove.
29554 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29555 string_types. Call resolve_variable_values.
29556 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29557 (resolve_addr_in_expr): Likewise. Add A argument.
29558 (copy_deref_exprloc): Remove deref argument. Adjust for the
29559 original expression being DW_OP_GNU_variable_value with optionally
29560 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29561 optionally after it.
29562 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29563 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29564 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29565 (variable_value_hasher::hash, variable_value_hasher::equal): New
29567 (resolve_variable_value_in_expr, resolve_variable_value,
29568 resolve_variable_values, note_variable_value_in_expr,
29569 note_variable_value): New functions.
29570 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29573 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29576 * opts.h (handle_generated_option): Add GENERATED_P argument.
29577 * opts-common.c (handle_option): Adjust function comment.
29578 (handle_generated_option): Add GENERATED_P argument, pass it to
29580 (control_warning_option): Pass false to handle_generated_option
29582 * opts.c (maybe_default_option): Pass true to handle_generated_option
29584 * optc-gen.awk: Likewise.
29586 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29588 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29589 a REG, look at the REG it is a SUBREG of.
29590 (splitter for cmpeqsi_t): Ditto.
29592 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29594 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29595 the special USEs with the pattern of the insn, not the insn itself.
29597 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29600 * doc/invoke.texi: Document -mload-store-pairs.
29602 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29603 Sandra Loosemore <sandra@codesourcery.com>
29605 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29606 argument isn't a CONST_INT.
29607 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29608 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29609 (nios2_validate_compare): Bypass alternate compare logic if *op2
29610 is not a CONST_INT.
29611 (ldstwm_operation_p): Return false if first_base is not a REG or
29612 if first_offset is not a CONST_INT.
29614 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29616 * config/cris/cris.md: Use correct operand in a define_peephole2.
29618 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29620 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29622 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29624 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29625 this_insn if it is an INSN or JUMP_INSN.
29626 (force_offsettable): Look at base, not at addr.
29627 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29628 on things that aren't necessarily CONST_INTs.
29630 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29632 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29633 -mfpmath=sse is the default also for x86-32 targets with SSE2
29634 instruction set when @option{-ffast-math} is enabled
29636 2017-02-24 Jeff Law <law@redhat.com>
29638 PR rtl-optimizatoin/79286
29639 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29642 2017-02-24 Richard Biener <rguenther@suse.de>
29644 PR tree-optimization/79389
29645 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29648 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29650 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29651 function comment to reflect reality.
29652 (loop_exits_before_overflow): Fix typo in function description.
29654 2017-02-24 Richard Biener <rguenther@suse.de>
29656 PR tree-optimization/79389
29657 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29658 properly that a threading opportunity exists. Detect conditional
29659 copy/constant propagation opportunities.
29661 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29663 * config/visium/visium.md (type): Add trap.
29664 (b): New mode attribute.
29665 (*btst): Rename into...
29666 (*btst<mode>): ...this and adjust.
29667 (*cbranchsi4_btst_insn): Rename into...
29668 (*cbranch<mode>4_btst_insn): ...this and adjust.
29669 (trap): New define_insn.
29671 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29673 PR tree-optimization/79389
29674 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29675 (noce_reversed_cond_code): New function.
29676 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29677 reversed_comparison_code. Formatting fix.
29678 (noce_try_store_flag): Test rev_cond != NULL in addition to
29679 reversed_comparison_code.
29680 (noce_try_store_flag_constants): Likewise.
29681 (noce_try_store_flag_mask): Likewise.
29682 (noce_try_addcc): Use rev_cond if non-NULL instead of
29683 reversed_comparison_code.
29684 (noce_try_cmove_arith): Likewise. Formatting fixes.
29685 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29686 (noce_find_if_block): Initialize rev_cond.
29687 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29688 instead of false as last argument never attempt to reverse it
29691 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29693 PR tree-optimization/79663
29694 * tree-predcom.c (combine_chains): Process refs in reverse order
29695 only for ZERO length chains, and add explaining comment.
29697 2017-02-23 Jeff Law <law@redhat.com>
29699 PR tree-optimization/79578
29700 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29701 in call to operand_equal_p.
29703 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29706 * config/i386/cpuid.h: Fix another undefined behavior.
29708 2017-02-23 Richard Biener <rguenther@suse.de>
29710 PR tree-optimization/79683
29711 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29712 vector types for data-refs.
29714 2017-02-23 Martin Liska <mliska@suse.cz>
29716 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29718 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29720 PR middle-end/79665
29721 * internal-fn.c (get_range_pos_neg): Moved to ...
29722 * tree.c (get_range_pos_neg): ... here. No longer static.
29723 * tree.h (get_range_pos_neg): New prototype.
29724 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29725 are known to be in between 0 and signed maximum inclusive, try to
29726 expand both unsigned and signed divmod and use the cheaper one from
29729 2017-02-22 Jeff Law <law@redhat.com>
29731 PR tree-optimization/79578
29732 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29733 to compare base operands.
29735 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29738 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29739 gpc_reg_operand instead of fpr_reg_operand.
29741 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29743 * config/mips/mips.c (mips_return_in_memory): Force FP
29744 vector types to be returned in memory for o32 ABI.
29746 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29748 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29749 instead of DW_TAG_member for static data member declarations and don't
29750 set no_linkage_name for static inline data members.
29751 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29754 2017-02-22 Martin Liska <mliska@suse.cz>
29756 * doc/invoke.texi: Replace inequality signs with square brackets
29759 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29762 * lra-constraints.c (simplify_operand_subreg): Handle
29763 WORD_REGISTER_OPERATIONS targets.
29765 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29768 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29769 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29770 elimination by swapping fld*.
29772 2017-02-22 Richard Biener <rguenther@suse.de>
29774 PR tree-optimization/79673
29775 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29776 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29777 irrelevant address-space qualifiers and avoiding a
29778 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29780 2017-02-22 Richard Biener <rguenther@suse.de>
29782 PR tree-optimization/79666
29783 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29784 to not symbolically negate if that may introduce undefined
29787 2017-02-22 Martin Liska <mliska@suse.cz>
29790 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29791 * data-streamer-out.c (streamer_write_gcov_count_stream):
29793 * value-prof.c (stream_out_histogram_value): Make assert more
29794 precise based on type of counter.
29796 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29799 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29800 Use nonimmediate_operand instead of memory_operand for operand 1.
29801 (float-extend standard_x87sse_constant_load splitter): Ditto.
29803 2017-02-21 Jeff Law <law@redhat.com>
29805 PR tree-optimization/79621
29806 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29807 blocks with edges to themselves.
29809 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29812 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29813 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29814 Use gimple_call_builtin_p.
29817 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29818 on temporarily removed DEBUG_INSNs.
29820 PR tree-optimization/79649
29821 * tree-loop-distribution.c (classify_partition): Give up on
29822 non-generic address space loads/stores.
29824 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29826 * doc/loop.texi (Loop manipulation): Remove nonexistent
29827 tree_ssa_loop_version from the documentation.
29828 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29830 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29833 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29834 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29835 * config/rs6000/rs6000.c: Include except.h.
29836 (rs6000_expand_split_stack_prologue): Call
29837 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29839 2017-02-21 Martin Jambor <mjambor@suse.cz>
29842 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29843 have been analyzed.
29845 2017-02-21 Martin Jambor <mjambor@suse.cz>
29847 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29848 for backward compatibility only.
29849 * doc/invoke.texi (Option Summary): Remove all references to
29850 -fipa-cp-alignment.
29852 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29856 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29858 * lra-constraints.c (curr_insn_transform): Handle
29859 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29861 2017-02-21 Martin Liska <mliska@suse.cz>
29863 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29865 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29867 PR translation/79638
29868 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29870 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29873 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29874 (arm_function_ok_for_sibcall): Return false for an indirect call by
29875 descriptor if all the argument registers are used.
29876 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29877 alignment of the function.
29879 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29881 PR tree-optimization/61441
29882 * simplify-rtx.c (simplify_const_unary_operation): For
29883 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29884 the sNaN unmodified.
29886 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29888 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29889 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29890 instead of SYSTEM_HEADER_DIR.
29892 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29893 Martin Liška <mliska@suse.cz>
29895 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29896 Fix typos and grammar, use active voice, and clarify.
29898 2017-02-20 Marek Polacek <polacek@redhat.com>
29900 PR middle-end/79537
29901 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29904 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29906 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29909 * config/i386/i386.c (ix86_expand_builtin): Handle
29910 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29911 ix86_builtins_isa[fcode].isa as a requirement of those
29912 flags and any other flag in the bitmask.
29913 (ix86_init_mmx_sse_builtins): Use 0 instead of
29914 ~OPTION_MASK_ISA_64BIT as mask.
29915 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29916 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29917 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29918 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29920 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29923 * lra-constraints.c (split_reg): Check requested split mode
29924 is supported by the register.
29926 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29928 * lra-constraints.c (simplify_operand_subreg): Remove early
29931 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29934 * lra-constraints.c (curr_insn_transform): Tighten condition
29935 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29937 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29940 * lra-constraints.c (curr_insn_transform): Handle
29941 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29943 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29946 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29948 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29950 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29953 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29956 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29958 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29961 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29963 * final.c (last_columnnum, override_columnnum): New variables.
29964 (final_start_function): Set last_columnnum, pass it to begin_prologue
29965 hook and pass 0 to dwarf2out_begin_prologue.
29966 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29967 to source_line debug hook.
29968 (notice_source_line): Compute last_columnnum and for debug_column_info
29969 return true on column changes.
29970 * debug.h (struct gcc_debug_hooks): Add column argument to
29971 source_line and begin_prologue hooks.
29972 (debug_nothing_int_charstar_int_bool): Remove prototype.
29973 (debug_nothing_int_int_charstar,
29974 debug_nothing_int_int_charstar_int_bool): New prototypes.
29975 (dwarf2out_begin_prologue): Add column argument.
29976 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29977 begin_prologue hooks.
29978 (debug_nothing_int_charstar_int_bool): Remove.
29979 (debug_nothing_int_int_charstar,
29980 debug_nothing_int_int_charstar_int_bool): New functions.
29981 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29982 through to dwarf2out_source_line.
29983 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29984 (dwarf2out_source_line): Add column argument, emit it if requested.
29985 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29987 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29988 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29989 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29990 through to dwarf2out_begin_prologue.
29991 (vmsdbgout_source_line): Add column argument, pass it through to
29992 dwarf2out_source_line.
29993 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29994 dbxout_source_line caller.
29995 (dbxout_source_line): Add column argument.
29997 * common.opt (gno-column-info, gcolumn-info): New options.
29998 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29999 (check_die): Also test for multiple DW_AT_decl_column attributes.
30000 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30001 DW_AT_decl_column if requested.
30002 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30004 (gen_variable_die): Likewise.
30005 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30006 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30009 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30010 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30011 (ix86_handle_option): Handle OPT_m3dnowa.
30012 * doc/invoke.texi (-m3dnowa): Document.
30013 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30014 -m3dnowa instead of -m3dnow -march=athlon.
30017 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30018 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30020 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30023 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30024 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30025 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30026 generator for vsx_xxpermdi_<mode>_be.
30027 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30028 force big-endian semantics.
30029 (vsx_xxpermdi_<mode>_be): New define_expand with same
30030 implementation as previous version of vsx_xxpermdi_<mode>.
30032 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30034 PR tree-optimization/79327
30035 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30036 variable, its initialization and use.
30038 2017-02-17 Julia Koval <julia.koval@intel.com>
30040 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30041 (OPTION_MASK_ISA_PKU_UNSET): New.
30042 (ix86_handle_option): Handle -mrdpid.
30043 * config/i386/cpuid.h (bit_RDPID): New.
30044 * config/i386/driver-i386.c (host_detect_local_cpu):
30045 Detect RDPID feature.
30046 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30047 * config/i386/i386-c.c (ix86_target_macros_internal):
30049 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30050 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30051 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30052 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30053 * config/i386/i386.md (define_insn "rdpid"): New.
30054 * config/i386/i386.opt Add -mrdpid.
30055 * config/i386/immintrin.h (_rdpid_u32): New.
30057 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30059 PR rtl-optimization/79541
30060 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30061 instead of transforming it into USE.
30063 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30065 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30066 If HONOR_SNANS (SFmode) force the input to a register.
30067 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30068 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30069 an frsp or similar insn.
30071 2017-02-17 Martin Liska <mliska@suse.cz>
30073 PR rtl-optimization/79577
30074 * params.def (selsched-max-sched-times): Increase minimum to 1.
30076 2017-02-17 Martin Liska <mliska@suse.cz>
30078 PR rtl-optimization/79574
30079 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30081 2017-02-17 Martin Liska <mliska@suse.cz>
30083 PR tree-optimization/79529
30084 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30085 ssa_defined_default_def_p to handle cases which are implicitly
30087 * tree-ssa.c (ssa_defined_default_def_p): New function.
30088 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30089 which are implicitly defined.
30090 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30092 2017-02-17 Richard Biener <rguenther@suse.de>
30094 PR middle-end/79576
30095 * params.def (max-ssa-name-query-depth): Limit to 10.
30097 2017-02-17 Richard Biener <rguenther@suse.de>
30099 PR tree-optimization/79552
30100 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30103 2017-02-17 Richard Biener <rguenther@suse.de>
30106 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30108 2017-02-17 Marek Polacek <polacek@redhat.com>
30110 PR middle-end/79536
30111 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30112 (fold_negate_expr): New wrapper.
30114 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30116 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30117 Correct terminology and de-emphasize pre-standard behavior.
30119 2017-02-16 Alan Modra <amodra@gmail.com>
30121 PR rtl-optimization/79286
30122 * ira.c (def_dominates_uses): New function.
30123 (update_equiv_regs): Don't create an equivalence for insns that
30124 may trap where the register def does not dominate the use.
30126 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30128 PR rtl-optimization/78127
30129 * lra.c (lra): Call lra_eliminate before finish the loop after
30132 2017-02-16 Richard Biener <rguenther@suse.de>
30134 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30136 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30137 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30138 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30139 (isl_val_int_from_wi): New function.
30140 (extract_affine_gmp): Rename to ...
30141 (extract_affine_wi): ... this, take a widest_int.
30142 (extract_affine_int): Just wrap extract_affine_wi.
30143 (add_param_constraints): Use isl_val_int_from_wi.
30144 (add_loop_constraints): Likewise, and extract_affine_wi.
30146 2017-02-15 Jeff Law <law@redhat.com>
30148 PR middle-end/79521
30149 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30150 ira_init_register_move_cost_if_necessary.
30152 2017-02-15 Martin Sebor <msebor@redhat.com>
30154 PR middle-end/32003
30155 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30156 removed in a prior commit.
30158 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30160 PR tree-optimization/79347
30161 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30162 counters during peeling.
30164 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30166 * Makefile.in (site.exp): Remove "set ISLVER".
30168 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30171 * real.c (real_from_integer): Call real_convert even for decimal.
30173 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30176 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30178 2017-02-14 Andrew Pinski <apinski@cavium.com>
30180 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30181 cores and change the partno/implementer to be correct.
30182 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30183 the 'B" as the implementer.
30184 * config/aarch64/aarch64-tune.md: Regenerate.
30186 2017-02-14 Carl Love <cel@us.ibm.com>
30188 * config/rs6000/rs6000.c: Add case statement entry to make the
30189 xvcvuxdsp built-in argument unsigned.
30190 * config/rs6000/vsx.md: Fix the source and return operand types so they
30191 match the instruction definitions from the ISA document. Fix typo
30192 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30195 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30198 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30199 member early_clobber_alts.
30200 * lra-lives.c (reg_early_clobber_p): New.
