1 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4 (get_mem_refs_of_builtin_call): Likewise.
5 * builtins.c (expand_builtin_apply): Adjust call to
6 allocate_dynamic_stack_space.
7 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8 the third argument to allocate_dynamic_stack_space, otherwise -1.
9 (expand_builtin): Deal with all alloca variants.
10 (is_inexpensive_builtin): Likewise.
11 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
12 * calls.c (special_function_p): Deal with all alloca variants.
13 (initialize_argument_information): Adjust call to
14 allocate_dynamic_stack_space.
15 (expand_call): Likewise.
16 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
17 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
18 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
19 use it for the stack usage computation.
20 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
21 * function.c (gimplify_parameters): Turn BUILT_IN_ALLOCA_WITH_ALIGN
22 into BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX and pass maximum size.
23 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
24 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
25 (in_loop_p): Remove first argument and useless check.
26 (pass_walloca::execute): Remove useless test and adjust call to above.
27 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
28 * gimplify.c (gimplify_vla_decl): Turn BUILT_IN_ALLOCA_WITH_ALIGN into
29 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX and pass maximum size.
30 (gimplify_call_expr): Deal with all alloca variants.
31 * hsa-gen.c (gen_hsa_alloca): Likewise.
32 (gen_hsa_insns_for_call): Likewise.
33 * ipa-pure-const.c (special_builtin_state): Likewise.
34 * tree-chkp.c (chkp_build_returned_bound): Likewise.
35 * tree-object-size.c (alloc_object_size): Likewise.
36 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
37 (call_may_clobber_ref_p_1): Likewise.
38 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
39 (ccp_fold_stmt): Likewise.
40 (optimize_stack_restore): Likewise.
41 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
42 (mark_all_reaching_defs_necessary_1): Likewise.
43 (propagate_necessity): Likewise.
44 (eliminate_unnecessary_stmts): Likewise.
45 * tree.c (build_common_builtin_nodes): Build
46 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
47 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
48 (CASE_BUILT_IN_ALLOCA): Likewise.
49 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
51 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
54 * dwarf2out.c (new_die_raw): New static inline function.
55 (new_die): Use it to create the DIE.
56 (add_AT_external_die_ref): Likewise.
57 (clone_die): Likewise.
58 (clone_as_declaration): Likewise.
59 (dwarf2out_vms_debug_main_pointer): Likewise.
60 (base_type_die): Likewise. Remove early return for corner cases.
61 Do not call add_pubtype on the DIE here.
62 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
63 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
64 typedefs for base types with DW_AT_endianity. Make sure a DIE with
65 native order exists for base types, attach the DIE manually and call
66 add_pubtype on it. Do not equate a reverse order DIE to the type.
68 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
70 * config/arm/arm.c (align_ok_ldrd_strd): New function.
71 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
73 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
75 2017-10-19 Jakub Jelinek <jakub@redhat.com>
77 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
78 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
79 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
80 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
81 * opts.c (sanitizer_opts): Add builtin.
82 * ubsan.c (instrument_builtin): New function.
83 (pass_ubsan::execute): Call it.
84 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
85 * doc/invoke.texi: Document -fsanitize=builtin.
87 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
88 builtins, store max (log2 (align), 0) into uchar field instead of
89 align into uptr field.
90 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
91 store uchar 0 field instead of uptr 0 field.
92 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
93 instead of passing one address of struct with 2 locations pass
94 two addresses of structs with 1 location each.
95 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
96 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
97 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
98 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
99 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
100 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
101 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
102 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
104 2017-10-19 Martin Liska <mliska@suse.cz>
107 * file-find.c (remove_prefix): Remove.
108 * file-find.h (remove_prefix): Likewise.
109 * gcc-ar.c: Remove smartness of lookup.
111 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
113 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
114 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
115 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
117 2017-10-19 Jakub Jelinek <jakub@redhat.com>
120 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
121 (setcc + and to xor + setcc): New peephole2.
123 2017-10-19 Tom de Vries <tom@codesourcery.com>
125 * doc/sourcebuild.texi (Test Directives, Variants of
126 dg-require-support): Add dg-require-stack-size.
128 2017-10-19 Martin Liska <mliska@suse.cz>
131 * gimplify.c (gimplify_decl_expr): Do not instrument variables
132 that have a large alignment.
133 (gimplify_target_expr): Likewise.
135 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
137 PR rtl-optimization/82602
138 * ira.c (rtx_moveable_p): Return false for volatile asm.
140 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
143 * config/i386/i386-modes.def (CCGZ): New CC mode.
144 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
145 * config/i386/predicates.md (ix86_comparison_operator):
147 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
148 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
149 with double-word subtraction.
150 (put_condition_code): Handle CCGZmode.
152 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
154 * wide-int.cc (debug (const wide_int &)): New.
155 (debug (const wide_int *)): New.
156 (debug (const widest_int &)): New.
157 (debug (const widest_int *)): New.
159 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
162 * lra-constraints.c (curr_insn_transform): Use non-input operand
163 instead of output one for matched reload.
165 2017-10-18 Bin Cheng <bin.cheng@arm.com>
167 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
168 (tree-ssa-loop-ivopts.h): New header file.
169 (struct builtin_info): New fields.
170 (classify_builtin_1): Compute and record base and offset parts for
171 memset builtin partition by calling strip_offset.
172 (offset_cmp, fuse_memset_builtins): New functions.
173 (finalize_partitions): Fuse adjacent memset partitions by calling
175 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
176 Expose the interface.
177 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
179 2017-10-18 Bin Cheng <bin.cheng@arm.com>
181 PR tree-optimization/82574
182 * tree-loop-distribution.c (find_single_drs): New parameter. Check
183 that data reference must be executed exactly once per iteration
184 against the outermost loop in nest.
185 (classify_partition): Update call to above function.
187 2017-10-18 Richard Biener <rguenther@suse.de>
189 PR tree-optimization/82591
190 * graphite.c (graphite_transform_loops): Move code gen message
192 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
193 Here. Handle scop_to_isl_ast failing.
194 (scop_to_isl_ast): Limit the number of ISL operations.
196 2017-10-18 Richard Biener <rguenther@suse.de>
198 * graphite-isl-ast-to-gimple.c
199 (translate_isl_ast_to_gimple::set_rename): Simplify.
200 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
201 (graphite_copy_stmts_from_block): ... here.
202 (copy_bb_and_scalar_dependences): Simplify.
203 (add_parameters_to_ivs_params): Canonicalize.
204 (generate_entry_out_of_ssa_copies): Simplify.
205 * graphite-sese-to-poly.c (extract_affine_name): Simplify
206 by passing in ISL dimension.
207 (parameter_index_in_region_1): Rename to ...
208 (parameter_index_in_region): ... this.
209 (extract_affine): Adjust assert, pass down parameter index.
210 (add_param_constraints): Use range-info when available.
211 (build_scop_context): Adjust.
212 * sese.c (new_sese_info): Adjust.
213 (free_sese_info): Likewise.
214 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
215 Remove unused typedefs.
216 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
218 2017-10-18 Martin Liska <mliska@suse.cz>
220 * combine.c (simplify_compare_const): Add gcc_fallthrough.
222 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
224 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
225 (s390_sched_init): Do not reset s390_sched_state if we entered the
226 current basic block via a fallthru edge and all others are unlikely.
228 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
230 * config/s390/s390.c (NUM_SIDES): New variable.
231 (LONGRUNNING_THRESHOLD): New variable.
232 (LATENCY_FACTOR): New variable.
233 (s390_sched_score): Decrease score for long-running instructions on
235 (s390_sched_variable_issue): Perform bookkeeping for long-running
238 2017-10-18 Richard Biener <rguenther@suse.de>
240 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
241 Simplify with removal of the parameter rename map.
242 (set_rename): Likewise.
243 (should_copy_to_new_region): Likewise.
244 (graphite_copy_stmts_from_block): Likewise.
245 (copy_bb_and_scalar_dependences): Remove initialization of
246 unused copied_bb_map.
248 (copy_internal_parameters): Likewise.
249 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
250 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
252 (parameter_index_in_region_1): Rename to ...
253 (assign_parameter_index_in_region): ... this. Assert we have
254 a parameter we handle.
255 (scan_tree_for_params): Adjust.
256 * sese.h (parameter_rename_map_t): Remove.
257 (struct sese_info_t): Remove unused parameter_rename_map and
258 copied_bb_map members.
259 * sese.c (new_sese_info): Adjust.
260 (free_sese_info): Likewise.
262 2017-10-18 Martin Liska <mliska@suse.cz>
265 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
268 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
270 * doc/invoke.texi (ffunction-sections and fdata-sections):
273 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
275 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
276 the use statement can throw internally.
278 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
280 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
281 any RTX present on the RHS of a SET.
282 * compare-elim.c (try_eliminate_compare): Restore comment.
284 2017-10-17 Jakub Jelinek <jakub@redhat.com>
286 * langhooks.h (struct lang_hooks): Document that tree_size langhook
287 may be also called on tcc_type nodes.
288 * langhooks.c (lhd_tree_size): Likewise.
290 2017-10-17 David Malcolm <dmalcolm@redhat.com>
292 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
293 format_warning_at_substring.
294 (maybe_warn): Convert source_range * param to a location_t. Pass
295 UNKNOWN_LOCATION rather than NULL to fmtwarn.
296 (format_directive): Remove code to extract source_ranges and
297 source_range * in favor of just a location_t.
298 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
300 * substring-locations.c (format_warning_va): Convert
301 source_range * param to a location_t.
302 (format_warning_at_substring): Likewise.
303 * substring-locations.h (format_warning_va): Likewise.
304 (format_warning_at_substring): Likewise.
306 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
308 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
310 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
311 and scatter/gather ops.
313 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
314 vec_gather_load and vec_scatter_store.
315 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
316 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
317 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
318 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
319 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
320 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
322 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
324 * reg-stack.c (compare_for_stack_reg): Add bool argument.
325 Detect FTST instruction and handle its register pops. Only pop
326 second operand if can_pop_second_op is true.
327 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
328 set can_pop_second_op to false in the compare_for_stack_reg call.
330 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
331 output_fp_compare for stack register operands.
332 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
333 instructions here. Do not emit stack register pops here. Assert
334 that FCOMPP pops next to top stack register. Rewrite function.
336 2017-10-17 Nathan Sidwell <nathan@acm.org>
339 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
340 use DECL_ASSEMBLER_NAME_RAW.
343 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
346 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
347 Wilco Dijkstra <wilco.dijkstra@arm.com>
349 * builtins.c (expand_builtin_update_setjmp_buf): Add a
350 converstion to Pmode from the buf_addr.
352 2017-10-17 Richard Biener <rguenther@suse.de>
354 * graphite-dependences.c (scop_get_reads_and_writes): Change
355 output parameters to references.
357 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
359 PR 71026/tree-optimization
360 * fold-const.c (distribute_real_division): Removed.
361 (fold_binary_loc): Remove calls to distribute_real_divison.
363 2017-10-17 Richard Biener <rguenther@suse.de>
365 * graphite-scop-detection.c
366 (scop_detection::stmt_has_simple_data_refs_p): Always use
367 the full nest as region.
368 (try_generate_gimple_bb): Likewise.
369 * sese.c (scalar_evolution_in_region): Simplify now that
370 SCEV can handle instantiation in regions.
371 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
372 in the non-loop part of a function if requested.
374 2017-10-17 Richard Biener <rguenther@suse.de>
376 PR tree-optimization/82563
377 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
379 (graphite_regenerate_ast_isl): Call it.
380 * graphite-scop-detection.c (build_scops): Remove entry edge split.
382 2017-10-17 Jakub Jelinek <jakub@redhat.com>
384 PR tree-optimization/82549
385 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
386 Formatting fixes. Instead of calling make_bit_field_ref with negative
389 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
391 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
392 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
393 _mm_maskz_reduce_ss): New.
394 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
395 __builtin_ia32_reducess_mask): Ditto..