30201 (process_bb_lives): Use it.
30202 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30203 (debug_operand_data): Initialize early_clobber_alts.
30204 (setup_operand_alternative): Set up early_clobber_alts.
30205 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30206 alternatives to new_insn_reg.
30207 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30209 (lra_update_insn_regno_info): Pass the new arg.
30211 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30213 PR middle-end/79505
30214 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30215 (new_oacc_loop_raw): Don't clear already cleared fields.
30218 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30219 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30220 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30222 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30225 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30227 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30230 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30231 the extra instruction to the right place to store 128-bit constant
30234 2017-02-14 Martin Sebor <msebor@redhat.com>
30236 PR middle-end/79448
30237 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30238 warning for strings of unknown length.
30240 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30242 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30244 2017-02-14 Jeff Law <law@redhat.com>
30247 * ira-costs.c (scan_one_insn): Initialize register move costs
30248 for pseudos seen in USE/CLOBBER insns.
30250 PR tree-optimization/79095
30251 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30252 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30253 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30254 if the operands are known to be not equal, then the resulting range
30256 (intersect_ranges): If the new range is ~[0,0] and the old range is
30257 wide, then prefer ~[0,0].
30258 * tree-vrp.c (overflow_comparison_p_1): New function.
30259 (overflow_comparison_p): New function.
30260 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30261 if NAME is used in an overflow test.
30262 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30263 overflow check that can be expressed as an equality test, then adjust
30264 ops to be that equality test.
30266 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30268 * config/s390/s390-builtin-types.def: Remove flags argument.
30269 * config/s390/s390.c (s390_init_builtins): Likewise.
30271 2017-02-14 Martin Liska <mliska@suse.cz>
30273 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30274 vector. Fix trailing white spaces.
30276 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30278 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30281 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30283 PR rtl-optimization/68664
30284 * config/arm/arm.c (arm_sched_can_speculate_insn):
30285 New function. Declare prototype.
30286 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30288 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30290 PR rtl-optimization/68664
30291 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30293 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30295 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30297 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30298 max skip bytes for function, loop and jump.
30300 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30302 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30303 ABS_EXPR for gimple dump.
30305 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30308 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30310 PR tree-optimization/79408
30311 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30312 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30313 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30314 also if rhs1 is INTEGER_CST.
30316 2017-02-14 Richard Biener <rguenther@suse.de>
30318 PR middle-end/79432
30319 * tree-into-ssa.c (insert_phi_nodes): When the function can
30320 have abnormal edges rewrite SSA names with broken use-def
30321 dominance out of SSA and register them for PHI insertion.
30323 2017-02-13 Martin Sebor <msebor@redhat.com>
30325 PR middle-end/79496
30326 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30327 clearing info.nowrite flag when snprintf size argument is a range.
30329 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30331 * cprop.c (cprop_jump): Add missing space in string literal.
30332 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30333 (get_constraint_for_component_ref): Likewise.
30334 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30335 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30336 * lra-constraints.c (process_alt_operands): Likewise.
30337 * ipa-inline.c (inline_small_functions): Likewise.
30338 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30339 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30340 * trans-mem.c (diagnose_tm_1_op): Likewise.
30341 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30342 (grid_parallel_clauses_gridifiable): Likewise.
30344 * config/nvptx/mkoffload.c (process): Add space in between
30347 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30348 "MOD4_SSE_REGS" and "ALL_REGS".
30350 * spellcheck.c (test_data): Add , in between "foo" and "food".
30352 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30355 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30356 boundary crossing check and subsequent code generation agree.
30358 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30360 * config/aarch64/aarch64.c (has_memory_op): Delete.
30361 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30364 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30366 PR rtl-optimization/79388
30367 PR rtl-optimization/79450
30368 * combine.c (distribute_notes): When removing TEM_INSN for which
30369 corresponding dest has last value recorded, invalidate that last
30372 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30374 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30375 of explicit '@'. Add missing assembly comment marker on branch costs
30378 2017-02-13 Nathan Sidwell <nathan@acm.org>
30380 * gengtype-lex.l (<in_struct>): Add '/'.
30382 2017-02-13 Martin Liska <mliska@suse.cz>
30385 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30387 2017-02-13 Richard Biener <rguenther@suse.de>
30389 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30391 * configure: Re-generate.
30392 * config.in: Likewise.
30393 * graphite-dependences.c: Simplify as if
30394 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30395 * graphite-isl-ast-to-gimple.c: Likewise.
30396 * graphite-optimize-isl.c: Likewise.
30397 * graphite-poly.c: Likewise.
30398 * graphite-sese-to-poly.c: Likewise.
30399 * graphite.h: Likewise.
30400 * toplev.c: Include isl/version.h and use isl_version () for
30401 printing the ISL version.
30402 * doc/install.texi: Update ISL requirement.
30404 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30406 * doc/standards.texi (Standards): Update reference to
30409 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30411 * doc/extend.texi (Named Address Spaces): sourceware.org now
30413 * doc/install.texi (Binaries): Ditto.
30416 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30418 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30419 standard terminology "stringize"/"stringizing" throughout.
30420 * doc/cppinternals.texi: Likewise.
30422 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30424 * doc/extend.texi: Fix some spelling mistakes and typos.
30425 * doc/invoke.texi: Likewise.
30427 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30430 * params.def (inline-min-speedup) Change from 10 to 8.
30432 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30434 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30437 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30440 * ipa-inline-analysis.c (get_minimal_bb): New function.
30441 (record_modified): Use it.
30442 (remap_edge_change_prob): Handle also ancestor functions.
30444 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30446 * doc/contrib.texi (Contributors): Remove broken link into
30447 the Mauve CVS repository.
30449 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30451 PR middle-end/79454
30452 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30453 result computation whenever lhs doesn't have vector mode, not
30454 just when it has BLKmode.
30456 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30458 * doc/makefile.texi (profiledbootstrap): Refer to the
30459 installation instructions only in textual form.
30461 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30464 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30466 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30468 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30469 (Specific): Update mingw-w64 reference.
30471 (Specific): Remove broken link to Renesas RX processor.
30473 2017-02-10 Richard Biener <rguenther@suse.de>
30475 * toplev.c (process_options): Do not mention obsolete graphite
30476 options when printing sorry message about missing graphite support.
30477 Mention -floop-nest-optimize.
30479 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30481 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30482 (vtst_p16): Likewise.
30483 (vtstq_p8): Likewise.
30484 (vtstq_p16): Likewise.
30486 (vtstq_p64): Likewise.
30487 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30488 (vset_lane_p64): New.
30489 (vsetq_lane_p64): New.
30491 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30493 PR tree-optimization/79411
30494 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30495 stmt operands are SSA_NAMEs used in abnormal phis.
30496 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30499 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30502 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30505 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30507 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30509 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30511 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30515 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30516 not arbitrary TREE_CONSTANT.
30519 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30520 "omp declare target link" attribute unless is_global_var.
30521 * omp-offload.c (find_link_var_op): Likewise.
30523 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30524 Chung-Lin Tang <cltang@codesourcery.com>
30526 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30528 (gimplify_adjust_omp_clauses): Don't delete TILE.
30529 (gimplify_omp_for): Deal with TILE.
30530 * internal-fn.c (expand_GOACC_TILE): New function.
30531 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30533 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30534 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30536 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30537 avoid DIV for outermost collapse var.
30538 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30539 Remove out of date comments, fix whitespace.
30540 * omp-general.c (omp_extract_for_data): Deal with tiling.
30541 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30542 adjust OLF_DIM_BASE value.
30543 (struct omp_for_data): Add tiling field.
30544 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30545 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30546 for auto loops. Remove default auto determining, moved to
30547 oacc_loop_fixed_partitions.
30548 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30549 stmts, add e_mask field.
30550 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30551 (oacc_thread_numbers): Use oacc_dim_call.
30552 (oacc_xform_tile): New.
30553 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30554 (finish_oacc_loop): Adjust for ifns vector.
30555 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30556 add case for GOACC_TILE fns.
30557 (oacc_loop_xform_loop): Delete.
30558 (oacc_loop_process): Iterate over call list directly, and add
30559 handling for GOACC_TILE fns.
30560 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30562 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30563 vector partitioning to outer loops. Assign 2 partitions to loops
30564 when available. Add TILE handling.
30565 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30566 (execite_oacc_device_lower): Process GOACC_TILE fns,
30567 ignore unknown specs.
30568 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30569 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30570 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30572 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30574 * configure.ac (ACX_BUGURL): Update.
30575 * configure: Regenerate.
30577 2017-02-09 Richard Biener <rguenther@suse.de>
30579 PR tree-optimization/69823
30580 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30581 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30583 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30585 * config/arc/arc-c.def: Add __NPS400__ definition.
30586 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30587 (TARGET_NPS400): Define.
30589 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30591 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30593 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30594 pointer, arch_info.
30595 (arc_cpu_types): Fill the arch_info field with a pointer into the
30596 arc_arch_types table.
30597 (arc_selected_cpu): Declare.
30598 * config/arc/arc.c (arc_selected_cpu): Make global.
30599 (arc_selected_arch): Delete.
30600 (arc_base_cpu): Delete.
30601 (arc_override_options): Remove references to deleted variables,
30602 update access to arch information.
30603 (ARC_OPT): Update access to arch information.
30604 (ARC_OPTX): Likewise.
30605 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30606 (TARGET_ARC600): Update access to arch information.
30607 (TARGET_ARC601): Likewise.
30608 (TARGET_ARC700): Likewise.
30609 (TARGET_EM): Likewise.
30610 (TARGET_HS): Likewise.
30611 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30614 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30617 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30618 condition/operands for integer GE/LE/GEU/LEU operations.
30620 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30622 PR translation/79397
30623 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30626 2017-02-08 Martin Jambor <mjambor@suse.cz>
30629 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30630 whether allocation happened.
30631 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30632 nothing was allocated.
30634 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30636 PR tree-optimization/79408
30637 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30638 constant, but SSA_NAME with a known integer range, use the minimum
30639 of that range instead of op1 to determine if modulo can be replaced
30640 with its first operand.
30642 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30644 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30646 2017-02-08 Richard Biener <rguenther@suse.de>
30648 PR tree-optimization/71824
30649 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30650 Check all loops contained in the merged region.
30652 2017-02-07 Andrew Pinski <apinski@cavium.com>
30654 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30656 2017-02-07 Andrew Pinski <apinski@cavium.com>
30658 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30659 (thunderxt88): Likewise.
30660 (thunderxt81): Disable LSE and change v8.1 to v8.
30661 (thunderxt83): Likewise.
30663 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30664 Richard Biener <rguenther@suse.de>
30666 PR middle-end/79399
30667 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30668 type from int to size_t.
30669 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30671 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30673 PR rtl-optimization/79386
30674 * cprop.c (bypass_conditional_jumps): Initialize
30675 bypass_last_basic_block already before splitting bbs after
30676 unconditional traps...
30677 (bypass_conditional_jumps): ... rather than here.
30680 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30681 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30682 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30683 fix -masm=intel patterns.
30685 2017-02-07 Richard Biener <rguenther@suse.de>
30687 PR tree-optimization/79256
30688 PR middle-end/79278
30689 * builtins.c (get_object_alignment_2): Use min_align_of_type
30690 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30691 and ADJUST_FIELD_ALIGN.
30693 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30695 * doc/tm.texi: Regenerate.
30696 * stor-layout.c (layout_decl): Adjust.
30697 (update_alignment_for_field): Likewise.
30698 (place_field): Likewise.
30699 (min_align_of_type): Likewise.
30700 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30701 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30702 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30703 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30704 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30705 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30706 * config/i386/i386.c (x86_field_alignment): Likewise.
30707 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30708 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30709 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30710 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30711 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30712 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30716 2017-01-30 Richard Biener <rguenther@suse.de>
30718 PR tree-optimization/79256
30719 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30720 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30723 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30725 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30726 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30727 builtins to SImode and emit a zero-extend, if necessary.
30729 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30731 * docs/invoke.texi (RISC-V Options): Alphabetize.
30733 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30735 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30738 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30740 * config/riscv/riscv.c: New file.
30741 * common/config/riscv/riscv-common.c: Likewise.
30742 * config.gcc: Likewise.
30743 * config/riscv/constraints.md: Likewise.
30744 * config/riscv/elf.h: Likewise.
30745 * config/riscv/generic.md: Likewise.
30746 * config/riscv/linux.h: Likewise.
30747 * config/riscv/multilib-generator: Likewise.
30748 * config/riscv/peephole.md: Likewise.
30749 * config/riscv/pic.md: Likewise.
30750 * config/riscv/predicates.md: Likewise.
30751 * config/riscv/riscv-builtins.c: Likewise.
30752 * config/riscv/riscv-c.c: Likewise.
30753 * config/riscv/riscv-ftypes.def: Likewise.
30754 * config/riscv/riscv-modes.def: Likewise.
30755 * config/riscv/riscv-opts.h: Likewise.
30756 * config/riscv/riscv-protos.h: Likewise.
30757 * config/riscv/riscv.h: Likewise.
30758 * config/riscv/riscv.md: Likewise.
30759 * config/riscv/riscv.opt: Likewise.
30760 * config/riscv/sync.md: Likewise.
30761 * config/riscv/t-elf-multilib: Likewise.
30762 * config/riscv/t-linux: Likewise.
30763 * config/riscv/t-linux-multilib: Likewise.
30764 * config/riscv/t-riscv: Likewise.
30765 * configure.ac: Likewise.
30766 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30767 Waterman as RISC-V maintainers.
30768 * doc/install.texi: Add RISC-V entries.
30769 * doc/invoke.texi: Add RISC-V options section.
30770 * doc/md.texi: Add RISC-V constraints section.
30771 * configure: Regenerated.
30773 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30776 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30777 false values to be constant vectors with all 0 or all 1 bits set.
30778 (vcondu<mode><mode>): Likewise.
30779 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30781 (fpmask_comparison_operator): Update comment.
30782 (vecint_comparison_operator): New predicate.
30783 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30784 vector conditionals when the true and false values are constant
30785 vectors with all 0 bits or all 1 bits set.
30787 2017-02-06 Martin Sebor <msebor@redhat.com>
30789 PR tree-optimization/79376
30790 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30792 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30794 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30795 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30796 to simplify split condition.
30798 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30800 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30801 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30804 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30806 PR rtl-optimization/68664
30807 * target.def (can_speculate_insn): New hook.
30808 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30809 * doc/tm.texi: Regenerate.
30810 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30811 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30812 (rs6000_sched_can_speculate_insn): New function.
30814 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30816 PR tree-optimization/79284
30817 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30818 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30819 vectorizable_mask_load_store, vectorizable_operation,
30820 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30821 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30822 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30823 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30824 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30825 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30826 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30827 is_gimple_assign (stmt). Replace another such test with
30828 is_gimple_assign (stmt).
30830 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30833 * config/avr/avr.c (rtl-iter.h): Include it.
30834 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30835 (avr_legitimate_combined_insn): ...and implementation.
30837 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30839 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30840 * config/s390/s390.c (s390_const_operand_ok)
30841 (s390_canonicalize_comparison, s390_extract_part)
30842 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30843 (s390_contiguous_bitmask_p, s390_rtx_costs)
30844 (legitimize_pic_address): Likewise.
30845 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30846 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30847 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30848 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30849 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30851 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30853 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30854 REGNO($0) == REGNO($1).
30856 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30858 * config/s390/linux.h(SIZE_TYPE): Add comment.
30860 2017-02-06 Julian Brown <julian@codesourcery.com>
30861 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30862 Virendra Pathak <virendra.pathak@broadcom.com>
30864 * config/aarch64/aarch64-cores.def: Change the scheduler
30866 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30867 * config/aarch64/thunderx2t99.md: New file.
30869 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30871 * doc/standards.texi (Go Language): Update link to language
30874 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30876 * tree-eh.c (lower_resx): Sanitize profile.
30877 (cleanup_empty_eh_move_lp): Likewise.
30879 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30882 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30884 * cfgloopmanip.h (loop_version): Update prototype.
30885 * modulo-sched.c (sms_schedule): Update call of loop_version.
30886 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30887 * tree-parloops.c (gen_parallel_loop): Likewise.
30888 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30889 * tree-ssa-loop-split.c (split_loop): Likewise.
30890 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30891 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30893 2017-02-05 Martin Liska <mliska@suse.cz>
30896 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30898 (print_operand_address): Initialize a struct to zero.
30900 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30902 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30903 garbage collector only in textual form.
30905 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30907 * doc/extend.texi (x86 specific memory model extensions for
30908 transactional memory): Simplify a phrase.
30910 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30913 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30914 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30915 (atomic_storedi_1): Likewise.
30917 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30919 PR tree-optimization/79338
30920 * tree-parloops.c (gather_scalar_reductions): Don't call
30921 vect_analyze_loop_form for loop->inner before destroying loop's
30924 2017-02-03 Martin Sebor <msebor@redhat.com>
30926 PR tree-optimization/79327
30927 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30928 when precision has resulted in leading zeros.
30929 (format_integer): Adjust the likely counter to assume an unknown
30930 argument that may be zero is non-zero.
30932 2017-02-03 Jason Merrill <jason@redhat.com>
30935 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30936 avoid copying non-taken branch.
30938 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30940 PR tree-optimization/79340
30941 * tree-vect-loop.c (vectorizable_reduction): Release
30942 vec_defs elements after safe_splicing them into other vectors.
30945 PR tree-optimization/79327
30946 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30947 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30949 (format_integer): Use wide_int_to_tree instead of build_int_cst
30950 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30951 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30952 of shortest and longest sequence.
30954 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30956 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30957 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30959 2017-02-03 Walter Lee <walt@tilera.com>
30962 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30963 after initial stackframe link reg save.
30964 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30966 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30969 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30970 wu for stxssp alternative.
30972 2017-02-03 Martin Sebor <msebor@redhat.com>
30974 PR tree-optimization/79352
30975 * gimple-fold.c (get_range_strlen): Add argument.
30976 (get_range_strlen): Change return type to bool.
30977 (get_maxval_strlen): Pass in a dummy argument.
30978 * gimple-fold.h (get_range_strlen): Change return type to bool.
30979 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30980 * tree.h (array_at_struct_end_p): Add argument.
30981 * tree.c (array_at_struct_end_p): Handle it.
30983 2017-02-03 Martin Liska <mliska@suse.cz>
30986 * multiple_target.c (create_dispatcher_calls): Redirect edge
30987 from a caller of a dispatcher.
30988 (expand_target_clones): Make the clones local.
30989 (ipa_target_clone): Do both target clones and resolvers.
30990 (ipa_dispatcher_calls): Remove the pass.
30991 (pass_dispatcher_calls::gate): Likewise.
30992 (make_pass_dispatcher_calls): Likewise.
30993 * passes.def (pass_target_clone): Put as very first IPA early
30996 2017-02-03 Martin Liska <mliska@suse.cz>
30998 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30999 in case of a function with ifunc attribute.
31001 2017-02-03 Martin Liska <mliska@suse.cz>
31003 * cgraph.c (cgraph_node::dump): Dump function version info.
31004 * symtab.c (symtab_node::dump_base): Add missing new line.
31006 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31008 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31009 (ifcombine_ifandif): Use it.
31011 2017-02-03 Martin Liska <mliska@suse.cz>
31013 * doc/invoke.texi: Document default value for
31014 use-after-scope-direct-emission-threshold.
31016 2017-02-03 Martin Liska <mliska@suse.cz>
31018 PR tree-optimization/79339
31019 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31020 (format_floating): Likewise.
31022 2017-02-03 Martin Liska <mliska@suse.cz>
31025 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31027 (ipa_node_params_t::remove): Likewise.
31028 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31029 initialization from removed ipa_node_params_t::insert.
31030 (ipa_node_params::~ipa_node_params): Move from removed
31031 ipa_node_params_t::release.
31032 * symbol-summary.h (symbol_summary::m_released): New member.
31033 Do not release a summary twice. Do not allow to call finalizer
31034 for types of a summary that live in GGC memory.
31036 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31038 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31041 2017-02-02 Martin Sebor <msebor@redhat.com>
31043 PR middle-end/79275
31044 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31045 (format_string): Tighten up the range of output for non-constant
31046 strings and correct the expected range for wide non-constant strings.
31048 2017-02-02 Martin Sebor <msebor@redhat.com>
31050 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31052 PR middle-end/32003
31053 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31055 (-fdump-tree-@var): Add to index and document how to come up
31056 with pass-specific option and dump file names.
31057 (-fdump-passes): Clarify where to look for output.
31059 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31061 PR middle-end/77445
31062 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31063 statistics of the analyzed path; allow threading for speed when
31064 any of BBs along the path are optimized for speed.
31066 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31068 PR middle-end/78468
31069 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31070 settings of the virtual registers.
31073 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31075 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31076 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31079 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31081 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31082 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31084 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31086 * config/s390/s390.md: Add missing comments with the expanded
31088 * config/s390/vector.md: Likewise.
31089 * config/s390/vx-builtins.md: Likewise.
31091 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31094 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31095 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31096 conditions on a single line.
31098 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31100 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31101 __S390_VX__ to __VX__.
31103 2017-02-01 Andrew Pinski <apinski@cavium.com>
31105 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31106 stmt_info to record_stmt_cost.
31107 (vect_get_known_peeling_cost): Pass stmt_info if known to
31109 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31110 cpu_vector_cost field into
31111 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31112 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31113 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31114 splitting of scalar_stmt_cost and vec_stmt_cost.
31115 (thunderx_vector_cost): Likewise.
31116 (cortexa57_vector_cost): LIkewise.
31117 (exynosm1_vector_cost): Likewise.
31118 (xgene1_vector_cost): Likewise.
31119 (thunderx2t99_vector_cost): Improve after the splitting of the two
31121 (aarch64_builtin_vectorization_cost): Update for the splitting of
31122 scalar_stmt_cost and vec_stmt_cost.
31124 2017-02-01 Torvald Riegel <triegel@redhat.com>
31125 Richard Henderson <rth@redhat.com>
31127 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31128 conditional on existance of a fast atomic load.
31129 * optabs-query.c (can_atomic_load_p): New function.
31130 * optabs-query.h (can_atomic_load_p): Declare it.
31131 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31132 no fast atomic load is available for the particular size of access.
31133 (expand_atomic_compare_and_swap): Likewise.
31134 (expand_atomic_load): Likewise.
31135 (expand_atomic_store): Likewise.
31136 (expand_atomic_fetch_op): Likewise.
31137 * testsuite/lib/target-supports.exp
31138 (check_effective_target_sync_int_128): Remove x86 because it provides
31139 no fast atomic load.
31140 (check_effective_target_sync_int_128_runtime): Likewise.
31142 2017-02-01 Richard Biener <rguenther@suse.de>
31144 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31145 (graphite_transform_loops): Provide opt-info for optimized nests.
31146 * tree-parloop.c (parallelize_loops): Provide opt-info for
31147 parallelized loops.
31149 2017-02-01 Richard Biener <rguenther@suse.de>
31151 PR middle-end/79315
31152 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31153 was not set before.
31155 2017-02-01 Richard Biener <rguenther@suse.de>
31157 PR tree-optimization/71824
31158 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31159 Verify the loops are valid in the merged SESE region.
31160 (scop_detection::can_represent_loop_1): Check analyzing the
31161 evolution of the number of iterations in the region succeeds.
31163 2017-01-31 Ian Lance Taylor <iant@golang.org>
31165 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31166 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31168 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31170 PR preprocessor/79210
31171 * input.c (get_substring_ranges_for_loc): Replace line_width
31172 assertion with error-handling.
31174 2017-01-31 Richard Biener <rguenther@suse.de>
31176 PR tree-optimization/77318
31177 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31178 (create_pw_aff_from_tree): Take loop parameter.
31179 (add_condition_to_pbb): Pass loop of the condition to
31180 create_pw_aff_from_tree.
31182 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31184 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31185 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31187 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31191 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31193 (convert_int_to_float128): Likewise.
31194 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31195 (convert_int_to_float128): Likewise.
31196 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31197 (UNSPEC_IEEE128_CONVERT): Likewise.
31198 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31199 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31200 Use local variables for IBM extended format.
31201 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31202 (fix_trunc<mode>si2_fprs): Likewise.
31203 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31204 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31205 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31206 to know that we can now have integers of all sizes in vector
31208 (fix<uns>_<mode>di2_hw): Likewise.
31209 (float<uns>_<mode>si2_hw): Likewise.
31210 (fix_<mode>si2_hw): Likewise.
31211 (fixuns_<mode>si2_hw): Likewise.
31212 (float<uns>_<mode>di2_hw): Likewise.
31213 (float_<mode>di2_hw): Likewise.
31214 (float_<mode>si2_hw): Likewise.
31215 (floatuns_<mode>di2_hw): Likewise.
31216 (floatuns_<mode>si2_hw): Likewise.
31217 (xscvqp<su>wz_<mode>): Delete, no longer used.
31218 (xscvqp<su>dz_<mode>): Likewise.
31219 (xscv<su>dqp_<mode>): Likewise.
31220 (ieee128_mfvsrd_64bit): Likewise.
31221 (ieee128_mfvsrd_32bit): Likewise.
31222 (ieee128_mfvsrwz): Likewise.
31223 (ieee128_mtvsrw): Likewise.
31224 (ieee128_mtvsrd_64bit): Likewise.
31225 (ieee128_mtvsrd_32bit): Likewise.
31227 2017-01-31 Martin Liska <mliska@suse.cz>
31230 * ipa-prop.c (ipa_free_all_node_params): Call release method
31231 instead of ~sumbol_summary to not to trigger double times
31234 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31236 PR tree-optimization/71691
31237 * bitmap.h (class auto_bitmap): New.
31238 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31239 is_maybe_undefined instead of ssa_undefined_value_p.
31241 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31243 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31244 __S390_ARCH_LEVEL__ to __ARCH__.
31246 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31248 PR tree-optimization/79267
31249 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31250 if should_remove_lhs_p is true.
31252 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31255 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31256 (add_alignment_attribute): New.
31257 (base_type_die): Add alignment attribute.
31258 (subrange_type_die): Likewise.
31259 (modified_type_die): Likewise.
31260 (gen_array_type_die): Likewise.
31261 (gen_descr_array_type_die: Likewise.
31262 (gen_enumeration_type_die): Likewise.
31263 (gen_subprogram_die): Likewise.
31264 (gen_variable_die): Likewise.
31265 (gen_field_die): Likewise.
31266 (gen_ptr_to_mbr_type_die): Likewise.
31267 (gen_struct_or_union_type_die): Likewise.
31268 (gen_subroutine_type_die): Likewise.
31269 (gen_typedef_die): Likewise.
31270 (base_type_cmp): Compare alignment attribute.
31272 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31275 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31276 (setb_unsigned) New pattern for setb with CCUNS.
31277 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31278 subfc./subfe sequence to avoid overflow problems. Generate a
31279 shorter sequence with cmpld/setb for power9.
31280 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31281 for generating subfc. instruction.
31282 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31283 now uses this instruction.
31285 2017-01-30 Ian Lance Taylor <iant@google.com>
31288 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31289 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31291 2017-01-30 Martin Sebor <msebor@redhat.com>
31293 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31294 Move constant to the right of a relational operator.
31295 (get_mpfr_format_length, format_character, format_string): Ditto.
31296 (should_warn_p, maybe_warn): Same.
31298 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31300 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31303 * asan.c (get_translation_unit_decl): Remove function.
31304 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31306 2017-01-30 Martin Liska <mliska@suse.cz>
31308 PR gcov-profile/79259
31309 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31310 -fprofile-generate.
31312 2017-01-30 Martin Liska <mliska@suse.cz>
31315 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31316 Initialize variables with NULL value.
31318 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31321 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31323 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31325 2017-01-30 Richard Biener <rguenther@suse.de>
31327 PR tree-optimization/79276
31328 * tree-vrp.c (process_assert_insertions): Properly adjust common
31329 when removing a duplicate.
31331 2017-01-30 Richard Biener <rguenther@suse.de>
31333 PR tree-optimization/79256
31334 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31335 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31337 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31339 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31342 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31343 ("*r<noxa>sbg_<mode>_sll_bitmask")
31344 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31345 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31346 Use contiguous_bitmask_nowrap_operand.
31348 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31351 * config/rs6000/altivec.h (vec_xl): Revise #define.
31352 (vec_xst): Likewise.
31354 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31356 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31358 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31360 PR rtl-optimization/79194
31361 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31362 traps before call to bypass_conditional_jumps.
31364 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31366 PR tree-optimization/71374
31367 * lra-constraints.c (check_conflict_input_operands): New.
31368 (match_reload): Use it.
31370 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31373 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31374 account to calculate conflict_set.
31376 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31378 PR rtl-optimization/78559
31379 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31380 other_insn in combine.
31382 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31384 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31385 uint16_type_node for BT_UINT16.
31387 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31389 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31390 "RTL Tests" to menu.
31391 (GIMPLE Tests): New node.
31392 (RTL Tests): New node.
31394 2017-01-27 Richard Biener <rguenther@suse.de>
31396 PR tree-optimization/79245
31397 * tree-loop-distribution.c (distribute_loop): Apply cost
31398 modeling also to detected patterns.
31400 2017-01-27 Richard Biener <rguenther@suse.de>
31402 PR tree-optimization/71433
31403 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31404 (compare_assert_loc): New function.
31405 (process_assert_insertions): Sort and optimize assert locations
31406 to remove duplicates and push down identical assertions on
31407 edges to their destination block.
31409 2017-01-27 Richard Biener <rguenther@suse.de>
31411 PR tree-optimization/79244
31412 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31413 out SSA names even if abnormal.
31415 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31417 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31418 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31419 instead of MPFR_RNDN.
31421 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31424 * arm.c (arm_option_override): Don't call build_target_option_node
31425 until after doing all option overrides.
31426 (arm_valid_target_attribute_tree): Likewise.
31428 2017-01-27 Martin Liska <mliska@suse.cz>
31430 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31431 for {cd}tors and C++ {cd}tors.
31433 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31435 * config/s390/s390.md ("*setmem_long_and")
31436 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31438 2017-01-26 Martin Sebor <msebor@redhat.com>
31440 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31443 2017-01-26 Martin Sebor <msebor@redhat.com>
31445 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31446 HAVE_DFmode before using XFmode or DFmode.
31447 (parse_directive): Avoid using the z length modifier to avoid
31448 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31450 PR middle-end/78703
31451 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31452 to accept adjustment as an array.
31453 (get_int_range): New function.
31454 (struct directive): Make width and prec arrays.
31455 (directive::set_width, directive::set_precision): Call get_int_range.
31456 (format_integer, format_floating): Handle width and precision ranges.
31457 (format_string, parse_directive): Same.
31459 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31462 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31463 comdat_type_p set, just clone them, but keep the children in the
31467 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31468 which have direct callers with -fvar-tracking-assignments enabled
31470 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31471 inside of type units.
31473 2017-01-26 Martin Sebor <msebor@redhat.com>
31475 PR middle-end/78703
31476 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31478 (struct format_result): Replace number_chars, number_chars_min,
31479 and number_chars_max with a single member of struct result_range.