396 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
397 * config/i386/sse.md (reduces<mode>): Renamed to ...
398 (reduces<mode><mask_scalar_name>): ... this.
399 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
401 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
402 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
404 2017-10-16 David Malcolm <dmalcolm@redhat.com>
406 * Makefile.in (OBJS): Add unique-ptr-tests.o.
407 * selftest-run-tests.c (selftest::run_tests): Call
408 selftest::unique_ptr_tests_cc_tests.
409 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
410 * unique-ptr-tests.cc: New file.
412 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
415 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
417 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
418 (make_hard_regno_born, make_hard_regno_dead): Update
419 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
421 2017-10-16 Jeff Law <law@redhat.com>
423 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
425 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
427 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
429 2017-10-16 Olivier Hainque <hainque@adacore.com>
431 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
432 with_cpu if we were configured for an e500v2 target cpu name.
434 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
436 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
437 * doc/invoke.texi: Document +nodsp as a valid extension for
440 2017-10-16 Martin Liska <mliska@suse.cz>
442 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
443 (test_set_range): Likewise.
444 (test_range_functions): Rename to ...
445 (test_bit_in_range): ... this.
446 (sbitmap_c_tests): Add new test.
448 2017-10-16 Tamar Christina <tamar.christina@arm.com>
450 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
451 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
452 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
454 2017-10-16 Tamar Christina <tamar.christina@arm.com>
456 * config/aarch64/aarch64-builtins.c
457 (aarch64_types_quadopu_lane_qualifiers): New.
458 (TYPES_QUADOPU_LANE): New.
459 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
460 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
461 (aarch64_<sur>dot_laneq<vsi2qi>): New.
462 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
463 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
464 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
465 (Vdottype, DOTPROD): New.
466 (sur): Add SDOT and UDOT.
468 2017-10-16 Tamar Christina <tamar.christina@arm.com>
470 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
471 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
472 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
473 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
474 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
475 (cortex-a75.cortex-a55): Likewise.
476 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
478 2017-10-16 Tamar Christina <tamar.christina@arm.com>
480 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
481 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
482 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
483 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
484 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
485 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
486 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
487 * config/arm/types.md (neon_dot, neon_dot_q): New.
488 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
490 2017-10-16 Tamar Christina <tamar.christina@arm.com>
492 * config/arm/arm.h (TARGET_DOTPROD): New.
493 * config/arm/arm.c (arm_arch_dotprod): New.
494 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
495 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
496 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
497 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
498 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
499 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
500 * doc/invoke.texi (armv8.2-a): Document dotprod
502 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
504 * i386.c (ix86_vec_cost): New function.
505 (ix86_rtx_costs): Handle vector operations better.
506 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
507 * x86-tune-costs.h: Add new costs to all tables.
509 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
511 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
513 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
514 divsd, sqrtss and sqrtsd
515 * x86-tune-costs.h: Add new entries to all costs.
516 (znver1_cost): Fix to match real instruction latencies.
518 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
519 Michael Collison <michael.collison@arm.com>
521 * compare-elim.c: Include emit-rtl.h.
522 (can_merge_compare_into_arith): New function.
523 (try_validate_parallel): Likewise.
524 (try_merge_compare): Likewise.
525 (try_eliminate_compare): Call the above when no previous clobber
527 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
530 2017-10-14 Jakub Jelinek <jakub@redhat.com>
534 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
535 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
539 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
540 to be any operand_equal_p operands. For & (B - 1) require
541 B to be power of 2. Recognize
542 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
544 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
547 * optabs.c (expand_memory_blockage): Fix call of
548 targetm.have_memory_blockage.
550 2017-10-14 Jakub Jelinek <jakub@redhat.com>
553 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
554 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
555 objects to extra_objs instead of overwriting it.
557 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
559 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
560 Use any_fp_register_operand as operand[3] predicate. Simplify
561 equality test for operands[2] and operands[4] memory location.
562 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
563 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
564 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
565 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
566 any_fp_register_operand as operand[1] predicate. Simplify
567 equality test for operands[0] and operands[3] memory location.
568 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
569 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
570 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
572 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
574 * target-insns.def: Add memory_blockage.
575 * optabs.c (expand_memory_blockage): New function.
576 (expand_asm_memory_barrier): Rename ...
577 (expand_asm_memory_blockage): ... to this.
578 (expand_mem_thread_fence): Call expand_memory_blockage
579 instead of expand_asm_memory_barrier.
580 (expand_mem_singnal_fence): Ditto.
581 (expand_atomic_load): Ditto.
582 (expand_atomic_store): Ditto.
583 * doc/md.texi (Standard Pattern Names For Generation):
584 Document memory_blockage instruction pattern.
586 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
588 * config/rl78/rl78.c (rl78_emit_libcall): New function.
589 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
590 * config/rl78/rl78.md: New define_expand "adddi3".
592 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
594 * cfghooks.c (verify_flow_info): Disable check that all probabilities
597 2017-10-13 Jeff Law <law@redhat.com>
599 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
601 2017-10-13 Jakub Jelinek <jakub@redhat.com>
604 * internal-fn.c (expand_mul_overflow): If both operands have
605 the same highpart of -1 or 0 and the topmost bit of lowpart
606 is different, overflow is if res <= 0 rather than res < 0.
608 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
610 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
611 TARGET_P9_VECTOR code for unaligned_load case.
613 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
615 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
617 2017-10-13 Nathan Sidwell <nathan@acm.org>
619 * tree-core.h (tree_contains_struct): Make bool.
620 * tree.c (tree_contains_struct): Likewise.
621 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
622 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
623 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
624 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
625 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
627 2017-10-13 Richard Biener <rguenther@suse.de>
629 * graphite-isl-ast-to-gimple.c
630 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
631 parameters and dominance check.
632 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
633 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
634 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
635 Do not update SSA form here or do intermediate IL verification.
636 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
637 (graphite_initialize): Remove check on the number of loops in
638 the function and inline into graphite_transform_loops.
639 (graphite_finalize): Inline into graphite_transform_loops.
640 (graphite_transform_loops): Perform SSA update and IL verification
642 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
644 2017-10-13 Richard Biener <rguenther@suse.de>
646 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
647 graphite_expression_type_precision): Avoid global constructor
649 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
650 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
651 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
652 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
653 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
654 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
656 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
659 * config/i386/i386.h (ix86_red_zone_size): New.
660 * config/i386/i386.md (push peephole2s): Replace
661 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
663 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
664 Alan Hayward <alan.hayward@arm.com>
665 David Sherwood <david.sherwood@arm.com>
667 * combine.c (can_change_dest_mode): Reject changes in
668 REGMODE_NATURAL_SIZE.
670 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
671 Alan Hayward <alan.hayward@arm.com>
672 David Sherwood <david.sherwood@arm.com>
674 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
675 (expand_debug_source_expr): Likewise.
676 * combine.c (combine_simplify_rtx): Likewise.
677 * cse.c (fold_rtx): Likewise.
678 * fwprop.c (canonicalize_address): Likewise.
679 * targhooks.c (default_shift_truncation_mask): Likewise.
681 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
682 Alan Hayward <alan.hayward@arm.com>
683 David Sherwood <david.sherwood@arm.com>
685 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
686 (widened_mode): Likewise.
687 (expand_unop): Likewise.
688 * ree.c (transform_ifelse): Likewise.
689 (merge_def_and_ext): Likewise.
690 (combine_reaching_defs): Likewise.
691 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
693 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
694 Alan Hayward <alan.hayward@arm.com>
695 David Sherwood <david.sherwood@arm.com>
697 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
698 * combine.c (gen_lowpart_for_combine): Likewise.
699 * dwarf2out.c (rtl_for_decl_location): Likewise.
700 * final.c (alter_subreg): Likewise.
701 * rtlhooks.c (gen_lowpart_general): Likewise.
702 (gen_lowpart_if_possible): Likewise.
704 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
705 Alan Hayward <alan.hayward@arm.com>
706 David Sherwood <david.sherwood@arm.com>
708 * calls.c (expand_call): Use subreg_lowpart_offset.
709 * cse.c (cse_insn): Likewise.
710 * regcprop.c (copy_value): Likewise.
711 (copyprop_hardreg_forward_1): Likewise.
713 2017-10-13 Jakub Jelinek <jakub@redhat.com>
716 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
717 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
718 =Q constraints to +Q and into insn condition add check
719 that operands[0] and operands[1] are equal.
720 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
721 =Q constraints to +Q and into insn condition add check
722 that operands[0] is equal to either operands[1] or operands[2].
725 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
726 instead of handling MINUS_EXPR twice (once for each argument),
727 canonicalize operand order and handle just once, use rtype where
728 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
731 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
732 any values of __C while still being pattern recognizable as a simple
735 2017-10-13 Richard Biener <rguenther@suse.de>
737 PR tree-optimization/82451
739 2017-10-02 Richard Biener <rguenther@suse.de>
741 PR tree-optimization/82355
742 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
743 a mapping for the enclosing loop but avoid generating one for
745 (copy_bb_and_scalar_dependences): Remove premature codegen
746 error on PHIs in blocks duplicated into multiple places.
747 * graphite-scop-detection.c
748 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
749 in the region use it as loop and nest to analyze the DR in.
750 (try_generate_gimple_bb): Likewise.
751 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
752 (add_loop_constraints): For blocks in a loop not in the region
753 create a dimension with a single iteration.
754 * sese.h (gbb_loop_at_index): Remove assert.
756 * cfgloop.c (loop_preheader_edge): For the loop tree root
757 return the single successor of the entry block.
758 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
759 Reset the SCEV hashtable and niters.
760 * graphite-scop-detection.c
761 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
762 assert that we only have POLYNOMIAL_CHREC that vary in loops
763 contained in the region.
764 (scop_detection::graphite_can_represent_expr): Adjust.
765 (scop_detection::stmt_has_simple_data_refs_p): For loops
766 not in the region set loop to NULL. The nest is now the
767 entry edge to the region.
768 (try_generate_gimple_bb): Likewise.
769 * sese.c (scalar_evolution_in_region): Adjust for
770 instantiate_scev change.
771 * tree-data-ref.h (graphite_find_data_references_in_stmt):
772 Make nest parameter the edge into the region.
773 (create_data_ref): Likewise.
774 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
775 entry edge into a region and adjust instantiate_scev calls.
776 (create_data_ref): Likewise.
777 (graphite_find_data_references_in_stmt): Likewise.
778 (find_data_references_in_stmt): Pass the loop preheader edge
779 from the nest argument.
780 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
781 parameter the edge into the region.
782 (instantiate_parameters): Use the loop preheader edge as entry.
783 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
785 (get_instantiated_value_entry): Make instantiate_below parameter
786 the edge into the region.
787 (instantiate_scev_name): Likewise. Adjust dominance checks,
788 when we cannot use loop-based instantiation instantiate by
789 walking use-def chains.
790 (instantiate_scev_poly): Adjust.
791 (instantiate_scev_binary): Likewise.
792 (instantiate_scev_convert): Likewise.
793 (instantiate_scev_not): Likewise.
794 (instantiate_array_ref): Remove.
795 (instantiate_scev_3): Likewise.
796 (instantiate_scev_2): Likewise.
797 (instantiate_scev_1): Likewise.
798 (instantiate_scev_r): Do not blindly handle N-operand trees.
799 Do not instantiate array-refs. Handle all constants and invariants.
800 (instantiate_scev): Make instantiate_below parameter
801 the edge into the region.
802 (resolve_mixers): Use the loop preheader edge for the region
803 parameter to instantiate_scev_r.
804 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
806 2017-10-13 Richard Biener <rguenther@suse.de>
808 PR tree-optimization/82525
809 * graphite-isl-ast-to-gimple.c
810 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
812 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
813 Fail code generation when we cannot represent the isl integer.
814 (binary_op_to_tree): Elide modulo operations that are no-ops
815 in the type we code generate. Remove now superfluous code
818 2017-10-13 Richard Biener <rguenther@suse.de>
820 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
821 (scop_detection::harmful_loop_in_region): Remove premature
823 (scop_detection::graphite_can_represent_scev): We can handle
824 pointer IVs just fine.