31481 (format_result::operator+=): Adjust.
31482 (struct fmtresult): Remove bounded. Handle likely and unlikely
31484 (fmtresult::adjust_for_width_or_precision): New function.
31485 (fmtresult:type_max_digits): New function.
31486 (bytes_remaining): Handle likely and unlikely counters.
31487 (min_bytes_remaining): Remove.
31488 (format_percent): Simplify.
31489 (format_integer, format_floating): Set likely and unlikely counters.
31490 (get_string_length, format_character, format_string): Same.
31491 (format_plain, should_warn_p): New function.
31492 (maybe_warn): Call should_warn_p. Update diagnostic messages
31493 and handle those for all directives, including plain strings.
31494 (format_directive): Handle likely and unlikely counters.
31495 Remove unnecessary quoting from diagnostics. Add an informational
31497 (add_bytes): Remove.
31498 (pass_sprintf_length::compute_format_length): Simplify.
31499 (try_substitute_return_value): Handle likely and unlikely counters.
31501 2017-01-26 Carl Love <cel@us.ibm.com>
31503 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31504 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31506 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31509 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31510 endianess for subregs into account.
31511 * lra-constraints.c (lra_constraints): Do risky transformations
31512 always on the first iteration.
31513 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31514 last_call_used_reg_set.
31515 (process_bb_lives): Define and use last_call_used_reg_set.
31516 * lra.c (lra): Always continue after lra_constraints on the first
31519 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31521 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31523 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31525 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31527 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31528 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31529 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31530 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31531 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31532 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31533 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31534 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31535 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31537 2017-01-26 Marek Polacek <polacek@redhat.com>
31540 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31541 for the third operand.
31543 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31545 PR middle-end/79236
31546 * omp-low.c (struct omp_context): Add simt_stmt field.
31547 (scan_omp_for): Return omp_context *.
31548 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31549 context to the _simt_ SIMD stmt.
31550 (lower_omp_for): For combined SIMD with sibling _simt_
31551 SIMD, make sure to use the same decls in _looptemp_
31552 clauses as in the sibling.
31554 2017-01-26 David Sherwood <david.sherwood@arm.com>
31556 PR middle-end/79212
31557 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31560 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31563 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31564 emit fld b; fld a; if possible.
31566 * brig-builtins.def: Update copyright years.
31567 * config/arm/arm_acle_builtins.def: Update copyright years.
31569 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31572 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31573 constraint instead of o for the stxsd instruction.
31575 2017-01-25 Carl Love <cel@us.ibm.com>
31577 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31578 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31580 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31582 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31584 2017-01-25 Richard Biener <rguenther@suse.de>
31586 PR tree-optimization/69264
31587 * target.def (vector_alignment_reachable): Improve documentation.
31588 * doc/tm.texi: Regenerate.
31589 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31591 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31592 earlier changes with respect to TYPE_USER_ALIGN.
31593 (vector_alignment_reachable_p): Likewise. Improve dumping.
31595 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31598 * config/arm/arm.md (xordi3): Force constant operand into a register
31601 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31603 * doc/invoke.texi (-fstore-merging): Correct default optimization
31604 levels at which it is enabled.
31605 (-O): Move -fstore-merging from list to...
31608 2017-01-25 Richard Biener <rguenther@suse.de>
31611 * omp-expand.c: Include debug.h.
31612 (expand_omp_taskreg): Make sure to generate early debug before
31613 outlining anything from a function.
31614 (expand_omp_target): Likewise.
31615 (grid_expand_target_grid_body): Likewise.
31617 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31620 * asan.c (get_translation_unit_decl): New function.
31621 (asan_add_global): Extract modules file name from globals
31622 TRANSLATION_UNIT_DECL name.
31624 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31627 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31628 for long calls with APCS frame and VFP.
31630 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31632 * cfg.c (original_copy_tables_initialized_p): New function.
31633 * cfg.h (original_copy_tables_initialized_p): New decl.
31634 * cfgrtl.c (relink_block_chain): Guard the call to
31635 free_original_copy_tables with a call to
31636 original_copy_tables_initialized_p.
31637 * cgraph.h (symtab_node::native_rtl_p): New decl.
31638 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31639 (symtab_node::needed_p): Don't assert for early assembly output
31640 for __RTL functions.
31641 (cgraph_node::finalize_function): Set "force_output" for __RTL
31643 (cgraph_node::analyze): Bail out early for __RTL functions.
31644 (analyze_functions): Update assertion to support __RTL functions.
31645 (cgraph_node::expand): Bail out early for __RTL functions.
31646 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31648 * function.h (struct function): Update comment for field
31650 * gimple-expr.c: Include "tree-pass.h".
31651 (gimple_has_body_p): Return false for __RTL functions.
31652 * Makefile.in (OBJS): Add run-rtl-passes.o.
31653 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31655 (gcc::pass_manager::get_clean_slate): New accessor.
31656 * passes.c: Include "insn-addr.h".
31657 (should_skip_pass_p): Add logging. Update logic for running
31658 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31659 property-provider override so it is only done for gimple passes.
31661 (skip_pass): New function.
31662 (execute_one_pass): Call skip_pass when skipping passes.
31663 * read-md.c (md_reader::read_char): Support filtering
31664 the input to a subset of line numbers.
31665 (md_reader::md_reader): Initialize fields
31666 m_first_line and m_last_line.
31667 (md_reader::read_file_fragment): New function.
31668 * read-md.h (md_reader::read_file_fragment): New decl.
31669 (md_reader::m_first_line): New field.
31670 (md_reader::m_last_line): New field.
31671 * read-rtl-function.c (function_reader::create_function): Only
31672 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31673 curr_properties. Set DECL_INITIAL to a dummy block.
31674 (read_rtl_function_body_from_file_range): New function.
31675 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31677 * run-rtl-passes.c: New file.
31678 * run-rtl-passes.h: New file.
31680 2017-01-24 Jeff Law <law@redhat.com>
31682 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31685 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31687 PR tree-optimization/79159
31688 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31689 (record_nonwrapping_iv): Improve boundary using above function if no
31690 value range information.
31692 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31693 Martin Jambor <mjambor@suse.cz>
31695 * brig-builtins.def: New file.
31696 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31697 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31698 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31699 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31700 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31701 * builtin-types.def (BT_INT8): New.
31702 (BT_INT16): Likewise.
31703 (BT_UINT8): Likewise.
31704 (BT_UINT16): Likewise.
31705 (BT_FN_ULONG): Likewise.
31706 (BT_FN_UINT_INT): Likewise.
31707 (BT_FN_UINT_ULONG): Likewise.
31708 (BT_FN_UINT_LONG): Likewise.
31709 (BT_FN_UINT_PTR): Likewise.
31710 (BT_FN_ULONG_PTR): Likewise.
31711 (BT_FN_INT8_FLOAT): Likewise.
31712 (BT_FN_INT16_FLOAT): Likewise.
31713 (BT_FN_UINT32_FLOAT): Likewise.
31714 (BT_FN_UINT16_FLOAT): Likewise.
31715 (BT_FN_UINT8_FLOAT): Likewise.
31716 (BT_FN_UINT64_FLOAT): Likewise.
31717 (BT_FN_UINT16_UINT32): Likewise.
31718 (BT_FN_UINT32_UINT16): Likewise.
31719 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31720 (BT_FN_INT_PTR_INT): Likewise.
31721 (BT_FN_UINT_PTR_UINT): Likewise.
31722 (BT_FN_LONG_PTR_LONG): Likewise.
31723 (BT_FN_ULONG_PTR_ULONG): Likewise.
31724 (BT_FN_VOID_UINT64_UINT64): Likewise.
31725 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31726 (BT_FN_INT8_INT8_INT8): Likewise.
31727 (BT_FN_INT16_INT16_INT16): Likewise.
31728 (BT_FN_INT_INT_INT): Likewise.
31729 (BT_FN_UINT_FLOAT_UINT): Likewise.
31730 (BT_FN_FLOAT_UINT_UINT): Likewise.
31731 (BT_FN_ULONG_UINT_UINT): Likewise.
31732 (BT_FN_ULONG_UINT_PTR): Likewise.
31733 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31734 (BT_FN_UINT_UINT_UINT): Likewise.
31735 (BT_FN_VOID_UINT_PTR): Likewise.
31736 (BT_FN_UINT_UINT_PTR: Likewise.
31737 (BT_FN_UINT32_UINT64_PTR): Likewise.
31738 (BT_FN_INT_INT_UINT_UINT): Likewise.
31739 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31740 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31741 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31742 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31743 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31744 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31745 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31746 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31747 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31748 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31749 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31750 * doc/frontends.texi: List BRIG FE.
31751 * doc/install.texi (Testing): Add BRIG tesring requirements.
31752 * doc/invoke.texi (Overall Options): Mention BRIG.
31753 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31755 2017-01-24 Richard Biener <rguenther@suse.de>
31757 PR translation/79208
31758 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31760 2017-01-24 Martin Jambor <mjambor@suse.cz>
31763 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31764 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31765 and known_contexts.
31767 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31769 PR middle-end/79123
31770 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31771 casts from signed to unsigned really don't have a range.
31773 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31775 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31776 GMP_RNDx for compatiblity.
31778 2017-01-24 Martin Liska <mliska@suse.cz>
31781 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31782 that would prevent us to call alloca with -1 as argument.
31784 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31786 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31787 Avoid -Wformat-security warning.
31789 2017-01-23 Andrew Pinski <apinski@cavium.com>
31791 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31794 2017-01-23 Martin Sebor <msebor@redhat.com>
31796 PR middle-end/78703
31797 * gimple-ssa-sprintf.c (warn_level): New global.
31798 (format_integer): Use it here and throughout the rest of the file.
31799 Use the same switch to compute sign as base.
31800 (maybe_warn): New function.
31801 (format_directive): Factor out warnings into maybe_warn.
31802 Add debugging output. Use warn_level.
31803 (add_bytes): Use warn_level.
31804 (pass_sprintf_length::compute_format_length): Add debugging output.
31805 (try_substitute_return_value): Same.
31806 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31808 PR middle-end/78703
31809 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31810 (struct fmtresult, format_integer, format_floating): Adjust.
31811 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31812 (get_string_length, format_string,format_directive): Same.
31813 (pass_sprintf_length::compute_format_length): Same.
31814 (try_substitute_return_value): Simplify slightly.
31816 PR middle-end/78703
31817 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31818 (fmtresult::operator+=): Outlined.
31819 (struct fmtresult): Add ctors.
31820 (struct conversion_spec): Rename...
31821 (struct directive): ...to this. Add and remove data members.
31822 (directive::set_width, directive::set_precision): New functions.
31823 (format_percent): Use fmtresult ctor.
31824 (get_width_and_precision): Remove.
31825 (format_integer): Make naming changes. Avoid computing width and
31827 (format_floating): Same. Adjust indentation.
31828 (format_character, format_none): New functions.
31829 (format_string): Moved character handling to format_character.
31830 (format_directive): Remove arguments, change return type.
31831 (parse_directive): New function.
31832 (pass_sprintf_length::compute_format_length): Move directive
31833 parsing to parse_directive.
31835 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31837 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31838 (assign_assembler_name_if_needed): ... this.
31839 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31840 (assign_assembler_name_if_needed): ... this.
31841 (free_lang_data_in_cgraph): Adjust callers.
31842 * cgraphunit.c (cgraph_node::analyze): Likewise.
31843 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31846 2017-01-23 Richard Biener <rguenther@suse.de>
31848 PR tree-optimization/79088
31849 PR tree-optimization/79188
31850 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31851 resetting loop bounds after last path deletion. Reset loop
31852 bounds of the target loop, make code match the comments.
31853 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31854 Make sure loops need no fixups.
31856 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31858 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31859 exponent support with double type for first argument.
31860 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31861 type returned by __builtin_vec_extract_sig,
31862 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31863 functions from "vector int" to "vector unsigned int" or from
31864 "vector long long int" to "vector unsigned long long int".
31865 Changed type returned by __builtin_vec_extract_exp,
31866 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31867 functions from "vector int" to "vector unsigned int" or from
31868 "vector long long int" to "vector unsigned long long int".
31869 Changed return type of __builtin_vec_test_data_class,
31870 __builtin_vec_test_data_class_sp, and
31871 __builtin_vec_test_data_class_dp from "vector int" to
31872 "vector bool int" or from "vector long long int" to "vector bool
31873 long long int" and changed second argument type from "unsigned
31874 int" to "int". Added new overloaded function forms "vector float
31875 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31876 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31877 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31878 double, vector unsigned long long int)" and "vector double
31879 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31880 long int)". Changed return type of
31881 __builtin_scalar_test_data_class and
31882 __builtin_scalar_test_data_class_sp and
31883 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31884 int" and changed second argument from "unsigned int" to "int".
31885 Changed type returned by __builtin_scalar_test_neg,
31886 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31887 from "int" to "bool int". Added new overloaded function form
31888 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31889 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31890 exponent double-precision with floating point first argument.
31891 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31892 documentation of scalar_test_data_class, scalar_test_neg,
31893 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31894 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31895 vec_test_data_class built-in functions to reflect refinements in
31896 their type signatures.
31898 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31900 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31902 (aarch64_elf_asm_destructor): Likewise.
31904 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31906 PR rtl-optimization/78634
31907 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31908 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31909 * ifcvt.c (noce_try_cmove): Add missing cost check.
31911 PR rtl-optimization/71724
31912 * combine.c (if_then_else_cond): Look for situations where it is
31913 beneficial to undo the work of one of the recursive calls.
31915 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31917 PR tree-optimization/70754
31918 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31919 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31920 combined stmt before it if not NULL.
31921 (combine_chains): Process refs reversely and compute dominance point
31924 2017-01-23 Martin Liska <mliska@suse.cz>
31926 PR tree-optimization/79196
31927 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31928 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31930 (strlen_optimize_stmt): Call the renamed function.
31932 2017-01-23 Michael Matz <matz@suse.de>
31934 PR tree-optimization/78384
31935 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31937 2017-01-23 Richard Biener <rguenther@suse.de>
31939 PR tree-optimization/79186
31940 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31941 both incoming edges before moving an assert.
31943 2017-01-23 Martin Jambor <mjambor@suse.cz>
31945 * ipa-prop.c (load_from_param_1): Removed.
31946 (load_from_unmodified_param): Bits from load_from_param_1 put back
31948 (load_from_param): Removed.
31949 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31950 with stmt. Reverted back to use of load_from_unmodified_param.
31952 2017-01-23 Martin Jambor <mjambor@suse.cz>
31955 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31956 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31957 field a pointer to garbage collected vector, mark lattices and
31958 ipcp_orig_node with GTY((skip)).
31959 (ipa_get_param_count): Adjust to descriptors being a pointer.
31960 (ipa_get_param): Likewise.
31961 (ipa_get_type): Likewise.
31962 (ipa_get_param_move_cost): Likewise.
31963 (ipa_set_param_used): Likewise.
31964 (ipa_get_controlled_uses): Likewise.
31965 (ipa_set_controlled_uses): Likewise.
31966 (ipa_is_param_used): Likewise.
31967 (ipa_node_params_t): Move into garbage collector. New methods insert
31969 (ipa_node_params_sum): Annotate wth GTY(()).
31970 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31972 (ipa_load_from_parm_agg): Adjust declaration.
31973 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31974 * ipa-profile.c (ipa_profile): Likewise.
31975 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31976 (ipa_populate_param_decls): Make descriptors parameter garbage
31978 (ipa_dump_param): Adjust to descriptors being a pointer.
31979 (ipa_alloc_node_params): Likewise.
31980 (ipa_initialize_node_params): Likewise.