826 2017-10-13 Alan Modra <amodra@gmail.com>
828 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
829 "Clobbers and Scratch Registers". Add paragraph on
830 alternative to clobbers for scratch registers and OpenBLAS
833 2017-10-13 Alan Modra <amodra@gmail.com>
835 * doc/extend.texi (Clobbers): Correct vax example. Delete old
836 example of a memory input for a string of known length. Move
837 commentary out of table. Add a number of new examples
838 covering array memory inputs.
840 2017-10-12 Martin Liska <mliska@suse.cz>
842 PR tree-optimization/82493
843 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
844 (test_range_functions): New function.
845 (sbitmap_c_tests): Likewise.
846 * selftest-run-tests.c (selftest::run_tests): Run new tests.
847 * selftest.h (sbitmap_c_tests): New function.
849 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
851 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
853 * config/rs6000/amo.h: Fix spacing issue.
855 2017-10-12 Jakub Jelinek <jakub@redhat.com>
858 * config/i386/i386.md (*ashl<mode>3_mask_1,
859 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
860 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
863 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
865 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
866 (profile_probability): Set max_probability
867 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
869 * profile-count.c (profile_probability::differs_from_p): Do not
870 rely on max_probaiblity == 10000
872 2017-10-12 Jeff Law <law@redhat.com>
874 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
877 2017-10-12 Martin Sebor <msebor@redhat.com>
881 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
882 (handle_alias_pairs): Call it.
883 * common.opt (-Wattribute-alias): New option.
884 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
885 * doc/invoke.texi (-Wattribute-alias): Document.
887 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
890 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
892 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
894 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
895 (make_hard_regno_born, make_hard_regno_dead): Update
896 bb_killed_pseudos and bb_gen_pseudos.
898 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
900 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
902 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
904 * config/alpha/alpha.c (alpha_split_conditional_move):
905 Use std::swap instead of manually swapping.
906 (alpha_stdarg_optimize_hook): Ditto.
907 (alpha_canonicalize_comparison): Ditto.
909 2017-10-12 Bin Cheng <bin.cheng@arm.com>
911 * tree-loop-distribution.c (struct builtin_info): New struct.
912 (struct partition): Refactor fields into struct builtin_info.
913 (partition_free): Free struct builtin_info.
914 (build_size_arg_loc, build_addr_arg_loc): Delete.
915 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
916 information from struct builtin_info.
917 (find_single_drs): New function refactored from classify_partition.
918 Also moved builtin validity checks to this function.
919 (compute_access_range, alloc_builtin): New functions.
920 (classify_builtin_st, classify_builtin_ldst): New functions.
921 (classify_partition): Refactor code into functions find_single_drs,
922 classify_builtin_st and classify_builtin_ldst.
923 (distribute_loop): Don't do runtime alias check when distributing
925 (find_seed_stmts_for_distribution): New function.
926 (pass_loop_distribution::execute): Refactor code finding seed
927 stmts into above function. Support distribution for the innermost
928 two-level loop nest. Adjust dump information.
930 2017-10-12 Bin Cheng <bin.cheng@arm.com>
932 * tree-loop-distribution.c: Adjust the general comment.
933 (NUM_PARTITION_THRESHOLD): New macro.
934 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
935 (classify_partition): Skip builtin pattern of loop nest's inner loop.
936 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
937 in call to build_partition_graph.
938 (finalize_partitions): New parameter. Make loop distribution more
939 conservative by fusing more partitions.
940 (distribute_loop): Don't do runtime alias check in case of loop nest
942 (find_seed_stmts_for_distribution): New function.
943 (prepare_perfect_loop_nest): New function.
944 (pass_loop_distribution::execute): Refactor code finding seed stmts
945 and loop nest into above functions. Support loop nest distribution.
946 Adjust dump information accordingly.
948 2017-10-12 Bin Cheng <bin.cheng@arm.com>
950 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
951 and set PTYPE_SEQUENTIAL for merged partition.
953 2017-10-12 Richard Biener <rguenther@suse.de>
955 PR tree-optimization/69728
957 2017-09-19 Richard Biener <rguenther@suse.de>
959 PR tree-optimization/69728
960 * graphite-sese-to-poly.c (schedule_error): New global.
961 (add_loop_schedule): Handle empty domain by failing the
963 (build_original_schedule): Handle schedule_error.
965 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
966 domain by returning an unchanged schedule.
968 2017-10-12 Jakub Jelinek <jakub@redhat.com>
970 * genrecog.c (validate_pattern): For VEC_SELECT verify that
971 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
973 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
975 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
978 2017-10-12 Jakub Jelinek <jakub@redhat.com>
981 * expr.c (store_field): Don't optimize away bitsize == 0 store
982 from CALL_EXPR with addressable return type.
984 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
986 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
987 * config/rs6000/rs6000.md (sel): Delete mode attribute.
988 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
989 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
990 TARGET_ISEL instead of TARGET_ISEL<sel>.
992 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
994 * config/rs6000/rs6000.c
995 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
997 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
999 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
1001 (reg_or_cint_operand): Fix comment.
1002 (reg_or_zero_operand): New predicate.
1003 * config/rs6000/rs6000-protos.h (output_isel): Delete.
1004 * config/rs6000/rs6000.c (output_isel): Delete.
1005 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
1006 instead of reg_or_cint_operand. Output instruction directly (not via
1008 (isel_unsigned_<mode>): Ditto.
1009 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
1010 gpc_reg_operand. Add an instruction alternative for this. Output
1011 instruction directly.
1012 (*isel_reversed_unsigned_<mode>): Ditto.
1014 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1016 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
1017 (TARGET_CANONICALIZE_COMPARISON): Define.
1019 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
1022 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1023 Check whether the dest is REG before adding REG_EQUIV note.
1025 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1028 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1030 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1031 (make_hard_regno_born, make_hard_regno_dead): Update
1032 bb_killed_pseudos and bb_gen_pseudos.
1034 2017-10-11 Nathan Sidwell <nathan@acm.org>
1036 * incpath.h (enum incpath_kind): Name enum, prefix values.
1037 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
1038 * incpath.c (heads, tails): Use INC_MAX.
1039 (add_env_var_paths, add_standard_paths): Use incpath_kind.
1040 (merge_include_chains, split_quote_chain,
1041 register_include_chains): Update incpath_kind names.
1042 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
1043 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
1045 (add_framework_path, darwin_register_objc_includes): Likewise.
1046 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
1048 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1050 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
1051 Do not use float_operator operator predicate.
1052 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1053 * config/i386/predicates.md (float_operator): Remove predicate.
1055 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1057 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
1058 (*jccxf_i387): Ditto.
1059 (*jcc<mode>_i387): Ditto.
1060 (*jccu<mode>_i387): Ditto.
1061 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
1062 (*jcc_*_i387 splitters): Remove.
1063 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
1064 * config/i386/i386.c (ix86_split_fp_branch): Remove.
1065 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
1068 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1070 * profile-count.h (slow_safe_scale_64bit): New function.
1071 (safe_scale_64bit): New inline.
1072 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
1073 * profile-count.c: Include wide-int.h
1074 (slow_safe_scale_64bit): New.
1076 2017-10-11 Nathan Sidwell <nathan@acm.org>
1078 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
1079 HAS_DECL_ASSEMBLER_NAME_P.
1080 * gimple-expr.c (gimple_decl_printable_name: Check
1081 HAS_DECL_ASSEMBLER_NAME_P too.
1082 * ipa-utils.h (type_in_anonymous_namespace_p): Check
1083 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
1084 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
1085 * passes.c (rest_of_decl_compilation): Check
1086 HAS_DECL_ASSEMBLER_NAME_P too.
1087 * recog.c (verify_changes): Likewise.
1088 * tree-pretty-print.c (dump_decl_name): Likewise.
1089 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
1091 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
1092 (SET_DECL_ASSEMBLER_NAME): Use it.
1093 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
1094 (COPY_DECL_ASSEMBLER_NAME): Likewise.
1095 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
1097 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1099 * config.gcc (i386, x86_64): Add extra objects.
1100 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
1101 (ix86_min_insn_size): Declare.
1102 (ix86_issue_rate): Declare.
1103 (ix86_adjust_cost): Declare.
1104 (ia32_multipass_dfa_lookahead): Declare.
1105 (ix86_macro_fusion_p): Declare.
1106 (ix86_macro_fusion_pair_p): Declare.
1107 (ix86_bd_has_dispatch): Declare.
1108 (ix86_bd_do_dispatch): Declare.
1109 (ix86_core2i7_init_hooks): Declare.
1110 (ix86_atom_sched_reorder): Declare.
1111 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
1112 (COSTS_N_BYTES): Move to x86-tune-costs.h.
1113 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
1114 (rip_relative_addr_p): Rename to ...
1115 (ix86_rip_relative_addr_p): ... this one; export.
1116 (memory_address_length): Update.
1117 (ix86_issue_rate): Move to x86-tune-sched.c.
1118 (ix86_flags_dependent): Move to x86-tune-sched.c.
1119 (ix86_agi_dependent): Move to x86-tune-sched.c.
1120 (exact_dependency_1): Move to x86-tune-sched.c.
1121 (exact_store_load_dependency): Move to x86-tune-sched.c.
1122 (ix86_adjust_cost): Move to x86-tune-sched.c.
1123 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
1124 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
1125 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
1126 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
1127 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
1128 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
1129 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
1130 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
1131 (min_insn_size): Rename to ...
1132 (ix86_min_insn_size): ... this one; export.
1133 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
1134 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
1135 (core2i7_first_cycle_multipass_backtrack): Move to
1136 x86-tune-sched-core.c.
1137 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
1138 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
1139 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
1140 (ix86_avoid_jump_mispredicts): Update.
1141 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
1142 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
1143 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
1144 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
1145 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
1146 (MAX_INSN): Move to ix86-tune-sched-bd.c.
1147 (MAX_IMM): Move to ix86-tune-sched-bd.c.
1148 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
1149 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
1150 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
1151 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
1152 (MAX_STORE): Move to ix86-tune-sched-bd.c.
1153 (BIG): Move to ix86-tune-sched-bd.c.
1154 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
1155 (enum insn_path): Move to ix86-tune-sched-bd.c.
1156 (get_mem_group): Move to ix86-tune-sched-bd.c.
1157 (is_cmp): Move to ix86-tune-sched-bd.c.
1158 (dispatch_violation): Move to ix86-tune-sched-bd.c.
1159 (is_branch): Move to ix86-tune-sched-bd.c.
1160 (is_prefetch): Move to ix86-tune-sched-bd.c.
1161 (init_window): Move to ix86-tune-sched-bd.c.
1162 (allocate_window): Move to ix86-tune-sched-bd.c.
1163 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
1164 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
1165 (process_end_window): Move to ix86-tune-sched-bd.c.
1166 (allocate_next_window): Move to ix86-tune-sched-bd.c.
1167 (find_constant): Move to ix86-tune-sched-bd.c.
1168 (get_num_immediates): Move to ix86-tune-sched-bd.c.
1169 (has_immediate): Move to ix86-tune-sched-bd.c.
1170 (get_insn_path): Move to ix86-tune-sched-bd.c.
1171 (get_insn_group): Move to ix86-tune-sched-bd.c.
1172 (count_num_restricted): Move to ix86-tune-sched-bd.c.
1173 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
1174 (add_insn_window): Move to ix86-tune-sched-bd.c.
1175 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
1176 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
1177 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
1178 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
1179 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
1180 (do_dispatch): Move to ix86-tune-sched-bd.c.
1181 (has_dispatch): Move to ix86-tune-sched-bd.c.
1182 * config/i386/t-i386: Add new object files.
1183 * config/i386/x86-tune-costs.h: New file.
1184 * config/i386/x86-tune-sched-atom.c: New file.
1185 * config/i386/x86-tune-sched-bd.c: New file.
1186 * config/i386/x86-tune-sched-core.c: New file.
1187 * config/i386/x86-tune-sched.c: New file.
1189 2017-10-11 Liu Hao <lh_mouse@126.com>
1191 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
1192 the generic version below instead.
1193 (should_colorize): Recognize Windows consoles as terminals
1195 * pretty-print.c [__MINGW32__] (write_all): New function.
1196 [__MINGW32__] (find_esc_head): Likewise.
1197 [__MINGW32__] (find_esc_terminator): Likewise.
1198 [__MINGW32__] (eat_esc_sequence): Likewise.
1199 [__MINGW32__] (mingw_ansi_fputs): New function that handles
1201 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
1204 2017-10-11 Richard Biener <rguenther@suse.de>
1206 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
1207 Properly call analyze_scalar_evolution with the loop of the stmt.
1209 2017-10-11 Richard Biener <rguenther@suse.de>
1211 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
1212 * tree-core.h (tree_base): Add chrec_var union member.
1213 * tree.h (CHREC_VAR): Remove.
1214 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
1215 * tree-chrec.h (build_polynomial_chrec): Adjust.
1216 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
1217 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
1219 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
1221 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
1222 * match.pd: ... here.
1223 ((T) X == (T) Y): Relax condition.
1225 2017-10-11 Bin Cheng <bin.cheng@arm.com>
1227 PR tree-optimization/82472
1228 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
1230 (break_alias_scc_partitions): Update postorder number.
1232 2017-10-11 Martin Liska <mliska@suse.cz>
1235 * opts.c (parse_no_sanitize_attribute): Do not use error_value
1237 * opts.h (parse_no_sanitize_attribute): Remove last argument.
1239 2017-10-11 Martin Liska <mliska@suse.cz>
1241 * print-rtl.c (print_insn): Move declaration of idbuf
1242 to same scope as name.
1244 2017-10-11 Martin Liska <mliska@suse.cz>
1249 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1252 2017-10-11 Martin Liska <mliska@suse.cz>
1255 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1258 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
1260 * config.gcc (default_gnu_indirect_function): Default to yes for
1261 arm*-*-linux* with glibc.
1263 2017-10-11 Richard Biener <rguenther@suse.de>
1265 * tree-scalar-evolution.c (get_scalar_evolution): Handle
1266 default-defs and types we do not want to analyze.
1267 (interpret_loop_phi): Replace unreachable code with an assert.
1268 (compute_scalar_evolution_in_loop): Remove and inline ...
1269 (analyze_scalar_evolution_1): ... here, replacing condition with
1270 what makes the intent clearer. Remove handling of cases
1271 get_scalar_evolution now handles.
1273 2017-10-10 Jim Wilson <wilson@tuliptree.org>
1275 PR rtl-optimization/81434
1276 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
1277 comment for main loop. In sched_group_found if, also add checks for
1278 pass and min_cost_group.
1280 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
1282 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
1283 (rs6000_insn_cost): New function.
1284 * config/rs6000/rs6000.md (cost): New attribute.
1286 2017-10-10 Jakub Jelinek <jakub@redhat.com>
1287 H.J. Lu <hongjiu.lu@intel.com>
1291 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
1292 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
1293 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
1294 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
1295 to OPTION_MASK_ISA_AVX512VL - builtins that have both
1296 OPTION_MASK_ISA_MMX and some other bit set require both
1297 mmx and the ISAs without the mmx bit.
1298 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
1299 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
1300 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
1301 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
1302 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
1303 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
1304 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
1305 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
1306 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
1307 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
1308 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
1309 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
1310 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
1311 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
1312 Add OPTION_MASK_ISA_MMX.
1314 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
1316 * config.gcc (armv7*-*-freebsd*): New target.
1317 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
1319 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
1321 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
1322 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
1325 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
1327 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
1328 a template parameter.
1329 (WIDE_INT_REF_FOR): Update accordingly.
1330 * tree.h (wi::int_traits <const_tree>): Delete.
1331 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
1332 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
1333 (wi::tree_to_wide_ref): New typedef.
1334 (wi::to_wide): New function.
1335 * calls.c (get_size_range): Use wi::to_wide when operating on
1337 * cgraph.c (cgraph_node::create_thunk): Likewise.
1338 * config/i386/i386.c (ix86_data_alignment): Likewise.
1339 (ix86_local_alignment): Likewise.
1340 * dbxout.c (stabstr_O): Likewise.
1341 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
1342 * expr.c (const_vector_from_tree): Likewise.
1343 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
1344 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
1345 (fold_negate_expr_1, int_const_binop_1, const_binop)
1346 (fold_convert_const_int_from_real, optimize_bit_field_compare)
1347 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
1348 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
1349 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
1350 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
1351 (fold_not_const, round_up_loc): Likewise.
1352 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
1353 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
1354 (alloca_call_type): Likewise.
1355 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
1356 * godump.c (go_output_typedef): Likewise.
1357 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
1358 * internal-fn.c (get_min_precision): Likewise.
1359 * ipa-cp.c (ipcp_store_vr_results): Likewise.
1360 * ipa-polymorphic-call.c
1361 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1362 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
1363 (ipa_modify_call_arguments): Likewise.
1364 * match.pd: Likewise.
1365 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
1366 * print-tree.c (print_node_brief, print_node): Likewise.
1367 * stmt.c (expand_case): Likewise.
1368 * stor-layout.c (layout_type): Likewise.
1369 * tree-affine.c (tree_to_aff_combination): Likewise.
1370 * tree-cfg.c (group_case_labels_stmt): Likewise.
1371 * tree-data-ref.c (dr_analyze_indices): Likewise.
1372 (prune_runtime_alias_test_list): Likewise.
1373 * tree-dump.c (dequeue_and_dump): Likewise.
1374 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
1375 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
1376 * tree-pretty-print.c (dump_generic_node): Likewise.
1377 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
1378 (simple_iv_with_niters): Likewise.
1379 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1380 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
1381 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
1382 * tree-ssa-loop-niter.c (split_to_var_and_offset)
1383 (refine_value_range_using_guard, number_of_iterations_ne_max)
1384 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
1385 (get_cst_init_from_scev, record_nonwrapping_iv)
1386 (scev_var_range_cant_overflow): Likewise.
1387 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
1388 * tree-ssa-pre.c (compute_avail): Likewise.
1389 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1390 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
1391 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
1392 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
1393 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
1394 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
1395 (dump_case_nodes, try_switch_expansion): Likewise.
1396 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
1397 (vect_do_peeling): Likewise.
1398 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
1399 * tree-vect-stmts.c (vectorizable_load): Likewise.
1400 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
1401 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
1402 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
1403 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
1404 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
1405 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
1406 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
1407 (evrp_dom_walker::before_dom_children): Likewise.
1408 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
1409 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
1410 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
1411 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
1412 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
1413 (get_range_pos_neg): Likewise.
1414 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
1415 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
1416 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
1417 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
1418 * config/avr/avr.c (avr_fold_builtin): Likewise.
1419 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
1420 * config/msp430/msp430.c (msp430_attr): Likewise.
1421 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
1422 * config/powerpcspe/powerpcspe-c.c
1423 (altivec_resolve_overloaded_builtin): Likewise.
1424 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
1425 (rs6000_expand_ternop_builtin): Likewise.
1426 * config/rs6000/rs6000-c.c
1427 (altivec_resolve_overloaded_builtin): Likewise.
1428 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
1429 (rs6000_expand_ternop_builtin): Likewise.
1430 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
1432 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1434 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
1435 when copying loop nest with only one inner loop.
1437 2017-10-10 Richard Biener <rguenther@suse.de>
1439 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
1440 blocks if SCEV is active.
1441 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
1443 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
1444 (scev_initialize): Assert we are not yet initialized.
1446 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1448 * tree-loop-distribution.c (generate_loops_for_partition): Remove
1449 inner loop's exit stmt by making it always exit the loop, otherwise
1450 we would generate an infinite empty loop.
1452 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1454 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
1455 renaming variables in new preheader if it's deleted.
1457 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1459 * tree-loop-distribution.c (struct partition): Remove unused field
1460 loops of the structure.
1461 (partition_alloc, partition_free): Ditto.
1462 (build_rdg_partition_for_vertex): Ditto.
1464 2017-10-09 Jeff Law <law@redhat.com>
1466 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
1467 return type to match prototype and documentation.
1469 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1471 * config/rs6000/rs6000.c (processor_costs): Move to ...
1472 * config/rs6000/rs6000.h: ... here.
1473 (rs6000_cost): Declare.
1475 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
1477 * except.c (setjmp_fn): New global variable.
1478 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
1479 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
1480 if DONT_USE_BUILTIN_SETJMP is defined.
1482 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1484 * target.def (insn_cost): New hook.
1485 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
1486 * doc/tm.texi: Regenerate.
1487 * rtlanal.c (insn_cost): Use the new hook.
1489 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1491 * combine.c (combine_validate_cost): Compute the new insn_cost,
1492 not just pattern_cost.
1493 (try_combine): Adjust comment.
1495 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1497 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
1499 * combine.c (uid_insn_cost): Adjust comment.
1500 (combine_validate_cost): Adjust comment. Use pattern_cost instead
1502 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
1503 * dse.c (find_shift_sequence): Ditto.
1504 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
1505 (bb_valid_for_noce_process_p): Use pattern_cost.
1506 * rtl.h (insn_rtx_cost): Delete.
1507 (pattern_cost): New prototype.
1508 (insn_cost): New prototype.
1509 * rtlanal.c (insn_rtx_cost): Rename to...
1510 (pattern_cost): ... this.
1513 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
1515 * config/i386/i386.md (*jcc_2): Remove insn pattern.
1516 (*jcc<mode>_0_r_i387): Ditto.
1517 (*jccxf_r_i387): Ditto.
1518 (*jcc<mode>_r_i387): Ditto.
1519 (*jccu<mode>_r_i387): Ditto.
1520 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
1521 (*jcc): Rename from *jcc_1.
1523 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1525 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
1526 deferred rescans after the lvx/stvx recombination pre-pass.
1528 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1530 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
1531 memory operation instruction support.
1532 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
1533 (rs6000-ibm-aix[789]*): Likewise.
1534 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
1535 Document new functions.
1537 2017-10-09 Richard Biener <rguenther@suse.de>
1539 PR tree-optimization/82397
1540 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
1541 equality only for semantically equal trees.
1543 2017-10-09 Richard Biener <rguenther@suse.de>
1545 PR tree-optimization/82449
1546 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
1547 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
1548 allow constant addresses.
1549 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
1552 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1554 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
1557 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1560 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
1563 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1566 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
1568 2017-10-09 Jakub Jelinek <jakub@redhat.com>
1571 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
1572 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
1574 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
1576 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
1577 (WI_BINARY_PREDICATE_RESULT): Likewise.
1578 (wi::binary_traits::operator_result): New type.
1579 (wi::binary_traits::predicate_result): Likewise.
1580 (generic_wide_int::operator~, unary generic_wide_int::operator-)
1581 (generic_wide_int::operator==, generic_wide_int::operator!=)
1582 (generic_wide_int::operator&, generic_wide_int::and_not)
1583 (generic_wide_int::operator|, generic_wide_int::or_not)
1584 (generic_wide_int::operator^, generic_wide_int::operator+
1585 (binary generic_wide_int::operator-, generic_wide_int::operator*):
1587 (operator~, unary operator-, operator==, operator!=, operator&)
1588 (operator|, operator^, operator+, binary operator-, operator*): New
1590 * expr.c (get_inner_reference): Use wi::bit_and_not.
1591 * fold-const.c (fold_binary_loc): Likewise.
1592 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
1593 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
1594 (bit_value_binop): Likewise.
1595 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
1596 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
1597 (extract_range_from_binary_expr_1): Likewise.
1598 (masked_increment): Likewise.