31981 (load_from_param_1): Make descriptors parameter garbage collected.
31982 (load_from_unmodified_param): Likewise.
31983 (load_from_param): Likewise.
31984 (ipa_load_from_parm_agg): Likewise.
31985 (ipa_node_params::~ipa_node_params): Removed.
31986 (ipa_free_all_node_params): Remove call to delete operator.
31987 (ipa_node_params_t::insert): New.
31988 (ipa_node_params_t::remove): Likewise.
31989 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31990 copy known_csts and known_contexts vectors.
31991 (ipa_read_node_info): Adjust to descriptors being a pointer.
31992 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31994 (ipcp_transform_function): Make descriptors variable garbage
31997 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31999 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32000 * config/i386/avx512dqintrin.h: Ditto.
32001 * config/i386/avx512fintrin.h: Ditto.
32002 * config/i386/i386.c: Handle new builtins.
32003 * config/i386/i386-builtin.def: Add new builtins.
32004 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32005 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32007 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32008 Martin Liska <mliska@suse.cz>
32010 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32011 * asan.c (asan_expand_poison_ifn): Support stores and use
32012 appropriate ASAN report function.
32013 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32014 * internal-fn.def (ASAN_POISON_USE): Declare.
32015 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32016 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32017 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32018 ASAN_POISON calls w/o LHS.
32019 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32020 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32021 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32022 * gimplify.c (asan_poison_variables): Add attribute
32023 use_after_scope_memory to variables that really needs to live
32025 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32026 having the attribute.
32028 2017-01-23 Martin Liska <mliska@suse.cz>
32030 * asan.c (create_asan_shadow_var): New function.
32031 (asan_expand_poison_ifn): Likewise.
32032 * asan.h (asan_expand_poison_ifn): New declaration.
32033 * internal-fn.c (expand_ASAN_POISON): Likewise.
32034 * internal-fn.def (ASAN_POISON): New builtin.
32035 * sanopt.c (pass_sanopt::execute): Expand
32036 asan_expand_poison_ifn.
32037 * tree-inline.c (copy_decl_for_dup_finish): Make function
32039 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32040 * tree-ssa.c (is_asan_mark_p): New function.
32041 (execute_update_addresses_taken): Rewrite local variables
32042 (identified just by use-after-scope as addressable) into SSA.
32044 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32046 * doc/install.texi (Specific): opensource.apple.com uses https
32047 now. Remove trailing slash.
32049 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32051 * README.Portability: Remove note on an Irix compatibility issue.
32053 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32055 * gcov.c (INCLUDE_ALGORITHM): Define.
32056 (INCLUDE_VECTOR): Define.
32057 No longer include <vector> and <algorithm> directly.
32059 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32061 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32063 * doc/invoke.texi (Code Gen Options): Ditto.
32065 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32068 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32070 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32072 rtl-optimization/79125
32073 * cprop.c (local_cprop_pass): Handle cases where we make an
32074 unconditional trap.
32076 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32080 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32081 read from, for big endian.
32083 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32085 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32086 register pauth builtins for LP64 only.
32088 2017-01-20 Marek Polacek <polacek@redhat.com>
32091 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32094 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32096 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32098 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32099 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32100 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32102 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32105 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32106 in big-endian mode when they are not a single duplicated value.
32108 2017-01-20 Richard Biener <rguenther@suse.de>
32110 * BASE-VER: Bump to 7.0.1.
32112 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32114 * omp-low.c (omplow_simd_context): New struct. Use it...
32115 (lower_rec_simd_input_clauses): ...here and...
32116 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32117 references to idx, lane, max_vf, is_simt.
32119 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32121 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32124 2017-01-20 Martin Jambor <mjambor@suse.cz>
32126 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32127 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32129 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32130 (GTFILES): Rename hsa.c to hsa-common.c.
32131 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32132 * hsa-dump.c: Likewise.
32133 * hsa-gen.c: Likewise.
32134 * hsa-regalloc.c: Likewise.
32135 * ipa-hsa.c: Likewise.
32136 * omp-expand.c: Likewise.
32137 * omp-low.c: Likewise.
32138 * toplev.c: Likewise.
32140 2017-01-20 Marek Polacek <polacek@redhat.com>
32143 * doc/invoke.texi: Document -Wduplicated-branches.
32144 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32145 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32146 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32147 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32148 return 0 only when not OEP_LEXICOGRAPHIC.
32149 (fold_build_cleanup_point_expr): Use the expression
32150 location when building CLEANUP_POINT_EXPR.
32151 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32152 * tree.c (add_expr): Handle error_mark_node.
32154 2017-01-20 Martin Liska <mliska@suse.cz>
32157 * tree-profile.c (init_ic_make_global_vars): Do not call
32159 (gimple_init_gcov_profiler): Likewise.
32161 2017-01-20 Martin Liska <mliska@suse.cz>
32164 * cgraph.h (maybe_create_reference): Remove argument and
32166 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32168 * ipa-cp.c (create_specialized_node): Likewise.
32169 * symtab.c (symtab_node::maybe_create_reference): Handle
32170 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32172 2017-01-20 Martin Liska <mliska@suse.cz>
32174 * read-rtl-function.c (function_reader::create_function): Use
32175 build_decl instread of build_decl_stat.
32177 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32179 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32180 * config/i386/avx512dqintrin.h: Ditto.
32181 * config/i386/avx512fintrin.h: Ditto.
32182 * config/i386/i386-builtin-types.def: Add new types.
32183 * config/i386/i386.c: Handle new types.
32184 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32185 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32186 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32187 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32188 (__builtin_ia32_kshiftridi): New.
32189 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32191 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32195 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32196 define to rs6000_init_stack_protect_guard.
32197 (rs6000_init_stack_protect_guard): New function.
32199 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32200 Yunqiang Su <yunqiang.su@imgtec.com>
32202 * config.gcc (supported_defaults): Add madd4.
32203 (with_madd4): Add validation.
32204 (all_defaults): Add madd4.
32205 * config/mips/mips.opt (mmadd4): New option.
32206 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32208 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32210 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32211 (ISA_HAS_FUSED_MADD4): Likewise.
32212 * doc/invoke.texi (-mmadd4): Document the new option.
32213 * doc/install.texi (--with-madd4): Document the new option.
32215 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32217 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32218 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32219 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32220 (aarch64_init_pauth_hint_builtins): New.
32221 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32222 (aarch64_expand_builtin): Expand new builtins.
32224 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32226 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32227 * combine-stack-adj.c (no_unhandled_cfa): Handle
32228 REG_CFA_TOGGLE_RA_MANGLE.
32229 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32230 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32231 info for return address signing.
32232 (aarch64_expand_epilogue): Likewise.
32234 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32236 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32237 * config/aarch64/aarch64-protos.h
32238 (aarch64_return_address_signing_enabled): New declaration.
32239 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32241 (aarch64_expand_prologue): Sign return address before it's pushed onto
32243 (aarch64_expand_epilogue): Authenticate return address fetched from
32245 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32246 (aarch64_attributes): New function attributes for "sign-return-address".
32247 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32248 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32249 ("*do_return"): Generate combined instructions according to key index.
32250 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32251 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32253 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32254 * config/aarch64/aarch64.opt (msign-return-address=): New.
32255 * doc/extend.texi (AArch64 Function Attributes): Documents
32256 "sign-return-address=".
32257 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32259 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32261 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32262 overall option summary.
32264 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32266 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32267 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32268 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32269 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32271 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32273 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32274 -mpower9-minmax by default for -mcpu=power9.
32275 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32276 128-bit floating point.
32278 2017-01-20 Alan Modra <amodra@gmail.com>
32280 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32281 optimizing for size.
32283 2017-01-20 Alan Modra <amodra@gmail.com>
32286 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32287 for strcmp and strncmp from corresponding builtin decl.
32289 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32291 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32292 instead of i386/rtems-64.h.
32293 * config/i386/rtems-64.h: Remove.
32295 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32299 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32301 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32303 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32305 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32306 Change int to HOST_WIDE_INT.
32307 * config/aarch64/aarch64-protos.h
32308 (aarch64_simd_gen_const_vector_dup): Likewise.
32309 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32311 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32313 * langhooks-def.h (lhd_type_for_size): New decl.
32314 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32315 * langhooks.c (lhd_type_for_size): New function, taken from
32318 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32320 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32321 define_bypass for CR latency.
32322 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32323 (power9-alu2): Add define_bypass for CR latency.
32325 (power9-mul): Update insn latency.
32326 (power9-mul-compare): Update insn latency, bypass latency and remove
32329 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32331 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32333 * config/aarch64/aarch64.md
32334 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32335 aarch64_nopcrelative_literal_loads.
32336 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32338 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32340 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32341 TARGET_LOONGSON_3A.
32342 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32344 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32347 * config.gcc (supported_defaults): Add lxc1-sxc1.
32348 (with_lxc1_sxc1): Add validation.
32349 (all_defaults): Add lxc1-sxc1.
32350 * config/mips/mips.opt (mlxc1-sxc1): New option.
32351 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32353 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32354 __mips_no_lxc1_sxc1.
32355 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32356 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32357 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32359 2017-01-19 Richard Biener <rguenther@suse.de>
32361 PR tree-optimization/72488
32362 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32363 sure to restore SSA info.
32364 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32366 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32368 PR rtl-optimization/79121
32369 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32370 of the inner type when shifting an extended value.
32372 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32375 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32376 interposable aliases.
32378 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32381 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32382 Use the evmergelohi instruction.
32383 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32384 (mov_sitf_e500_subreg8_2_be): Likewise.
32385 (mov_sitf_e500_subreg12_2_le): Likewise.
32386 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32387 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32388 (mov_sitf_e500_subreg8_2_le): Likewise.
32389 (mov_sitf_e500_subreg12_2_be): Likewise.
32391 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32393 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32394 attribute from vecsimple to vecperm.
32395 (altivec_vbpermq2): Likewise.
32397 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32400 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32402 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32403 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32404 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32405 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32406 case where N arg is SIZE_MAX.
32407 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32408 (cmpstrsi): Add pattern.
32410 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32412 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32413 __builtin_vec_revb builtins.
32414 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32415 built-in functions to support generation of the ISA 3.0 XXBR<x>
32416 vector byte reverse instructions.
32417 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32418 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32419 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32420 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32421 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32422 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32423 (P9V_BUILTIN_VEC_REVB): Likewise.
32424 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32425 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32426 (p9_xxbrq_v16qi): Likewise.
32427 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32428 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32429 (p9_xxbrh_v8hi): Likewise.
32430 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32431 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32432 vec_revb built-in functions.
32434 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32436 PR rtl-optimization/78952
32437 * config/i386/i386.md (any_extract): New code iterator.
32438 (*insvqi_2): Use any_extract for source operand.
32439 (*insvqi_3): Use any_shiftrt for source operand.
32441 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32443 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32445 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32447 2017-01-18 Matthias Klose <doko@ubuntu.com>
32449 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32451 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32453 * config/rs6000/altivec.h (vec_bperm): Change #define.
32454 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32455 (altivec_vbpermq2): New define_insn.
32456 (altivec_vbpermd): Likewise.
32457 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32458 function interface.
32459 (VBPERMD): Likewise.
32460 (VBPERM): New polymorphic function interface.
32461 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32462 Add entries for P9V_BUILTIN_VEC_VBPERM.
32463 * doc/extend.texi: Add interfaces for vec_bperm.
32465 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32467 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32468 first letter of error messages.
32469 (s390_resolve_overloaded_builtin): Likewise.
32470 * config/s390/s390.c (s390_expand_builtin): Likewise.
32471 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32472 (s390_valid_target_attribute_inner_p): Likewise.
32473 * config/s390/s390.md ("tabort"): Likewise.
32475 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32477 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32478 (ISA_AVOID_DIV_HILO): New macro.
32479 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32480 (ISA_HAS_DDIV): Likewise.
32482 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32484 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32486 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32488 * doc/invoke.texi (fabi-version): Spelling fix.
32490 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32493 * doc/invoke.texi (fabi-version): Mention mangling fix for
32496 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32499 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32501 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32504 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32505 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32507 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32509 (stack_protect_test): Ditto.
32510 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32511 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32513 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32514 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32515 -mstack-protector-guard-offset=.
32516 (RS/6000 and PowerPC Options): Ditto.
32518 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32520 * config/i386/i386.h (MASK_CLASS_P): New define.
32521 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32522 there are no registers from different register sets also when
32523 mask registers are used. Update function comment.
32524 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32525 to (*k/*r) and (*k/*km) alternatives.
32527 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32529 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32530 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32531 (EH_RETURN_HANDLER_RTX): New define.
32532 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32533 Force frame pointer in EH return functions.
32534 (aarch64_expand_epilogue): Add barrier for eh_return.
32535 (aarch64_final_eh_return_addr): Remove.
32536 (aarch64_eh_return_handler_rtx): New function.
32537 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32539 (aarch64_eh_return_handler_rtx): New prototype.
32541 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32543 * config/rs6000/altivec.h (vec_rlmi): New #define.
32544 (vec_vrlnm): Likewise.
32545 (vec_rlnm): Likewise.
32546 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32547 (UNSPEC_VRLNM): Likewise.
32548 (VIlong): New mode iterator.
32549 (altivec_vrl<VI_char>mi): New define_insn.
32550 (altivec_vrl<VI_char>nm): Likewise.
32551 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32553 (VRLDNM): Likewise.
32554 (RLNM): New polymorphic function entry.
32555 (VRLWMI): New monomorphic function entry.
32556 (VRLDMI): Likewise.
32557 (RLMI): New polymorphic function entry.
32558 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32559 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32560 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32563 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32566 * dwarf2out.c (field_byte_offset): Restore the
32567 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32568 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32569 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32572 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32575 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32577 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32580 * dwarf2out.c (add_data_member_location_attribute): For constant
32581 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32582 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32583 DW_AT_byte_size attributes.
32585 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32587 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32588 after forcing to constant memory when the code model is medium.
32590 2017-01-17 Julia Koval <julia.koval@intel.com>
32593 * config/i386/avx512fintrin.h
32594 (_mm512_i32gather_ps): Change __addr type to void const*.
32595 (_mm512_mask_i32gather_ps): Ditto.
32596 (_mm512_i32gather_pd): Ditto.
32597 (_mm512_mask_i32gather_pd): Ditto.
32598 (_mm512_i64gather_ps): Ditto.
32599 (_mm512_mask_i64gather_ps): Ditto.
32600 (_mm512_i64gather_pd): Ditto.
32601 (_mm512_mask_i64gather_pd): Ditto.
32602 (_mm512_i32gather_epi32): Ditto.
32603 (_mm512_mask_i32gather_epi32): Ditto.
32604 (_mm512_i32gather_epi64): Ditto.
32605 (_mm512_mask_i32gather_epi64): Ditto.
32606 (_mm512_i64gather_epi32): Ditto.
32607 (_mm512_mask_i64gather_epi32): Ditto.
32608 (_mm512_i64gather_epi64): Ditto.
32609 (_mm512_mask_i64gather_epi64): Ditto.
32610 (_mm512_i32scatter_ps): Change __addr type to void*.
32611 (_mm512_mask_i32scatter_ps): Ditto.
32612 (_mm512_i32scatter_pd): Ditto.
32613 (_mm512_mask_i32scatter_pd): Ditto.
32614 (_mm512_i64scatter_ps): Ditto.
32615 (_mm512_mask_i64scatter_ps): Ditto.
32616 (_mm512_i64scatter_pd): Ditto.
32617 (_mm512_mask_i64scatter_pd): Ditto.
32618 (_mm512_i32scatter_epi32): Ditto.
32619 (_mm512_mask_i32scatter_epi32): Ditto.