1599 (simplify_bit_ops_using_ranges): Likewise.
1601 2017-10-09 Martin Jambor <mjambor@suse.cz>
1604 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
1605 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
1606 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
1608 (hsa_fixup_mov_insn_type): New function.
1609 (hsa_op_with_type::get_in_type): Use it.
1610 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
1611 immediates in an assert.
1612 (hsa_op_with_type::extend_int_to_32bit): New method.
1613 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
1614 types. Convert to dest type if necessary.
1615 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
1616 (reg_for_gimple_ssa): Pass false as min32int to
1617 hsa_type_for_scalar_tree_type.
1618 (gen_hsa_addr): Fixup type when creating addresable temporary.
1619 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
1620 (gen_hsa_unary_operation): Extend operands and convert to dest type if
1621 necessary. Call hsa_fixup_mov_insn_type.
1622 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
1623 extend operands and convert to dest type if necessary.
1624 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
1625 to dest type if necessary.
1626 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
1627 if conversion nt necessary and size matches.
1628 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
1629 to dest type if necessary.
1630 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
1631 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
1633 (gen_hsa_clrsb): Likewise.
1634 (gen_hsa_ffs): Likewise.
1635 (gen_hsa_divmod): Extend operands and convert to dest type if
1637 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
1639 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1641 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
1642 Remove empty default arguments. Use a brace block as output
1644 (conditional return): Ditto.
1646 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
1647 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
1649 (group_ending_nop): Ditto.
1650 (doloop_end): Ditto.
1651 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
1652 (splitters for those): Ditto.
1654 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1656 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
1657 a conditional jump (and the compare for it) so that pc_rtx is the
1659 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
1660 for the deleted and renamed ctr<mode>_internal[234] patterns.
1661 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
1662 Delete second conditional return pattern.
1663 (ctr<mode>_internal2): Delete this second bdnz pattern.
1664 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
1665 (ctr<mode>_internal4): Delete this second bdz pattern.
1667 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1669 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
1670 (always_initialized_rtx_for_ssa_name_p): New predicate.
1671 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
1672 (finish_out_of_ssa): Free new field of SA.
1673 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
1674 * tree-ssa-coalesce.c: Include tree-ssa.h.
1675 (get_parm_default_def_partitions): Remove extern keyword.
1676 (get_undefined_value_partitions): New function.
1677 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
1678 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
1680 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
1682 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1684 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
1686 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1688 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
1689 for targets that preffer 128bit.
1691 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1693 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
1695 2017-10-08 Olivier Hainque <hainque@adacore.com>
1697 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
1698 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
1700 (thumb_set_return_address): Likewise.
1702 2017-10-08 Olivier Hainque <hainque@adacore.com>
1704 * common/config/arm/arm-common.c (arm_except_unwind_info):
1705 Handle DWARF2_UNWIND_INFO.
1707 2017-10-07 Michael Collison <michael.collison@arm.com>
1709 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
1712 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
1714 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
1715 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
1716 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
1717 defined, force the creation of a new block for a dispatch label.
1719 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
1721 * invoke.texi (Wsuggest-attribute=cold): Document.
1722 * common.opt (Wsuggest-attribute=cold): New
1723 * ipa-pure-const.c (warn_function_cold): New function.
1724 * predict.c (compute_function_frequency): Use it.
1725 * predict.h (warn_function_cold): Declare.
1727 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
1729 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
1732 2017-10-06 Martin Liska <mliska@suse.cz>
1734 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
1735 keyword for member functions.
1736 (struct sanopt_tree_couple): New struct.
1737 (struct sanopt_tree_couple_hash): New function.
1738 (struct sanopt_ctx): Add new hash_map.
1739 (has_dominating_ubsan_ptr_check): New function.
1740 (record_ubsan_ptr_check_stmt): Likewise.
1741 (maybe_optimize_ubsan_ptr_ifn): Likewise.
1742 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
1743 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
1745 2017-10-06 Sudakshina Das <sudi.das@arm.com>
1748 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
1749 aarch64_simd_valid_immediate on CONST_VECTORs.
1750 (aarch64_reg_or_bic_imm): Likewise.
1752 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
1754 PR rtl-optimization/82396
1755 * haifa-sched.c (ready_sort_real): Disable qsort checking.
1757 2017-10-06 Sebastian Pop <sebpop@gmail.com>
1759 * graphite-dependences.c (scop_get_reads): Move code to...
1760 (scop_get_must_writes): Move code to...
1761 (scop_get_may_writes): Move code to...
1762 (scop_get_reads_and_writes): ... here.
1763 (scop_get_dependences): Call scop_get_reads_and_writes.
1765 2017-10-06 Jakub Jelinek <jakub@redhat.com>
1767 PR tree-optimization/82434
1768 * fold-const.h (can_native_encode_type_p,
1769 can_native_encode_string_p): Remove.
1770 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
1771 don't encode anything, just return what would be otherwise returned.
1772 (native_encode_fixed, native_encode_complex, native_encode_vector):
1774 (native_encode_string): Likewise. Inline by hand
1775 can_native_encode_string_p.
1776 (can_native_encode_type_p): Remove.
1777 (can_native_encode_string_p): Remove.
1778 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
1779 STRING_CSTs using can_native_encode_string_p, test all
1780 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
1781 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
1782 argument from native_encode_expr.
1783 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
1784 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
1787 2017-10-06 Richard Biener <rguenther@suse.de>
1789 PR tree-optimization/82397
1790 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
1791 operand_equal_p but rely on data_ref_compare_tree for detecting
1793 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
1794 to match up with dr_group_sort_cmp.
1796 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1799 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
1801 * config/s390/s390-builtin-types.def: Regenerate.
1803 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1806 * config/s390/s390-builtin-types.def: Regenerate.
1807 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
1808 Change flag from B_VXE to B_VX.
1809 (s390_vec_min_dbl): Remove B_VXE flag.
1811 2017-10-06 Richard Biener <rguenther@suse.de>
1813 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
1814 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
1815 translate_isl_ast_to_gimple::is_valid_rename,
1816 translate_isl_ast_to_gimple::get_rename,
1817 translate_isl_ast_to_gimple::get_def_bb_for_const,
1818 translate_isl_ast_to_gimple::get_new_name,
1819 translate_isl_ast_to_gimple::collect_all_ssa_names,
1820 translate_isl_ast_to_gimple::copy_loop_phi_args,
1821 translate_isl_ast_to_gimple::collect_all_ssa_names,
1822 translate_isl_ast_to_gimple::copy_loop_phi_args,
1823 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
1824 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
1825 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
1826 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
1827 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
1828 translate_isl_ast_to_gimple::copy_cond_phi_args,
1829 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
1830 translate_isl_ast_to_gimple::edge_for_new_close_phis,
1831 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
1832 translate_isl_ast_to_gimple::rename_uses,
1833 translate_isl_ast_to_gimple::rename_all_uses): Remove.
1834 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
1835 (set_rename_for_each_def): Likewise.
1836 (graphite_copy_stmts_from_block): Handle debug stmt resetting
1837 here. Handle rewriting SCEV analyzable uses here.
1838 (copy_bb_and_scalar_dependences): Generate code for PHI
1840 (graphite_regenerate_ast_isl): Adjust.
1841 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
1842 (add_write, add_read): New functions.
1843 (build_cross_bb_scalars_def): Use it and simplify.
1844 (build_cross_bb_scalars_use): Likewise.
1845 (graphite_find_cross_bb_scalar_vars): Inline into...
1846 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
1847 simulating out-of-SSA. Compute liveout and add dependencies.
1848 (build_scops): Force an empty entry block.
1849 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
1851 (sese_build_liveouts): Declare.
1852 (sese_trivially_empty_bb_p): Likewise.
1853 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
1854 compute liveout and debug_liveout.
1855 (sese_bad_liveouts_use): Remove.
1856 (sese_reset_debug_liveouts_bb): Likewise.
1857 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
1858 (sese_build_liveouts): Build liveout and debug_liveout and store
1860 (new_sese_info): Adjust.
1861 (free_sese_info): Likewise.
1862 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
1863 do not build liveout here.
1864 (move_sese_in_condition): Adjust region entry.
1865 (scev_analyzable_p): Match up with chrec_apply requirements.
1866 (sese_trivially_empty_bb_p): New.
1867 * tree-into-ssa.c (get_reaching_def): Properly support generating
1868 default-defs for incremental rewrite of anonymous names.
1870 2017-10-06 Richard Biener <rguenther@suse.de>
1872 * graphite-sese-to-poly.c (extract_affine): For casts increasing
1873 precision do not perform modulo reduction.
1875 2017-10-06 Richard Biener <rguenther@suse.de>
1877 PR tree-optimization/82436
1878 * tree-vect-slp.c (vect_supported_load_permutation_p): More
1879 conservatively choose the vectorization factor when checking
1880 whether we can perform the required load permutation.
1881 (vect_transform_slp_perm_load): Assert when we may not fail.
1883 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1885 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
1886 message for incompatible -msdata=* and -mcall-* options.
1888 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1890 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
1891 rate for post-reload scheduling.
1893 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1895 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
1897 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1899 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
1900 to improve monte carlo in scimark.
1902 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1904 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
1905 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
1906 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
1907 pentium4_cost, nocona_cost): Set reassociation width to 1.
1908 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
1909 width to 2 for fp operations and 1 otherwise.
1910 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
1912 (atom_cost): Set reassociation width to 2.
1913 (slm_cost, generic_cost): Set fp reassociation width
1914 to 2 and 1 otherwise.
1915 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
1916 (core_cost): Set fp reassociation width to 4 and vector to 2.
1917 (ix86_reassociation_width): Rewrite using cost table; special case
1918 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
1919 and TARGET_AVX128_OPTIMAL.
1920 * config/i386/i386.h (processor_costs): Add
1921 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
1922 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
1923 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
1924 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
1925 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
1926 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
1928 2017-10-05 Nathan Sidwell <nathan@acm.org>
1930 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
1932 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1934 * config/arm/arm.c (arm_test_fpu_data): New.
1935 (arm_run_selftests): Call arm_test_fpu_data.
1937 2017-10-04 Nathan Sidwell <nathan@acm.org>
1939 * toplev.c (toplev::main): Remove excess parens on pretty_printer
1941 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
1943 2017-10-04 Sudakshina Das <sudi.das@arm.com>
1945 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
1946 check type for aarch64_simd_valid_immediate.
1947 (aarch64_output_simd_mov_immediate): Update prototype.
1948 (aarch64_simd_valid_immediate): Update prototype.
1949 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
1950 support for ORR-immediate.
1951 (and<mode>3): modified pattern to add support for BIC-immediate.
1952 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
1953 now checks for valid immediate for BIC and ORR based on new enum
1955 (aarch64_output_simd_mov_immediate): Function now used to output
1956 BIC/ORR imm as well based on new enum argument.
1957 * config/aarch64/constraints.md (Do): New vector immediate constraint.
1959 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
1960 (aarch64_reg_or_bic_imm): Likewise.
1962 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1964 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
1965 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
1967 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1971 PR rtl-optimization/82396
1972 * haifa-sched.c (autopref_multipass_init): Simplify
1974 (autopref_rank_data): Simplify sort order.
1975 * sched-int.h (autopref_multipass_data_): Remove
1976 multi_mem_insn_p, min_offset and max_offset.
1978 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1980 * doc/sourcebuild.texi: Document vect_peeling_profitable.
1982 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1984 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
1987 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1989 * doc/sourcebuild.texi: Document vect_long_mult.
1991 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
1993 PR tree-optimization/82413
1994 * fold-const.c (build_range_check): Use widest_int when comparing
1995 the maximum ETYPE value with HIGH.
1997 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1999 PR rtl-optimization/82396
2000 * haifa-sched.c (autopref_multipass_init): Simplify
2002 (autopref_rank_data): Simplify sort order.
2003 * sched-int.h (autopref_multipass_data_): Remove
2004 multi_mem_insn_p, min_offset and max_offset.
2006 2017-10-04 Jakub Jelinek <jakub@redhat.com>
2008 PR tree-optimization/82381
2009 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
2010 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
2013 PR tree-optimization/82374
2014 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
2015 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
2016 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
2017 current_function_decl to the new decl.
2019 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
2021 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
2022 helper macro for IEEE float128 hardware built-in functions.
2023 (SQRTF128_ODD): Add built-in functions with the round-to-odd
2025 (TRUNCF128_ODD): Likewise.
2026 (ADDF128_ODD): Likewise.
2027 (SUBF128_ODD): Likewise.
2028 (MULF128_ODD): Likewise.
2029 (DIVF128_ODD): Likewise.
2030 (FMAF128_ODD): Likewise.
2031 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
2032 UNSPEC_TRUNC_ROUND_TO_ODD.
2033 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
2034 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
2035 floating point round to odd instructions.
2036 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
2037 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
2038 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
2039 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
2040 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
2041 (trunc<mode>sf2_hw): Change the truncate with round to odd
2042 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
2043 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
2044 to odd hardware instructions.
2045 (sub<mode>3_odd): Likewise.
2046 (mul<mode>3_odd): Likewise.
2047 (div<mode>3_odd): Likewise.
2048 (sqrt<mode>2_odd): Likewise.
2049 (fma<mode>4_odd): Likewise.
2050 (fms<mode>4_odd): Likewise.
2051 (nfma<mode>4_odd): Likewise.
2052 (nfms<mode>4_odd): Likewise.
2053 (trunc<mode>df2_odd): Change the truncate with round to odd
2054 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
2056 * doc/extend.texi (PowerPC built-in functions): Update documentation
2057 for existing IEEE float128-bit built-in functions. Add built-in
2058 functions that generate the IEEE 128-bit floating point round to
2061 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
2063 PR rtl-optimization/77729
2064 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
2065 to (X&(C1&~C2))|C2 transformations.
2067 2017-10-03 Martin Jambor <mjambor@suse.cz>
2069 PR tree-optimization/82363
2070 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
2071 mismatch, mark lacc written regardless of racc.
2073 2017-10-03 Jakub Jelinek <jakub@redhat.com>
2075 PR tree-optimization/82381
2076 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
2077 stmt_to_insert nor wheather SSA_NAMEs are default defs.
2078 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
2079 fallthrough into reassoc_stmt_dominates_stmt_p.
2082 * combine.c (combine_instructions): Don't combine in unreachable
2085 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2088 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
2089 function to not use the have_cpu variable. Do not set cpu_index,
2090 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
2092 (rs6000_valid_attribute_p): Remove duplicate initializations of
2093 old_optimize and func_optimize.
2094 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
2095 (rs6000_activate_target_options): Make global.
2096 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
2099 2017-10-02 Jakub Jelinek <jakub@redhat.com>
2101 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
2102 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
2103 Set *poffset to 0 and *psize and *pmax_size to -1 if
2104 *poffset + *psize overflows in HOST_WIDE_INT.
2106 PR tree-optimization/82387
2107 PR tree-optimization/82388
2108 PR tree-optimization/82389
2109 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
2110 instead of live_bytes non-NULL.
2112 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
2115 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
2118 2017-10-02 Richard Biener <rguenther@suse.de>
2120 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
2121 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
2122 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
2124 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2126 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
2127 requested precision matches the type's.
2128 * calls.c (alloc_max_size): Calculate the new candidate size as
2129 a widest_int and use wi::to_widest when comparing it with the
2130 current candidate size.
2131 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
2132 zero rather than integer_zero_node.
2133 * match.pd: Check for a no-op conversion before using wi::add
2134 rather than after. Use tree_to_uhwi when summing small shift
2135 counts into an unsigned int.
2137 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2138 Alan Hayward <alan.hayward@arm.com>
2139 David Sherwood <david.sherwood@arm.com>
2142 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
2143 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
2144 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
2145 POINTER_AND_FP_REGS.
2147 2017-10-02 Richard Biener <rguenther@suse.de>
2149 PR tree-optimization/82355
2150 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2151 a mapping for the enclosing loop but avoid generating one for
2153 (copy_bb_and_scalar_dependences): Remove premature codegen
2154 error on PHIs in blocks duplicated into multiple places.
2155 * graphite-scop-detection.c
2156 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2157 in the region use it as loop and nest to analyze the DR in.
2158 (try_generate_gimple_bb): Likewise.
2159 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2160 (add_loop_constraints): For blocks in a loop not in the region
2161 create a dimension with a single iteration.
2162 * sese.h (gbb_loop_at_index): Remove assert.
2164 2017-10-01 Kevin Buettner <kevinb@redhat.com>
2166 * omp-expand.c (adjust_context_scope): New function.
2167 (expand_parallel_call): Call adjust_context_scope.
2169 2017-10-01 Jeff Law <law@redhat.com>
2171 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
2172 dom_opt_dom_walker class with direct access to private members.
2173 Add comments. Call test_for_singularity.
2174 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
2175 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
2176 m_dummy_cond anymore.
2177 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
2179 (pass_dominator:execute): Build the dummy_cond here and pass it
2180 to the dom_opt_dom_walker ctor.
2181 (test_for_singularity): New function.
2183 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
2184 Maya Rashish <coypu@sdf.org>
2186 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
2187 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
2188 (alpha*-*-netbsd*) Use nbsd_tm_file.
2189 (arm*-*-netbsdelf*) Likewise.
2190 (i[34567]86-*-netbsdelf*) Likewise.
2191 (x86_64-*-netbsd*) Likewise.
2192 (mips*-*-netbsd*) Likewise.
2193 (powerpc-*-netbsd*) Likewise.
2194 (sh*-*-netbsd*) Likewise.
2195 (sparc-*-netbsdelf*) Likewise.
2196 (sparc64-*-netbsd*) Likewise.
2197 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
2199 (vax-*-netbsdelf*) Likewise.
2200 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
2201 (UINT_FAST8_TYPE) Likewise.
2202 (INT_FAST16_TYPE) Check CHAR_FAST16.
2203 (UINT_FAST16_TYPE) Likewise.
2205 2017-09-30 Jakub Jelinek <jakub@redhat.com>
2208 * config/i386/i386.md
2209 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
2210 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
2211 *divmodsi4_zext_2): New define_insn_and_split.
2212 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
2213 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
2214 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
2215 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
2216 New define_insn_and_split.
2217 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
2218 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
2219 operands[1] having DImode when mode is SImode.
2221 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
2222 always SImode for DIV and MOD in REG_EQUAL notes.
2224 2017-09-29 Yury Gribov <tetra2005@gmail.com>
2227 * match.pd: Fix handling of NaNs in pattern.
2229 2017-09-29 Jeff Law <law@redhat.com>
2231 * sbitmap.c (bitmap_bit_in_range_p): New function.
2232 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
2233 * tree-ssa-dse.c (live_bytes_read): New function.
2234 (dse_classify_store): Ignore reads of dead bytes.
2236 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
2237 typos and whitespace errors.
2238 * config/i386/predicates.md (address_no_seg_operand): Likewise.
2239 * config/s390/s390.c (s390_emit_prologue): Likewise.
2241 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2244 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
2245 with a symbol for LRA.
2247 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2249 PR rtl-optimization/82338
2250 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
2252 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
2254 * genmodes.c (calc_wider_mode): Suppress qsort macro.
2255 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
2256 (qsort_chk): Declare.
2257 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
2258 (qsort_chk): New function.
2260 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2262 PR tree-optimization/82337
2263 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
2264 phi definition if the PHI result appears in an abnormal PHI.
2265 (find_basis_for_base_expr): Don't record a basis if the LHS of the
2266 basis appears in an abnormal PHI.
2268 2017-09-29 Richard Biener <rguenther@suse.de>
2270 * graphite-isl-ast-to-gimple.c
2271 (translate_isl_ast_to_gimple::set_codegen_error): New function.
2272 (binary_op_to_tree): Use it.
2273 (get_rename_from_scev): Likewise.
2274 (copy_loop_phi_nodes): Likewise.
2275 (copy_bb_and_scalar_dependences): Likewise.
2276 (translate_pending_phi_nodes): Likewise.
2278 2017-09-29 Jakub Jelinek <jakub@redhat.com>
2281 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
2282 for movabsq $(i32 << shift), r64.
2284 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
2286 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
2287 index when encoding %esp as %rsp to avoid 0x67 prefix.
2289 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
2291 * config/i386/i386.md (*movsf_internal, *movdf_internal):
2292 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
2294 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
2296 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
2297 Extensions with more than 16 double VFP registers.
2298 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
2299 to_clear_mask and all code related to it. Replace the remaining
2300 entry by a sbitmap and adapt code accordingly.
2302 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
2304 * brig-builtins.def: Change pure attributes to const.
2306 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2308 * config.gcc (default_gnu_indirect_function): Default to yes for
2309 sparc*-*-linux* with glibc.
2311 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2313 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
2314 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
2315 when creating .init_array and .fini_array sections with priority
2318 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
2321 * config/aarch64/aarch64.c
2322 (aarch64_builtin_support_vector_misalignment): Always return false
2323 when misalignment is unknown.
2325 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
2327 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
2328 this function to return false if the definition used by the swap
2329 instruction is artificial, or if the memory address from which the
2330 constant value is loaded is not represented by a base address held
2331 in a register or if the base address register is a frame or stack
2332 pointer. Additionally, return false if the base address of the
2333 loaded constant is a SYMBOL_REF but is not considered to be a
2335 (replace_swapped_load_constant): New function.
2336 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
2337 loaded constant vector with a load of a swapped constant vector.
2339 2017-09-27 Carl Love <cel@us.ibm.com>
2341 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
2342 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
2343 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
2346 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
2348 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
2349 first, always call autopref_rank_data otherwise.
2351 2017-09-27 Richard Biener <rguenther@suse.de>
2353 * graphite-scop-detection.c (find_scop_parameters): Move
2354 loop bound handling ...
2355 (gather_bbs::before_dom_children): ... here, avoiding the need
2356 to build scop_info->loop_nest.
2357 (record_loop_in_sese): Remove.
2358 * sese.h (sese_info_t::loop_nest): Remove.
2359 * sese.c (new_sese_info): Do not allocate loop_nest.
2360 (free_sese_info): Do not free loop_nest.
2362 2017-09-27 Jakub Jelinek <jakub@redhat.com>
2365 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
2366 lhs from calls if the lhs has addressable type.
2368 2017-09-27 Richard Biener <rguenther@suse.de>
2370 * graphite.h (scop::max_alias_set): New member.
2371 * graphite-scop-detection.c: Remove references to non-existing
2372 --param in comments.
2373 (build_alias_sets): Record the maximum alias set used for drs.
2374 (build_scops): Support zero as unlimited for
2375 --param graphite-max-arrays-per-scop.
2376 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
2378 (build_poly_sr_1): ... here. Compute alias set based on the
2379 maximum alias set used for drs rather than
2380 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
2382 2017-09-27 Richard Biener <rguenther@suse.de>
2384 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
2385 --param loop-block-tile-size=0 to disable tiling.
2387 2017-09-27 Richard Biener <rguenther@suse.de>
2389 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
2390 (graphite-max-nb-scop-params): Document special value zero.
2391 * domwalk.h (dom_walker::STOP): New symbolical constant.
2392 (dom_walker::dom_walker): Add optional parameter for bb to
2394 (dom_walker::~dom_walker): Declare.
2395 (dom_walker::before_dom_children): Document STOP return value.
2396 (dom_walker::m_user_bb_to_rpo): New member.
2397 (dom_walker::m_bb_to_rpo): Likewise.
2398 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
2399 mapping here if not provided by the user.
2400 (dom_walker::~dom_walker): Free bb to RPO mapping if not
2401 provided by the user.
2402 (dom_walker::STOP): Define.
2403 (dom_walker::walk): Do not compute bb to RPO mapping here.