32620 (_mm512_i32scatter_epi64): Ditto.
32621 (_mm512_mask_i32scatter_epi64): Ditto.
32622 (_mm512_i64scatter_epi32): Ditto.
32623 (_mm512_mask_i64scatter_epi32): Ditto.
32624 (_mm512_i64scatter_epi64): Ditto.
32625 (_mm512_mask_i64scatter_epi64): Ditto.
32626 * config/i386/avx512pfintrin.h
32627 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32628 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32629 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32630 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32631 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32632 (_mm512_prefetch_i32scatter_ps): Ditto.
32633 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32634 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32635 (_mm512_prefetch_i64scatter_pd): Ditto.
32636 (_mm512_prefetch_i64scatter_ps): Ditto.
32637 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32638 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32639 * config/i386/avx512vlintrin.h
32640 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32641 (_mm_mmask_i32gather_ps): Ditto.
32642 (_mm256_mmask_i32gather_pd): Ditto.
32643 (_mm_mmask_i32gather_pd): Ditto.
32644 (_mm256_mmask_i64gather_ps): Ditto.
32645 (_mm_mmask_i64gather_ps): Ditto.
32646 (_mm256_mmask_i64gather_pd): Ditto.
32647 (_mm_mmask_i64gather_pd): Ditto.
32648 (_mm256_mmask_i32gather_epi32): Ditto.
32649 (_mm_mmask_i32gather_epi32): Ditto.
32650 (_mm256_mmask_i32gather_epi64): Ditto.
32651 (_mm_mmask_i32gather_epi64): Ditto.
32652 (_mm256_mmask_i64gather_epi32): Ditto.
32653 (_mm_mmask_i64gather_epi32): Ditto.
32654 (_mm256_mmask_i64gather_epi64): Ditto.
32655 (_mm_mmask_i64gather_epi64): Ditto.
32656 (_mm256_i32scatter_ps): Change __addr type to void*.
32657 (_mm256_mask_i32scatter_ps): Ditto.
32658 (_mm_i32scatter_ps): Ditto.
32659 (_mm_mask_i32scatter_ps): Ditto.
32660 (_mm256_i32scatter_pd): Ditto.
32661 (_mm256_mask_i32scatter_pd): Ditto.
32662 (_mm_i32scatter_pd): Ditto.
32663 (_mm_mask_i32scatter_pd): Ditto.
32664 (_mm256_i64scatter_ps): Ditto.
32665 (_mm256_mask_i64scatter_ps): Ditto.
32666 (_mm_i64scatter_ps): Ditto.
32667 (_mm_mask_i64scatter_ps): Ditto.
32668 (_mm256_i64scatter_pd): Ditto.
32669 (_mm256_mask_i64scatter_pd): Ditto.
32670 (_mm_i64scatter_pd): Ditto.
32671 (_mm_mask_i64scatter_pd): Ditto.
32672 (_mm256_i32scatter_epi32): Ditto.
32673 (_mm256_mask_i32scatter_epi32): Ditto.
32674 (_mm_i32scatter_epi32): Ditto.
32675 (_mm_mask_i32scatter_epi32): Ditto.
32676 (_mm256_i32scatter_epi64): Ditto.
32677 (_mm256_mask_i32scatter_epi64): Ditto.
32678 (_mm_i32scatter_epi64): Ditto.
32679 (_mm_mask_i32scatter_epi64): Ditto.
32680 (_mm256_i64scatter_epi32): Ditto.
32681 (_mm256_mask_i64scatter_epi32): Ditto.
32682 (_mm_i64scatter_epi32): Ditto.
32683 (_mm_mask_i64scatter_epi32): Ditto.
32684 (_mm256_i64scatter_epi64): Ditto.
32685 (_mm256_mask_i64scatter_epi64): Ditto.
32686 (_mm_i64scatter_epi64): Ditto.
32687 (_mm_mask_i64scatter_epi64): Ditto.
32688 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32689 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32690 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32691 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32692 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32693 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32694 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32695 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32696 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32697 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32698 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32699 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32700 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32701 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32702 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32703 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32704 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32705 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32706 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32707 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32708 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32709 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32710 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32711 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32712 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32713 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32714 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32715 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32716 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32717 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32718 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32719 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32720 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32721 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32722 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32723 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32724 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32725 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32726 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32727 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32728 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32729 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32730 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32731 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32732 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32733 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32734 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32735 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32736 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32737 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32738 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32739 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32740 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32741 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32742 definitions accordingly.
32744 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32745 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32748 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32751 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32754 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32755 hard regs for inner regno.
32757 2017-01-17 Martin Liska <mliska@suse.cz>
32760 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32761 assumption and add comment.
32763 2017-01-17 Nathan Sidwell <nathan@acm.org>
32765 * ipa-visibility.c (localize_node): New function, broken out of ...
32766 (function_and_variable_visibility): ... here. Call it.
32768 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32770 PR middle-end/77445
32771 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32772 correctly set frequency of oudgoing edge.
32773 (duplicate_thread_path): Fix profile updating.
32775 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32778 * configure.ac: Add GCC_BASE_VER.
32779 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32780 version from BASE-VER file.
32781 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32782 (gcc.o): Depend on $(BASEVER).
32783 * common.opt (dumpfullversion): New option.
32784 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32785 * doc/invoke.texi: Document -dumpfullversion.
32786 * doc/install.texi: Document --with-gcc-major-version-only.
32787 * configure: Regenerated.
32789 2017-01-17 Richard Biener <rguenther@suse.de>
32791 PR tree-optimization/71433
32792 * tree-vrp.c (register_new_assert_for): Merge same asserts
32793 on all incoming edges.
32794 (process_assert_insertions_for): Handle insertions at the
32797 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32799 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32800 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32802 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32805 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32808 2017-01-17 Alan Modra <amodra@gmail.com>
32811 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32812 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32813 symbolic stack limit when pic.
32815 2017-01-16 Martin Sebor <msebor@redhat.com>
32817 PR tree-optimization/78608
32818 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32820 2017-01-16 Jeff Law <law@redhat.com>
32823 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32824 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32825 for several include directories that may be relative to sysroot.
32826 * config/i386/x-mingw32 (gplus_includedir): Define.
32827 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32828 (native_system_includedir): Likewise.
32829 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32830 override if TARGET_SYSTEM_ROOT is defined.
32831 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32833 PR tree-optimization/79090
32834 PR tree-optimization/33562
32835 PR tree-optimization/61912
32836 PR tree-optimization/77485
32837 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32838 and computed trims into the dump file.
32840 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32842 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32844 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32847 * gimplify.c (gimplify_init_constructor): If want_value and
32848 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32852 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32853 sequence. Formatting fixes.
32854 (doloop_optimize): Formatting fixes.
32857 * gcc.c (debug_level_greater_than_spec_func): New function.
32858 (static_spec_functions): Add debug-level-gt spec function.
32859 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32861 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32862 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32863 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32864 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32865 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32866 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32868 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32870 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32871 QImode fixups to general and mask registers only.
32873 2017-01-16 Carl Love <cel@us.ibm.com>
32875 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32876 for built-in functions
32877 vector signed char vec_nabs (vector signed char)
32878 vector signed short vec_nabs (vector signed short)
32879 vector signed int vec_nabs (vector signed int)
32880 vector signed long long vec_nabs (vector signed long long)
32881 vector float vec_nabs (vector float)
32882 vector double vec_nabs (vector double)
32883 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32885 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32886 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32887 * doc/extend.texi: Update the documentation file for the new built-in
32890 2017-01-16 Martin Sebor <msebor@redhat.com>
32892 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32895 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32897 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32898 UNSPEC_VSX__XXSPLTD to require special splat handling.
32900 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32903 * system.h: Poison strndup.
32905 2017-01-16 Alan Modra <amodra@gmail.com>
32908 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32911 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32913 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32915 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32917 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32918 call recog here. Assert that INSN_CODE (insn) is non-negative.
32920 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32923 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32925 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32926 in the currently scheduled RTL fragment.
32928 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32930 PR rtl-optimization/78751
32931 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32934 2017-01-14 Jeff Law <law@redhat.com>
32936 PR tree-optimization/79090
32937 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32938 variable length stores.
32939 (compute_trims): Delete dead assignment to *trim_tail.
32940 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32943 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32945 PR rtl-optimization/78626
32946 PR rtl-optimization/78727
32947 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32948 of a block, and split such blocks after everything else is finished.
32950 2017-01-14 Alan Modra <amodra@gmail.com>
32953 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32954 target legitimate_combined_insn.
32955 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32956 (rs6000_legitimate_combined_insn): New function.
32957 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32959 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32960 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32961 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32963 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32965 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32967 2017-01-13 Jeff Law <law@redhat.com>
32969 PR tree-optimization/33562
32970 PR tree-optimization/61912
32971 PR tree-optimization/77485
32972 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32974 (delete_dead_assignment): Likewise.
32975 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32976 statement to delete_dead_call and delete_dead_assignment.
32978 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32981 * substring-locations.c (format_warning_va): Strengthen case 1 so
32982 that both endpoints of the substring must be within the format
32983 range for just the substring to be printed.
32985 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32987 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32988 * config/i386/i386.c (ix86_target_string): Add missing options
32989 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32990 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32991 flags_other and ix86_target_other to flags2_other. Display unknown
32993 (ix86_valid_target_attribute_inner_p): Add missing options and
32994 reorder options by implied ISAs, as in ix86_target_string.
32996 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32998 * hash-table.h (hash_table::too_empty_p): New function.
32999 (hash_table::expand): Use it.
33000 (hash_table::traverse): Likewise.
33001 (hash_table::empty_slot): Use sizeof (value_type) instead of
33002 sizeof (PTR) to convert bytes to elements. Shrink the table
33003 if the current size is excessive for the current number of
33006 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33008 * ira-costs.c (record_reg_classes): Break from the inner loop
33009 early once alt_fail is known to be true. Update outer loop
33010 handling accordingly.
33012 2017-01-13 Jeff Law <law@redhat.com>
33014 * tree-ssa-dse.c (decrement_count): New function.
33015 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33016 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33017 when we know the partially dead statement is a mem* function.
33019 PR tree-optimization/61912
33020 PR tree-optimization/77485
33021 * tree-ssa-dse.c: Include expr.h.
33022 (maybe_trim_constructor_store): New function.
33023 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33025 PR tree-optimization/33562
33026 PR tree-optimization/61912
33027 PR tree-optimization/77485
33028 * doc/invoke.texi: Document new dse-max-object-size param.
33029 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33030 * tree-ssa-dse.c: Include params.h.
33031 (dse_store_status): New enum.
33032 (initialize_ao_ref_for_dse): New, partially extracted from
33034 (valid_ao_ref_for_dse, normalize_ref): New.
33035 (setup_live_bytes_from_ref, compute_trims): Likewise.
33036 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33037 (maybe_trim_partially_dead_store): Likewise.
33038 (maybe_trim_complex_store): Likewise.
33039 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33040 Track what bytes live from the original store. Return tri-state
33041 for dead, partially dead or live.
33042 (dse_dom_walker): Add constructor, destructor and new private members.
33043 (delete_dead_call, delete_dead_assignment): New extracted from
33045 (dse_optimize_stmt): Make a member of dse_dom_walker.
33046 Use initialize_ao_ref_for_dse.
33048 PR tree-optimization/33562
33049 PR tree-optimization/61912
33050 PR tree-optimization/77485
33051 * sbitmap.h (bitmap_count_bits): Prototype.
33052 (bitmap_clear_range, bitmap_set_range): Likewise.
33053 * sbitmap.c (bitmap_clear_range): New function.
33054 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33056 2017-01-13 Martin Liska <mliska@suse.cz>
33059 * function.c (set_cfun): Add new argument force.
33060 * function.h (set_cfun): Likewise.
33061 * ipa-inline-transform.c (inline_call): Use the function when
33062 strict alising from is dropped for function we inline to.
33064 2017-01-13 Richard Biener <rguenther@suse.de>
33066 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33067 for dumping GIMPLE INTEGER_CSTs.
33069 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33071 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33072 to 201112L since C++17.
33074 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33077 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33078 if -fsanitize=kernel-address is present.
33080 2017-01-13 Richard Biener <rguenther@suse.de>
33082 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33083 as _Literal ( type ) number in case usual suffixes do not
33084 preserve all information.
33086 2017-01-13 Richard Biener <rguenther@suse.de>
33088 PR tree-optimization/77283
33089 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33090 and ssa-iterators.h.
33091 (is_feasible_trace): Implement a cost model based on joiner
33094 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33097 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33098 char or short to __float128/_Float128 directly.
33100 2017-01-12 Martin Sebor <msebor@redhat.com>
33102 to -Wformat-overflow.
33103 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33104 (min_bytes_remaining): Same.
33105 (get_string_length): Same.
33106 (format_string): Same.
33107 (format_directive): Same.
33109 (pass_sprintf_length::handle_gimple_call): Same.
33111 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33113 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33114 info.nowrite calls with no lhs that can't throw. Return bool
33115 whether gsi_remove has been called or not.
33116 (pass_sprintf_length::handle_gimple_call): Return bool whether
33117 try_substitute_return_value called gsi_remove. Formatting fix.
33118 (pass_sprintf_length::execute): Don't use gsi_remove if
33119 handle_gimple_call returned true.
33122 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33123 be removed due to side-effects, don't remove following barrier nor
33124 turn the successor edge into fallthru edge.
33126 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33129 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33130 element-reversing loads and stores as not swappable.
33132 2017-01-12 Nathan Sidwell <nathan@acm.org>
33133 Nicolai Stange <nicstange@gmail.com>
33135 * combine.c (try_combine): Don't ignore result of overlap checking
33136 loop. Combine overlap & asm check into single loop.
33138 2017-01-12 Richard Biener <rguenther@suse.de>
33140 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33141 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33143 2017-01-12 Richard Biener <rguenther@suse.de>
33145 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33146 and TS_TARGET_OPTION directly derive from TS_BASE.
33147 * tree-core.h (tree_optimization_option): Derive from tree_base.
33148 (tree_target_option): Likewise.
33150 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33152 * config/i386/i386.c (memory_address_length): Increase len
33153 only when rip_relative_addr_p returns false.
33155 2017-01-11 Julia Koval <julia.koval@intel.com>
33157 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33158 (OPTION_MASK_ISA_SGX_SET): New.
33159 (ix86_handle_option): Handle OPT_msgx.
33160 * config.gcc: Added sgxintrin.h.
33161 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33162 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33163 * config/i386/i386.c (ix86_target_string): Add -msgx.
33165 (ix86_option_override_internal): Handle new options.
33166 (ix86_valid_target_attribute_inner_p): Add sgx.
33167 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33168 * config/i386/i386.opt: Add msgx.
33169 * config/i386/sgxintrin.h: New file.
33170 * config/i386/x86intrin.h: Add sgxintrin.h.
33172 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33175 * fold-const.c (maybe_nonzero_address): Return 1 for function
33177 (tree_single_nonzero_warnv_p): Don't handle function local objects
33181 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33184 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33187 * opts.c: Include "spellcheck.h"
33188 (struct string_fragment): New struct.
33189 (struct edit_distance_traits<const string_fragment &>): New
33191 (get_closest_sanitizer_option): New function.
33192 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33194 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33196 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33198 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33199 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33200 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33201 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33202 for initial die_offset if dwarf_split_debug_info.
33203 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33204 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33206 (output_skeleton_debug_sections): Formatting fix. Use
33207 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33208 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33210 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33212 * config/arm/cortex-a53.md: Add bypasses for
33213 cortex_a53_r2f_cvt.
33214 (cortex_a53_r2f): Only use for transfers.
33215 (cortex_a53_f2r): Likewise.
33216 (cortex_a53_r2f_cvt): Add reservation for conversions.
33217 (cortex_a53_f2r_cvt): Likewise.
33219 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33221 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33222 to all inlined functions, change static to extern.
33224 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33227 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33229 (arm_assemble_integer): Likewise.
33231 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33233 * config.gcc: Use new awk script to check CPU, FPU and architecture
33234 parameters for --with-... options.
33235 * config/arm/parsecpu.awk: New file
33236 * config/arm/arm-cpus.in: New file.
33237 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33239 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33241 * config/arm/t-arm: Update dependency rules.
33242 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33243 of processing .def files.
33244 * config/arm/genopt.sh: Deleted.
33245 * config/arm/gentune.sh: Deleted.
33246 * config/arm/arm-cores.def: Deleted.
33247 * config/arm/arm-arches.def: Deleted.
33248 * config/arm/arm-fpus.def: Deleted.
33249 * config/arm/arm-tune.md: Regenerated.
33250 * config/arm/arm-tables.opt: Regenerated.
33251 * config/arm/arm-cpu.h: New generated file.
33252 * config/arm/arm-cpu-data.h: New generated file.
33253 * config/arm/arm-cpu-cdata.h: New generated file.
33255 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33258 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33260 (input_varpool_node): Unpack dynamically_initialized bit.
33262 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33264 PR rtl-optimization/79032
33265 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33266 the alignment of the adjusted memory reference against that of MODE,
33267 instead of the alignment of the original memory reference.
33269 2017-01-11 Martin Jambor <mjambor@suse.cz>
33271 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33273 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33274 decorated functions.
33276 2017-01-11 Richard Biener <rguenther@suse.de>
33278 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33279 set range/nonnull info for PHI results. Do not set it on
33280 stmts marked for removal.
33282 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33284 * expr.c (store_field): In the bitfield case, fetch the return value
33285 from the registers before applying a single big-endian adjustment.
33286 Always do a final load for a BLKmode value not larger than a word.
33288 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33291 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33292 that we correctly handle column numbers greater than
33293 LINE_MAP_MAX_COLUMN_NUMBER.
33295 2017-01-10 Martin Sebor <msebor@redhat.com>
33297 PR middle-end/78245
33298 * gimple-ssa-sprintf.c (get_destination_size): Call
33299 {init,fini}object_sizes.
33300 * tree-object-size.c (addr_object_size): Adjust.
33301 (pass_through_call): Adjust.
33302 (pass_object_sizes::execute): Adjust.
33303 * tree-object-size.h (fini_object_sizes): Declare.
33305 2017-01-10 Martin Sebor <msebor@redhat.com>
33307 PR tree-optimization/78775
33308 * builtins.c (get_size_range): Move...
33309 * calls.c: ...to here.
33310 (alloc_max_size): Accept zero argument.
33311 (operand_signed_p): Remove.
33312 (maybe_warn_alloc_args_overflow): Call get_size_range.
33313 * calls.h (get_size_range): Declare.
33315 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33317 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33318 from TI's devices.csv file as of September 2016.
33319 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33321 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33323 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33324 * doc/invoke.texi: Likewise.
33325 * doc/md.texi: Likewise.
33326 * doc/objc.texi: Likewise.
33328 2017-01-10 Joshua Conner <joshconner@google.com>
33330 * config/arm/fuchsia-elf.h: New file.
33331 * config/fuchsia.h: New file.
33332 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33333 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33335 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33337 2016-01-10 Richard Biener <rguenther@suse.de>
33339 PR tree-optimization/79034
33340 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33341 Propagate out degenerate PHIs in the joiner.
33343 2017-01-10 Martin Liska <mliska@suse.cz>
33345 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33346 (sort_congruence_classes_by_decl_uid): Likewise.
33347 (sort_congruence_class_groups_by_decl_uid): Likewise.
33348 (sem_item_optimizer::merge_classes): Sort class, groups in these
33349 classes and members in the groups by DECL_UID of declarations.
33350 This would make merge operations stable.
33352 2017-01-10 Martin Liska <mliska@suse.cz>
33354 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33355 usage of m_classes_vec.
33356 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33357 (sem_item_optimizer::get_group_by_hash): Likewise.
33358 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33359 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33360 (sem_item_optimizer::verify_classes): Likewise.
33361 (sem_item_optimizer::process_cong_reduction): Likewise.
33362 (sem_item_optimizer::dump_cong_classes): Likewise.
33363 (sem_item_optimizer::merge_classes): Likewise.
33364 * ipa-icf.h (congruence_class_hash): Rename from
33365 congruence_class_group_hash. Remove declaration of m_classes_vec.
33367 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33369 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33370 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33371 * config.gcc: Add avx512vpopcntdqintrin.h.
33372 * config/i386/avx512vpopcntdqintrin.h: New.
33373 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33374 * config/i386/i386-builtin-types.def: Add new types.
33375 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33376 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33377 __builtin_ia32_vpopcountq_v8di_mask): New.
33378 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33379 __AVX512VPOPCNTDQ__.
33380 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33381 (PTA_AVX512VPOPCNTDQ): Define.
33382 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33383 TARGET_AVX512VPOPCNTDQ_P): Define.
33384 * config/i386/i386.opt: Add mavx512vpopcntdq.
33385 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33386 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33388 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33390 PR middle-end/77484
33391 * predict.def (PRED_CALL): Set to 67.
33393 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33395 * expr.c (store_field): In the bitfield case, if the value comes from
33396 a function call and is of an aggregate type returned in registers, do
33397 not modify the field mode; extract the value in all cases if the mode
33398 is BLKmode and the size is not larger than a word.
33400 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33403 * config/i386/cpuid.h: Fix undefined behavior.
33405 2017-01-04 Jeff Law <law@redhat.com>
33407 PR tree-optimization/79007
33408 PR tree-optimization/67955
33409 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33410 conservative for pt.null when flag_non_call_exceptions is on.
33412 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33414 PR translation/79019
33415 PR translation/79020
33416 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33417 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33418 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33420 * config/avr/avr.opt (maccumulate-args): Likewise.
33421 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33422 * common.opt (freport-bug): Likewise.
33423 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33424 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33425 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33426 translatable string.
33427 * config/i386/i386.c (function_value_32): Likewise.
33428 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33429 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33431 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33432 * common/config/msp430/msp430-common.c (msp430_handle_option):
33434 * symtab.c (symtab_node::verify_base): Likewise.
33435 * opts.c (set_debug_level): Likewise.
33436 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33437 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33438 missing whitespace to translatable strings.
33439 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33440 * config/sh/superh.h: Likewise.
33441 * config/i386/xopintrin.h: Likewise.
33442 * config/i386/znver1.md: Likewise.
33443 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33444 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33445 * double-int.h (struct double_int): Likewise.
33446 * double-int.c (div_and_round_double): Likewise.
33447 * wide-int.cc: Likewise.
33448 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33449 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33450 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33451 (crossjumps_occurred): ... this.
33452 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33455 PR tree-optimization/78899
33456 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33457 returning bool return struct loop *, NULL for failure and the new
33459 (versionable_outer_loop_p): Don't version outer loop if it has
33460 dont_vectorized bit set.
33461 (tree_if_conversion): When versioning outer loop, ensure
33462 tree_if_conversion is performed also on the inner loop of the
33463 non-vectorizable outer loop copy.
33464 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33465 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33466 prevent vectorization of it.
33467 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33468 the outer loop vectorization of the non-scalar version is attempted
33469 before vectorization of the inner loop in scalar version. If
33470 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33471 vectorization of its inner loop.
33472 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33473 has 2 inner loops, rename also on edges from bb whose single pred
33474 is outer_loop->header. Fix typo in function comment.
33476 2017-01-09 Martin Sebor <msebor@redhat.com>
33479 * asan.c (asan_emit_stack_protection): Increase local buffer size
33480 to avoid snprintf truncation warning.
33482 2017-01-09 Andrew Pinski <apinski@cavium.com>
33484 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33485 to reference thunderx2t99 for the tuning structure
33486 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33488 (thunderx2t99_extra_costs): This.
33489 * config/aarch64/aarch64-tune.md: Regenerate.
33490 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33491 (vulcan_addrcost_table): This.
33492 (vulcan_regmove_cost): Rename to ...
33493 (thunderx2t99_regmove_cost): This.
33494 (vulcan_vector_cost): Rename to ...
33495 (thunderx2t99_vector_cost): this.
33496 (vulcan_branch_cost): Rename to ...
33497 (thunderx2t99_branch_cost): This.
33498 (vulcan_tunings): Rename to ...
33499 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33500 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33502 2017-01-09 Martin Jambor <mjambor@suse.cz>
33506 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33507 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33508 (propagate_vr_accross_jump_function): Use the above function for all
33509 value range computations for pass-through jump functions and type
33510 converasion from explicit value range values.
33511 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33512 parameters from TYPE_ARG_TYPES.
33513 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33514 (ipa_write_node_info): Stream type of the actual argument.
33515 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33517 2017-01-09 Martin Liska <mliska@suse.cz>
33520 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33521 (lookup_compiler): Do not show error message with have_E.
33523 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33525 PR tree-optimization/78938
33526 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33527 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33528 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33529 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33532 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33534 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33537 2017-01-09 Richard Biener <rguenther@suse.de>
33539 PR tree-optimization/78997
33540 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33541 name condition properly.
33543 2017-01-09 Richard Biener <rguenther@suse.de>
33546 * dwarf2out.c (is_cxx): New overload with context.
33547 (is_naming_typedef_decl): Use it.
33549 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33551 * invoke.texi (Option Summary): Correct spacing in option lists
33552 and add line breaks to fix over-long lines.
33554 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33556 PR middle-end/17660
33558 * extend.texi (Common Variable Attributes): Add xref to GCC
33559 Internals manual to explain mode attribute keywords.
33561 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33564 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33565 and Preprocessor Options.
33566 (Options for Linking): Document -pthread here....
33567 (RS/6000 and PowerPC Options): ...not here.
33568 (Solaris 2 Options): ...or here.
33569 * doc/cppopts.texi: Document -pthread.
33571 2017-01-08 Martin Sebor <msebor@redhat.com>
33573 PR middle-end/77708
33574 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33575 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33576 New member functions.
33577 (format_directive): Used them.
33579 (pass_sprintf_length::handle_gimple_call): Same.
33580 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33581 to avoid truncation for any argument.
33582 (extract_affine_mul): Same.
33583 * tree.c (get_file_function_name): Same.
33585 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33587 PR middle-end/77484
33588 * predict.def (PRED_INDIR_CALL): Set to 86.
33590 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33592 PR preprocessor/54124
33593 * doc/cppopts.texi: Reformat -d subtable to list the full name
33594 of the options. Add cross-reference to the docs for the general
33595 compiler -d options.
33596 * doc/invoke.texi (Developer Options): Add cross-reference to the
33597 preprocessor-specific -d option documentation.
33599 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33601 PR preprocessor/13498
33602 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33603 redudant material, and reflect new command-line options.
33604 (System Headers): Likewise.
33606 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33608 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33609 -isystem, and -idirafter. Copy-edit.
33610 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33611 default for -ftrack-macro-expansion. Delete obsolete and
33612 badly-formatted implementation details about -fdebug-cpp output.
33613 * doc/cppwarnopts.texi: Copy-edit.
33615 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33618 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33619 that the transition from a max line width >= 1<<10 to narrower
33620 lines works correctly.
33622 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33624 * doc/options.texi (PerFunction): New.
33625 * opt-functions.awk (switch_flags): Map both Optimization and
33626 PerFunction to CL_OPTIMIZATION.
33627 * opth-gen.awk: Test for PerFunction flag along with
33629 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33630 it only when the latter is present. Skip those that don't in
33631 the hash function generator.
33632 * common.opt (fvar-tracking): Mark as PerFunction instead of
33634 (fvar-tracking-assignments): Likewise.
33635 (fvar-tracking-assignments-toggle): Likewise.
33636 (fvar-tracking-uninit): Likewise.
33638 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33640 PR translation/79018
33641 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33644 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33647 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33648 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33649 TARGET_LONG_JUMP_TABLE_OFFSETS.
33650 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33651 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33652 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33653 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33654 * config/m68k/m68k.md (tablejump expander): Likewise.
33655 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33656 TARGET_LONG_JUMP_TABLE_OFFSETS.
33657 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33658 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33660 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33661 David Holsgrove <david.holsgrove@xilinx.com>
33663 * common/config/microblaze/microblaze-common.c
33664 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33665 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33667 * config/microblaze/microblaze.c (microblaze_must_save_register)
33668 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33670 (microblaze_eh_return): New function.
33671 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33672 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33673 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33674 * config/microblaze/microblaze.md (eh_return): New pattern.
33676 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33678 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33679 GCC_DIAGNOSTIC_STRINGIFY): Define.
33681 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33683 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33685 * config/arm/arm.md (<mcrr>): New.
33687 * config/arm/arm.c (arm_arch5te): New.
33688 (arm_option_override): Set arm_arch5te.
33689 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33691 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33692 (arm_mcrr_qualifiers): ... this. New.
33693 (MRRC_QUALIFIERS): Define to...
33694 (arm_mrrc_qualifiers): ... this. New.
33695 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33697 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33698 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33699 (MRRCI, mrrc, MRRC): New.
33700 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33701 VUNSPEC_MRRC2): New.
33703 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33705 * config/arm/arm.md (<mcr>): New.
33707 * config/arm/arm.c (arm_coproc_builtin_available): Add
33708 support for mcr, mrc, mcr2 and mrc2.
33709 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33710 (arm_mcr_qualifiers): ... this. New.
33711 (MRC_QUALIFIERS): Define to ...
33712 (arm_mrc_qualifiers): ... this. New.
33713 (MCR_QUALIFIERS): Define to ...
33714 (arm_mcr_qualifiers): ... this. New.
33715 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33717 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33718 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33719 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33720 VUNSPEC_MRC2): New.
33722 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33724 * config/arm/arm.md (*ldc): New.
33728 * config/arm/arm.c (arm_coproc_builtin_available): Add
33729 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33730 (arm_coproc_ldc_stc_legitimate_address): New.
33731 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33732 'qualifier_const_pointer'.
33733 (LDC_QUALIFIERS): Define to...
33734 (arm_ldc_qualifiers): ... this. New.
33735 (STC_QUALIFIERS): Define to...
33736 (arm_stc_qualifiers): ... this. New.
33737 * config/arm/arm-protos.h
33738 (arm_coproc_ldc_stc_legitimate_address): New.
33739 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33740 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33741 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33742 stc2, stcl, stc2l): New.
33743 * config/arm/constraints.md (Uz): New.
33744 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33745 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33746 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33747 VUNSPEC_STC2L): New.
33749 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33751 * config/arm/arm.md (<cdp>): New.
33752 * config/arm/arm.c (neon_const_bounds): Rename this ...
33753 (arm_const_bounds): ... this.
33754 (arm_coproc_builtin_available): New.
33755 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33756 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33757 (CDP_QUALIFIERS): Define to...
33758 (arm_cdp_qualifiers): ... this. New.
33760 (arm_expand_builtin_args): Add case for 6 arguments.
33761 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33762 (arm_const_bounds): ... this.
33763 (arm_coproc_builtin_available): New.
33764 * config/arm/arm_acle.h (__arm_cdp): New.
33766 * config/arm/arm_acle_builtins.def (cdp): New.
33768 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33769 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33770 'arm_const_bounds'.
33771 * config/arm/types.md (coproc): New.
33772 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33773 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33774 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33775 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33777 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33779 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33780 (UBINOP_QUALIFIERS): New.
33782 (acle_builtin_data): New. Change comment.