2404 Support STOP return value from before_dom_children to stop
2406 * graphite-optimize-isl.c (optimize_isl): If the schedule
2407 is the same still generate code if -fgraphite-identity
2408 or -floop-parallelize-all are given.
2409 * graphite-scop-detection.c: Include cfganal.h.
2410 (gather_bbs::gather_bbs): Get and pass through bb to RPO
2412 (gather_bbs::before_dom_children): Return STOP for BBs
2414 (build_scops): Compute bb to RPO mapping and pass it to
2415 the domwalk. Treat --param graphite-max-nb-scop-params=0
2416 as not limiting the number of params.
2417 * graphite.c (graphite_initialize): Remove limit on the
2418 number of basic-blocks in a function.
2419 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
2420 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
2421 default value of 10.
2423 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2425 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
2426 Adjust code to eliminate needing to do the shift right 32-bits
2427 operation after XSCVDPSPN.
2429 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2431 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
2432 ((X / Y) != 0 -> X >= Y): Likewise.
2434 2017-09-26 Carl Love <cel@us.ibm.com>
2436 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
2437 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
2438 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
2439 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
2440 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
2441 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
2442 definitions and overloading.
2443 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
2444 statement for P9V_BUILTIN_XST_LEN_R.
2445 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
2446 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
2447 define_expand and define_insn for the instructions and builtins.
2448 * doc/extend.texi: Update the built-in documentation file for the new
2450 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
2451 define_insn for the instructions
2453 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
2456 * gcc/config/netbsd-protos.h: New file.
2457 * gcc/config/netbsd.c: New file.
2458 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
2459 * gcc/config/t-netbsd: New file.
2460 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
2461 (tmake_file) Add t-netbsd.
2462 (extra_objs) Add netbsd.o.
2464 2017-09-26 Janus Weil <janus@gcc.gnu.org>
2468 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
2470 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2472 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
2473 sign extension from a vector register to a GPR by doing a 32-bit
2474 direct move and then an EXTSW.
2475 (extendsi<mode>2 splitter): Likewise.
2476 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
2477 right or vector extract after doing XSCVDPSPN. Use
2478 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
2480 (movdi_from_sf_zero_ext): Likewise.
2481 (reload_gpr_from_vsxsf): Likewise.
2482 (p8_mfvsrd_4_disf): Delete, no longer used.
2483 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
2484 then needing to move the SFmode to a GPR to use the XSCVDPSP
2485 instruction instead of FRSP and XSCVDPSPN.
2486 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
2487 it is adjacent to the other XSCVSPDP insns.
2488 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
2489 SFmode to be in traditional Altivec registers.
2490 (vsx_xscvdpspn): Eliminate useless alternative constraint.
2491 (vsx_xscvspdpn): Likewise.
2492 (vsx_xscvspdpn_scalar): Likewise.
2494 2017-09-26 Martin Jambor <mjambor@suse.cz>
2496 * tree-sra.c (compare_access_positions): Put integral types first,
2497 stabilize sorting of integral types, remove conditions putting
2498 non-full-precision integers last.
2499 (sort_and_splice_var_accesses): Disable scalarization if a
2500 non-integert would be represented by a non-full-precision integer.
2502 2017-09-26 Joseph Myers <joseph@codesourcery.com>
2504 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
2505 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
2506 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
2507 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
2508 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
2509 conditionals inside the function instead of around it. Call
2510 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
2511 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
2513 2017-09-26 Richard Biener <rguenther@suse.de>
2515 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
2517 (scop_detection::build_scop_breadth): ... this. Removed.
2518 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
2519 (scop_detection::harmful_stmt_in_bb): Likewise.
2520 (scop_detection::graphite_can_represent_stmt): Likewise.
2521 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
2522 (scop_detection::can_represent_loop): Remove recursion, fold in ...
2523 (scop_detection::can_represent_loop_1): ... this. Removed.
2524 (scop_detection::harmful_loop_in_region): Simplify after inlining
2525 the above and remove more quadraticness.
2526 (build_scops): Adjust.
2527 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
2530 2017-09-26 Jakub Jelinek <jakub@redhat.com>
2533 * config/i386/i386.c (ix86_print_operand_address_as): Only test
2534 REGNO (base) == SP_REG if base is a REG.
2537 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
2538 if it is different SSA_NAME.
2539 (optimize_range_tests_cmp_bitwise): New function.
2540 (optimize_range_tests): Call it.
2542 2017-09-26 Richard Biener <rguenther@suse.de>
2544 PR tree-optimization/82321
2545 * graphite.c (canonicalize_loop_closed_ssa): Properly check
2546 for the def being inside the loop.
2548 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2550 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
2552 * config/s390/s390-builtins.def: Fix constraint on op4.
2554 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2556 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
2557 independent expanders.
2558 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
2559 ("vec_ordered", "vec_unordered"): New expanders.
2561 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2563 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
2566 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2568 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
2569 vec_unpacks_lo_v16qi.
2570 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
2572 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2574 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
2575 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
2576 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
2578 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2580 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
2582 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
2584 ("*vec_slb<mode>"): New insn pattern.
2585 ("vec_shr_<mode>"): New expander.
2586 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
2587 and force the shift count operand to V16QImode.
2588 ("vec_srb<mode>"): Set shift count mode to V16QI.
2590 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2592 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
2593 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
2594 ("vec_widen_smult_hi_<mode>"): New expander definitions.
2596 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
2599 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
2601 2017-09-26 Richard Biener <rguenther@suse.de>
2603 PR tree-optimization/82320
2604 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
2607 2017-09-25 Jeff Law <law@redhat.com>
2609 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
2610 prototype for new argument.
2611 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
2612 mostly extracted from rs6000_emit_allocate_stack.
2613 (rs6000_emit_probe_stack_range_stack_clash): New function.
2614 (rs6000_emit_allocate_stack): Call
2615 rs6000_emit_probe_stack_range_stack_clash as needed.
2616 (rs6000_emit_probe_stack_range): Add additional argument
2617 to call to gen_probe_stack_range{si,di}.
2618 (output_probe_stack_range): New.
2619 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
2620 (output_probe_stack_range_stack_clash): New.
2621 (rs6000_emit_prologue): Emit notes into dump file as requested.
2622 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
2623 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
2624 Add additional operand and pass it to output_probe_stack_range.
2626 2017-09-25 Bin Cheng <bin.cheng@arm.com>
2628 PR tree-optimization/82163
2629 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
2630 (checking_verify_loop_closed_ssa): New parameter.
2631 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
2632 (check_loop_closed_ssa_stmt): Delete.
2633 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
2634 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
2635 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
2638 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
2640 * brig-builtins.def: Treat HSAIL barrier builtins as
2641 setjmp/longjump style functions.
2643 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2645 * target.def (constant_alignment): New hook.
2646 * defaults.h (CONSTANT_ALIGNMENT): Delete.
2647 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
2648 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
2649 * doc/tm.texi: Regenerate.
2650 * targhooks.h (default_constant_alignment): Declare.
2651 (constant_alignment_word_strings): Likewise.
2652 * targhooks.c (default_constant_alignment): New function.
2653 (constant_alignment_word_strings): Likewise.
2654 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
2655 instead of CONSTANT_ALIGNMENT.
2656 * varasm.c (align_variable, get_variable_align, build_constant_desc)
2657 (force_const_mem): Likewise.
2658 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
2659 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
2660 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
2661 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2662 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
2664 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
2665 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2666 constant_alignment_word_strings.
2667 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
2668 (CONSTANT_ALIGNMENT): Likewise.
2669 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2670 (arm_constant_alignment): New function.
2671 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
2672 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2673 constant_alignment_word_strings.
2674 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
2675 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2676 constant_alignment_word_strings.
2677 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
2678 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2679 (cris_constant_alignment): New function.
2680 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
2681 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2682 (epiphany_constant_alignment): New function.
2683 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
2684 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2685 constant_alignment_word_strings.
2686 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
2687 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2688 constant_alignment_word_strings.
2689 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
2690 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2691 constant_alignment_word_strings.
2692 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
2693 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
2694 * config/i386/i386.c (ix86_constant_alignment): Make static.
2695 Use the same interface as the target hook.
2696 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2697 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
2698 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2699 constant_alignment_word_strings.
2700 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
2701 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
2702 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2703 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
2704 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2705 constant_alignment_word_strings.
2706 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
2707 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2708 constant_alignment_word_strings.
2709 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
2710 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2711 constant_alignment_word_strings.
2712 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
2713 * config/microblaze/microblaze.c (microblaze_constant_alignment):
2715 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2716 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
2717 * config/mips/mips.c (mips_constant_alignment): New function.
2718 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2719 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
2720 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
2721 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2722 (mmix_constant_alignment): Make static. Use the same interface
2724 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
2725 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2726 constant_alignment_word_strings.
2727 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
2728 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2729 constant_alignment_word_strings.
2730 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
2731 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2732 constant_alignment_word_strings.
2733 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
2734 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2735 (rs6000_constant_alignment): New function.
2736 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
2737 * config/riscv/riscv.c (riscv_constant_alignment): New function.
2738 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2739 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
2740 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2741 (rs6000_constant_alignment): New function.
2742 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
2743 * config/s390/s390.c (s390_constant_alignment): New function.
2744 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2745 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
2746 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2747 constant_alignment_word_strings.
2748 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
2749 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2750 (sparc_constant_alignment): New function.
2751 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
2752 * config/spu/spu.c (spu_constant_alignment): New function.
2753 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2754 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
2755 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2756 constant_alignment_word_strings.
2757 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
2758 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2759 constant_alignment_word_strings.
2760 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
2761 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2762 constant_alignment_word_strings.
2763 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
2764 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2765 (visium_constant_alignment): New function.
2766 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
2767 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2768 (xtensa_constant_alignment): New function.
2769 * system.h (CONSTANT_ALIGNMENT): Poison.
2771 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
2773 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2774 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
2775 (rs6000_builtin_valid_without_lhs): New helper function.
2776 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2777 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
2779 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2781 * target.h (vec_perm_indices): Use unsigned short rather than
2783 (auto_vec_perm_indices): Likewise.
2784 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
2785 Use unsigned int rather than unsigned char.
2786 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
2788 2017-09-25 Richard Biener <rguenther@suse.de>
2790 * cfgloop.h (sort_sibling_loops): Declare.
2791 * cfgloop.c (sort_sibling_loops_cmp): New helper.
2792 (sort_sibling_loops): New function sorting the sibling loop list
2794 * graphite.c (graphite_transform_loops): Sort sibling loops.
2796 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
2798 * target.def (vec_perm_const_ok): Change sel parameter to
2800 * optabs-query.c (can_vec_perm_p): Update accordingly.
2801 * doc/tm.texi: Regenerate.
2802 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
2803 auto_vec_perm_indices and remove separate nelt field.
2804 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
2805 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
2806 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
2807 (aarch64_expand_vec_perm_const): Update accordingly.
2808 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
2809 to vec_perm_indices.
2810 * config/arm/arm.c (expand_vec_perm_d): Change perm to
2811 auto_vec_perm_indices and remove separate nelt field.
2812 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
2813 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
2814 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
2816 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
2817 to vec_perm_indices.
2818 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
2819 sel to vec_perm_indices.
2820 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
2821 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
2822 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
2824 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
2827 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
2830 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
2831 on the FUNCTION_DECL function context if it has a DIE that is a
2834 2017-09-25 Richard Biener <rguenther@suse.de>
2836 PR tree-optimization/82285
2837 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
2840 2017-09-25 Tom de Vries <tom@codesourcery.com>
2844 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
2847 2017-09-25 Richard Biener <rguenther@suse.de>
2849 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
2850 ISL errors other than isl_error_quota happen. Dump if the
2851 schedule is the same.
2852 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
2853 errors instead of aborting inside ISL.
2855 2017-09-25 Iain Sandoe <iain@codesourcery.com>
2858 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
2859 of libgcc_eh for m64.
2860 * config/i386/darwin64.h: Likewise.