33783 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33784 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33785 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33786 arm_acle_builtins.def.
33787 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33788 (arm_init_acle_builtins): New.
33789 (CRC32_BUILTIN): Remove.
33790 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33791 crc32cb, crc32ch and crc32cw.
33792 (arm_init_crc32_builtins): Remove.
33793 (arm_init_builtins): Use arm_init_acle_builtins rather
33794 than arm_init_crc32_builtins.
33795 (arm_expand_acle_builtin): New.
33796 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33797 * config/arm/arm_acle_builtins.def: New.
33799 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33801 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33802 (arm_builtin_datum): ... this.
33803 (arm_init_neon_builtin): Rename to ...
33804 (arm_init_builtin): ... this. Add a new parameters PREFIX
33805 and USE_SIG_IN_NAME.
33806 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33807 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33808 'arm_builtin_datum'.
33809 (arm_init_vfp_builtins): Likewise.
33810 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33811 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33812 (arm_expand_neon_args): Rename to ...
33813 (arm_expand_builtin_args): ... this. Rename builtin_arg
33814 enum values and differentiate between ARG_BUILTIN_MEMORY
33815 and ARG_BUILTIN_NEON_MEMORY.
33816 (arm_expand_neon_builtin_1): Rename to ...
33817 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33818 values, arm_expand_builtin_args and add bool parameter NEON.
33819 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33820 (arm_expand_vfp_builtin): Likewise.
33821 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33823 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33825 PR middle-end/77484
33826 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33827 * predict.c (tree_estimate_probability_bb): Reverse direction of
33828 polymorphic call predictor.
33830 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33832 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33833 (determine_pass_name_match): ...this new function and...
33834 (should_skip_pass_p): ...this new function.
33836 2017-01-06 Nathan Sidwell <nathan@acm.org>
33838 * ipa-visibility.c (function_and_variable_visibility): Reformat
33839 comments and long lines. Remove extrneous if.
33840 * symtab.c (symtab_node::make_decl_local): Fix code format.
33841 (symtab_node::set_section_for_node): Fix comment typo.
33843 2017-01-06 Martin Liska <mliska@suse.cz>
33846 * lra-constraints.c: Rename invariant to lra_invariant.
33847 * predict.c (set_even_probabilities): Initialize e to NULL.
33849 2017-01-05 Martin Sebor <msebor@redhat.com>
33851 PR tree-optimization/78910
33852 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33853 (format_integer): Correct off-by-one error in the handling
33854 of precision with negative numbers in signed conversions..
33856 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33858 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33860 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33862 PR tree-optimization/71016
33863 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33864 factor_out_conditional_conversion. Formatting fix.
33865 (factor_out_conditional_conversion): Add cond_stmt argument.
33866 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33867 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33870 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33872 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33873 read-rtl-function.o, and selftest-rtl.o.
33874 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33875 (selftest::aarch64_test_loading_full_dump): New function.
33876 (selftest::aarch64_run_selftests): New function.
33877 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33878 selftest::aarch64_run_selftests.
33879 * config/i386/i386.c
33880 (selftest::ix86_test_loading_dump_fragment_1): New function.
33881 (selftest::ix86_test_loading_call_insn): New function.
33882 (selftest::ix86_test_loading_full_dump): New function.
33883 (selftest::ix86_test_loading_unspec): New function.
33884 (selftest::ix86_run_selftests): Call the new functions.
33885 * emit-rtl.c (maybe_set_max_label_num): New function.
33886 * emit-rtl.h (maybe_set_max_label_num): New decl.
33887 * function.c (instantiate_decls): Guard call to
33888 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33889 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33891 * gensupport.c (gen_reader::gen_reader): Pass "false"
33892 for new "compact" param of rtx_reader.
33893 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33894 rather than an empty string for NULL strings.
33895 * read-md.c: Potentially include config.h rather than bconfig.h.
33896 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33897 (have_error): New global, copied from errors.c.
33898 (md_reader::read_name): Rename to...
33899 (md_reader::read_name_1): ...this, adding "out_loc" param,
33900 and converting "missing name or number" to returning false, rather
33902 (md_reader::read_name): Reimplement in terms of read_name_1.
33903 (md_reader::read_name_or_nil): New function.
33904 (md_reader::read_string): Handle "(nil)" by returning NULL.
33905 (md_reader::md_reader): Add new param "compact".
33906 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33907 (md_reader::read_file): New method.
33908 * read-md.h (md_reader::md_reader): Add new param "compact".
33909 (md_reader::read_file): New method.
33910 (md_reader::is_compact): New accessor.
33911 (md_reader::read_name): Convert return type from void to file_location.
33912 (md_reader::read_name_or_nil): New decl.
33913 (md_reader::read_name_1): New decl.
33914 (md_reader::m_compact): New field.
33915 (noop_reader::noop_reader): Pass "false" for new "compact" param
33917 (rtx_reader::rtx_reader): Add new "compact" param.
33918 (rtx_reader::read_rtx_operand): Make virtual and convert return
33919 type from void to rtx.
33920 (rtx_reader::read_until): New decl.
33921 (rtx_reader::handle_any_trailing_information): New virtual function.
33922 (rtx_reader::postprocess): New virtual function.
33923 (rtx_reader::finalize_string): New virtual function.
33924 (rtx_reader::m_in_call_function_usage): New field.
33925 (rtx_reader::m_reuse_rtx_by_id): New field.
33926 * read-rtl-function.c: New file.
33927 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33928 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33929 (selftest::verify_three_block_rtl_cfg): New decl.
33930 * read-rtl-function.h: New file.
33931 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33932 For host, include function.h, memmodel.h, and emit-rtl.h.
33933 (one_time_initialization): New function.
33934 (struct compact_insn_name): New struct.
33935 (compact_insn_names): New array.
33936 (find_code): Handle insn codes in compact dumps.
33937 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33938 (bind_subst_iter_and_attr): Likewise.
33939 (add_condition_to_string): Likewise.
33940 (add_condition_to_rtx): Likewise.
33941 (apply_attribute_uses): Likewise.
33942 (add_current_iterators): Likewise.
33943 (apply_iterators): Likewise.
33944 (initialize_iterators): Guard usage of apply_subst_iterator with
33945 #ifdef GENERATOR_FILE.
33946 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33947 (md_reader::read_mapping): Likewise.
33948 (add_define_attr_for_define_subst): Likewise.
33949 (add_define_subst_attr): Likewise.
33950 (read_subst_mapping): Likewise.
33951 (check_code_iterator): Likewise.
33952 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33954 (one_time_initialization): New function.
33955 (rtx_reader::read_until): New method.
33956 (read_flags): New function.
33957 (parse_reg_note_name): New function.
33958 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33959 Handle reuse_rtx ids.
33960 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33961 Add parsing support for RTL dumps, mirroring the special-cases in
33962 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33963 values, and calling handle_any_trailing_information.
33964 (rtx_reader::read_rtx_operand): Convert return type from void
33965 to rtx, returning return_rtx. Handle case 'e'. Call
33966 finalize_string on XSTR and XTMPL fields.
33967 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33968 "(nil)" values were omitted. Call the postprocess vfunc on the
33970 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33971 class ctor. Initialize m_in_call_function_usage. Call
33972 one_time_initialization.
33973 * rtl-tests.c (selftest::test_uncond_jump): Call
33974 set_new_first_and_last_insn.
33975 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33976 * selftest-rtl.c: New file.
33977 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33978 (selftest::get_insn_by_uid): New decl.
33979 * selftest-run-tests.c (selftest::run_tests): Call
33980 read_rtl_function_c_tests.
33981 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33982 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33985 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33987 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33988 operands in a special way. Assert that pos+len <= mode precision.
33990 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33992 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33993 3 argument Alias with unlimited for the negative form.
33994 (fno-vect-cost-model): Removed.
33996 2017-01-05 Martin Liska <mliska@suse.cz>
33998 * hsa-gen.c (gen_hsa_divmod): New function.
33999 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34001 2017-01-05 Martin Liska <mliska@suse.cz>
34004 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34007 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34009 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34010 small constant length operands.
34012 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34014 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34015 between loop iterations.
34017 2017-01-05 Martin Liska <mliska@suse.cz>
34020 * gimplify.c (gimplify_decl_expr): Compare to
34021 asan_poisoned_variables instread of checking flags.
34022 (gimplify_target_expr): Likewise.
34023 (gimplify_expr): Likewise.
34024 (gimplify_function_tree): Conditionally initialize
34025 asan_poisoned_variables.
34027 2017-01-04 Jeff Law <law@redhat.com>
34029 PR tree-optimizatin/78812
34030 * rtl.h (contains_mem_rtx_p): Prototype.
34031 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34032 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34033 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34034 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34035 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34037 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34039 * input.c (assert_char_at_range): Default-initialize actual_range.
34041 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34043 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34044 to match the caller.
34046 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34048 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34049 insns after final jump in test to emit dummy move.
34051 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34053 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34054 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34056 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34058 * multiple_target.c (create_dispatcher_calls): Init e_next.
34059 * tree-ssa-loop-split.c (split_loop): Init border.
34060 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34063 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34068 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34069 (altivec_register_operand): Do not return true if the operand
34070 contains a SUBREG mixing SImode and SFmode.
34071 (vsx_register_operand): Likewise.
34072 (vsx_reg_sfsubreg_ok): New predicate.
34073 (vfloat_operand): Do not return true if the operand contains a
34074 SUBREG mixing SImode and SFmode.
34075 (vint_operand): Likewise.
34076 (vlogical_operand): Likewise.
34077 (gpc_reg_operand): Likewise.
34078 (int_reg_operand): Likewise.
34079 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34080 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34081 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34083 (rs6000_emit_move_si_sf_subreg): New helper function.
34084 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34085 fixup SUBREGs involving SImode and SFmode.
34086 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34087 numbers for the new peephole2 optimization.
34088 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34089 the GLIBC math library that do AND/IOR/XOR operations on single
34090 precision floating point.
34091 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34092 target macros to say whether we need to avoid SUBREGs mixing
34094 (TARGET_ALLOW_SF_SUBREG): Likewise.
34095 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34096 (UNSPEC_SI_FROM_SF): Likewise.
34097 (iorxor): Change spacing.
34098 (and_ior_xor): New iterator for AND, IOR, and XOR.
34099 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34100 (movdi_from_sf_zero_ext): Likewise.
34101 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34102 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34103 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34104 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34105 (fms<mode>4): Likewise.
34106 (fnma<mode>4): Likewise.
34107 (fnms<mode>4): Likewise.
34108 (nfma<mode>4): Likewise.
34109 (nfms<mode>4): Likewise.
34111 2017-01-04 Marek Polacek <polacek@redhat.com>
34114 * doc/invoke.texi: Document -Wpointer-compare.
34116 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34118 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34121 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34122 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34125 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34128 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34129 documentation of the powerpc_popcntb_ok attribute.
34130 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34131 code to issue warning messages if a requested CPU configuration is
34132 not supported by the binary (assembler and loader) toolchain.
34133 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34134 made to define a built-in function that has been disabled.
34135 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34136 made to define a built-in function that has been disabled.
34137 (altivec_init_builtins): Add comment explaining why definition
34138 of the DST built-in functions is not preceded by an assertion
34139 check. Add assertions to prevent ICE if attempts are made to
34140 define an altivec predicate or an abs* built-in function that has
34142 (htm_init_builtins): Add comment explaining why definition of the
34143 htm built-in functions is not preceded by an assertion check.
34145 2017-01-04 Jeff Law <law@redhat.com>
34147 PR tree-optimizatin/67955
34148 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34149 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34150 the points-to solution does not include pt_null. Use DECL_PT_UID
34153 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34155 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34156 Use gen_int_mode instead of gen_lopwart for const_int operands.
34158 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34160 PR tree-optimization/71563
34161 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34162 out of range value - has low bits known to be zero.
34164 2017-01-04 Alan Modra <amodra@gmail.com>
34166 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34167 * configure: Regenerate.
34168 * config.in: Regenerate.
34170 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34173 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34174 a substring of the message, but strcmp with the whole message. Ifdef
34175 ENABLE_NLS, translate the message first using dgettext.
34177 2017-01-03 Jeff Law <law@redhat.com>
34179 PR tree-optimizatin/78856
34180 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34181 (mark_threaded_blocks): Remove code to truncate thread paths that
34182 cross multiple loop headers. Instead invalidate the cached loop
34183 iteration information and handle case of a thread path walking
34184 into an irreducible region.
34186 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34189 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34190 assertions. Add support for doing the signbit if the IEEE 128-bit
34191 floating point value is in a GPR.
34192 * config/rs6000/rs6000.md (Fsignbit): Delete.
34193 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34194 Update the length attribute if the value is in a GPR.
34195 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34196 the sign or zero extension instruction, since the value is always 0/1.
34197 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34200 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34201 extracting SImode to a GPR register so that we can generate a
34202 store, limit the vector to be in a traditional Altivec register
34203 for the vextuwrx instruction.
34205 2017-01-03 Ian Lance Taylor <iant@google.com>
34207 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34209 2017-01-03 Martin Sebor <msebor@redhat.com>
34211 PR tree-optimization/78696
34212 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34213 precision. Use MPFR for %f for greater fidelity. Correct handling
34215 (pass_sprintf_length::compute_format_length): Set width and precision
34216 specified by asrerisk to void_node for vararg functions.
34217 (try_substitute_return_value): Adjust dump output.
34219 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34221 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34223 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34225 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34226 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34227 -mlra/-mno-lra was passed to the compiler.
34229 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34231 PR rtl-optimization/65618
34232 * emit-rtl.c (try_split): Move initialization of "before" and
34233 "after" to just before the call to emit_insn_after_setloc.
34235 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34237 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34239 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34241 * dwarf2out.c (gen_enumeration_type_die): When
34242 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34244 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34246 PR tree-optimization/78965
34247 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34248 Change first argument from const call_info & to call_info &. For %n
34249 set info.nowrite to false.
34251 PR middle-end/78901
34252 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34253 possibly throwing calls.
34255 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34256 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34257 and fns handling, rather than in a separate case SSA_NAME.
34259 2017-01-02 Jeff Law <law@redhat.com>
34261 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34262 fixes for first_period and second_period variables.
34264 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34267 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34268 (*insvqi_1): New insn pattern.
34269 (*insvqi_1_mem_rex64): Ditto.
34270 (*insvqi_2): Ditto.
34271 (*insvqi_3): Rename from *insvqi.
34273 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34275 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34277 * doc/cfg.texi (Edges): Remove reference to Java.
34278 (Maintaining the CFG): Ditto.
34280 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34282 PR middle-end/77674
34283 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34284 transparent aliases.
34286 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34288 PR middle-end/77484
34289 * predict.def (PRED_CALL): Update hitrate.
34290 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34291 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34292 into direct/indirect/polymorphic variants.
34294 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34296 Update copyright years.
34298 * gcc.c (process_command): Update copyright notice dates.
34299 * gcov-dump.c (print_version): Ditto.
34300 * gcov.c (print_version): Ditto.
34301 * gcov-tool.c (print_version): Ditto.
34302 * gengtype.c (create_file): Ditto.
34303 * doc/cpp.texi: Bump @copying's copyright year.
34304 * doc/cppinternals.texi: Ditto.
34305 * doc/gcc.texi: Ditto.
34306 * doc/gccint.texi: Ditto.
34307 * doc/gcov.texi: Ditto.
34308 * doc/install.texi: Ditto.
34309 * doc/invoke.texi: Ditto.
34311 Copyright (C) 2017 Free Software Foundation, Inc.
34313 Copying and distribution of this file, with or without modification,
34314 are permitted in any medium without royalty provided the copyright
34315 notice and this notice are preserved.