2862 2017-09-25 Richard Biener <rguenther@suse.de>
2865 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
2866 attribute for incomplete types nor twice for complete ones.
2868 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
2871 * config/i386/i386.c (ix86_print_operand_address_as): Encode
2872 %esp as %rsp to avoid 0x67 prefix if there is no index or base
2875 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
2878 * config/i386/i386.opt (mprefer-avx256): Use
2879 ix86_target_flags variable.
2880 * config/i386/i386.c (ix86_target_string): Move
2881 -mprefer-avx256 to flag2_opts.
2883 2017-09-22 Jakub Jelinek <jakub@redhat.com>
2886 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
2887 and x != -1 | y != -1 into (x & y) != -1.
2889 2017-09-22 Steve Ellcey <sellcey@cavium.com>
2891 * config.gcc: Add new case statement to set
2892 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
2893 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
2894 s390x-*-linux* case statements. Added aarch64 to the list of
2895 supported architectures.
2897 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2899 PR tree-optimization/82289
2900 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
2901 STMT_VINFO_RELEVANT_P.
2903 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2904 Alan Hayward <alan.hayward@arm.com>
2905 David Sherwood <david.sherwood@arm.com>
2907 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
2908 for VR_RANGE only; don't allow VR_ANTI_RANGE.
2909 (extract_range_from_binary_expr_1): Don't call
2910 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
2912 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2913 Alan Hayward <alan.hayward@arm.com>
2914 David Sherwood <david.sherwood@arm.com>
2916 * target.def (preferred_vector_alignment): New hook.
2917 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
2919 * doc/tm.texi: Regenerate.
2920 * targhooks.h (default_preferred_vector_alignment): Declare.
2921 * targhooks.c (default_preferred_vector_alignment): New function.
2922 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
2924 (DR_TARGET_ALIGNMENT): New macro.
2925 (aligned_access_p): Update commentary.
2926 (vect_known_alignment_in_bytes): New function.
2927 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
2929 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
2930 Calculate the misalignment based on the target alignment rather than
2932 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
2933 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
2934 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
2935 the target alignment, rather than masking the element misalignment
2936 with the number of elements in a vector. Also use the target
2937 alignment when calculating the maximum number of peels.
2938 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
2939 instead of TYPE_ALIGN_UNIT.
2940 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
2941 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
2942 (vect_create_addr_base_for_vector_ref): Update call accordingly.
2943 (vect_create_data_ref_ptr): Likewise.
2944 (vect_setup_realignment): Realign by ANDing with
2945 -DR_TARGET_MISALIGNMENT.
2946 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
2947 the number of peels based on DR_TARGET_ALIGNMENT.
2948 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
2949 with the guaranteed alignment boundary when deciding whether
2951 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
2952 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
2953 (ensure_base_align): Remove stmt_info parameter. Get the
2954 target base alignment from DR_TARGET_ALIGNMENT.
2955 (vectorizable_store): Update call accordingly. Interpret
2956 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
2958 (vectorizable_load): Likewise.
2960 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2961 Alan Hayward <alan.hayward@arm.com>
2962 David Sherwood <david.sherwood@arm.com>
2964 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
2965 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
2966 (vect_enhance_data_refs_alignment): Likewise.
2968 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2970 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2971 error. Only quit immediately if parsing is complete.
2972 (BEGIN): Initialize fatal_err and parse_done.
2973 (begin fpu, end fpu): Check number of arguments.
2974 (begin arch, end arch): Likewise.
2975 (begin cpu, end cpu): Likewise.
2976 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2977 (optalias): Likewise.
2979 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2981 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2982 * config/arm/arm-isa.h: Delete. Move definitions to ...
2983 * arm-cpus.in: ... here. Use new feature and fgroup values.
2984 * config/arm/arm.c (arm_option_override): Use lower case for feature
2986 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2987 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2988 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2990 (print_isa_bits_for): New function.
2991 (gen_isa): New function.
2992 (gen_comm_data): Use print_isa_bits_for.
2993 (define feature): New keyword.
2994 (define fgroup): New keyword.
2995 * config/arm/t-arm (TM_H): Remove.
2996 (GTM_H): Add arm-isa.h.
2997 (arm-isa.h): Add rule to generate file.
2998 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2999 case for feature bit names.
3001 2017-09-22 Richard Biener <rguenther@suse.de>
3003 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
3005 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
3006 print of no dependency loops ...
3007 * graphite.c (graphite_transform_loops): ... here.
3008 (canonicalize_loop_closed_ssa_form): Work from inner to outer
3010 (same_close_phi_node, remove_duplicate_close_phi,
3011 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
3012 (canonicalize_loop_closed_ssa): ... here and simplify.
3013 * graphite-optimize-isl.c: Include tree-vectorizer.h.
3014 (optimize_isl): Use dump_printf_loc to tell when we stopped
3015 optimizing because of an ISL timeout.
3017 2017-09-22 Richard Biener <rguenther@suse.de>
3019 PR tree-optimization/82291
3020 * tree-if-conv.c (predicate_mem_writes): Make sure to
3021 remove writes in blocks predicated with false.
3023 2017-09-22 Richard Biener <rguenther@suse.de>
3025 * sese.c: Include cfganal.h.
3026 (if_region_set_false_region): Remove.
3027 (create_if_region_on_edge): Likewise.
3028 (move_sese_in_condition): Re-implement without destroying
3031 2017-09-22 Richard Biener <rguenther@suse.de>
3033 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
3034 Verify both BBs contain loop PHI nodes before dispatching to
3036 (graphite_regenerate_ast_isl): Do not recompute dominators,
3037 do not verify three times. Restructure for clarity.
3038 * graphite-scop-detection.c (same_close_phi_node,
3039 remove_duplicate_close_phi, make_close_phi_nodes_unique,
3040 defined_in_loop_p, canonicalize_loop_closed_ssa,
3041 canonicalize_loop_closed_ssa_form): Simplify, remove excess
3042 checking and SSA rewrite, move to ...
3043 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
3044 (graphite_initialize): Do not pass in ctx, do not reset the
3045 SCEV cache, compute only dominators.
3046 (graphite_transform_loops): Allocate ISL ctx after
3047 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
3048 Maintain post-dominators only around build_scops.
3049 * sese.c (if_region_set_false_region): Make static. Free
3050 and recompute dominators.
3051 (move_sese_in_condition): Assert we don't get called with
3052 post-dominators computed.
3053 * sese.h (if_region_set_false_region): Remove.
3055 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
3057 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
3058 mode attribute for TARGET_AVX512VL.
3060 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
3062 * config/i386/i386.opt (mprefer-avx256): New option.
3063 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
3065 (ix86_preferred_simd_mode): Return 256-bit AVX modes
3066 for TARGET_PREFER_AVX256.
3067 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
3069 2017-09-21 Jeff Law <law@redhat.com>
3071 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
3072 Fix dump output if the only stack space is for pushed registers.
3074 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3076 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
3079 2017-09-21 Martin Sebor <msebor@redhat.com>
3082 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
3083 code (in C++) or code that triggers warnings.
3085 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
3087 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
3089 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
3091 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
3092 * sched-rgn.c: Ditto.
3093 * sel-sched-ir.c: Ditto.
3095 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
3097 * toplev.h (set_random_seed): Adjust return type.
3098 * toplev.c (init_local_tick): Move eager initialization of random_seed
3099 to get_random_seed. Adjust comment.
3100 (init_random_seed): Inline to get_random_seed, delete.
3101 (get_random_seed): Initialize random_seed lazily.
3102 (set_random_seed): Do not return previous value.
3103 (print_switch_value): Do not call get_random_seed.
3105 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
3107 * cgraph.c (delete_function_version): New, broken out from...
3108 (cgraph_node::delete_function_version): ...here. Rename to
3109 cgraph_node::delete_function_version_by_decl. Update all uses.
3110 (cgraph_node::remove): Call delete_function_version.
3112 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3115 * tree-inline.c (expand_call_inline): Emit clobber stmts for
3116 VAR_DECLs to which addressable non-volatile parameters are mapped
3117 and for id->retvar after the return value assignment. Clear
3118 id->retval and id->retbnd after inlining.
3120 2017-09-21 Richard Biener <rguenther@suse.de>
3122 PR tree-optimization/82276
3123 PR tree-optimization/82244
3124 * tree-vrp.c (build_assert_expr_for): Set
3125 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
3127 (remove_range_assertions): Revert earlier change.
3129 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
3132 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
3134 2017-09-21 Richard Biener <rguenther@suse.de>
3136 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3137 Restore valid IL after code generation errors.
3138 * graphite.c (graphite_transform_loops): Diagnose code
3139 generation issues as MSG_MISSED_OPTIMIZATION and continue
3140 with processing SCOPs.
3142 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3143 Alan Hayward <alan.hayward@arm.com>
3144 David Sherwood <david.sherwood@arm.com>
3146 * calls.c (compute_argument_addresses): Use simplify_gen_binary
3147 rather than choosing between plus_constant and gen_rtx_<CODE>.
3148 * expr.c (emit_push_insn): Likewise.
3149 (expand_expr_real_2): Likewise.
3151 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3152 Alan Hayward <alan.hayward@arm.com>
3153 David Sherwood <david.sherwood@arm.com>
3155 * loop-unroll.c (split_iv): Call copy_rtx on the step.
3157 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3158 Alan Hayward <alan.hayward@arm.com>
3159 David Sherwood <david.sherwood@arm.com>
3161 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
3162 calling tree_to_uhwi.
3164 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3165 Alan Hayward <alan.hayward@arm.com>
3166 David Sherwood <david.sherwood@arm.com>
3168 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
3169 INTEGER_CST rather than a negative test for ADDR_EXPR.
3171 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3172 Alan Hayward <alan.hayward@arm.com>
3173 David Sherwood <david.sherwood@arm.com>
3175 * tree-vrp.c (extract_range_from_binary_expr_1): Check
3176 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
3178 2017-09-21 Richard Biener <rguenther@suse.de>
3180 PR tree-optimization/71351
3181 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
3182 graphite_create_new_loop_guard): Remove, fold remaining parts
3184 (translate_isl_ast_node_for): ... here and simplify.
3186 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3189 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
3190 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
3191 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
3192 alternative always use QI mode, for -Os imov (=R,R) alternative
3193 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
3196 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3197 Jeff Law <law@redhat.com>
3199 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
3200 (allocate_stack_space): New function, partially extracted from
3202 (s390_emit_prologue): Track offset to most recent stack probe.
3203 Code to allocate space moved into allocate_stack_space.
3204 Dump actions when no stack is allocated.
3205 (s390_prologue_plus_offset): New function.
3206 (s390_emit_stack_probe): Likewise.
3208 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
3210 * common.opt (Wa, Wl, Wp, g, gz=): Add
3212 (gno-column-info): Remove.
3213 (gcolumn-info): Drop RejectNegative.
3215 (gno-record-gcc-switches): Remove.
3216 (grecord-gcc-switches): Drop RejectNegative.
3217 (gno-split-dwarf): Remove.
3218 (gsplit-dwarf): Drop RejectNegative.
3219 (gno-strict-dwarf): Remove.
3220 (gstrict-dwarf): Drop RejectNegative.
3221 * config/darwin.opt (gfull, gused): Add RejectNegative.
3222 * dwarf2out.c (gen_producer_string): Drop
3223 gno-record-gcc-switches handler.
3224 * optc-gen.awk: Add g to prefixes with negative forms.
3225 * opts-common.c (remapping_prefix_p): New.
3226 (find_opt): Check it.
3227 (generate_canonical_option): Test g prefix.
3228 (option_map): Add -gno- mapping.
3229 (add_misspelling_candidates): Check remapping_prefix_p.
3231 2017-09-20 Jeff Law <law@redhat.com>
3233 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
3234 thinko in stack clash protection support.
3236 * explow.c (compute_stack_clash_protection_loop_data): Use
3237 CONST_INT_P instead of explicit test. Verify object is a
3238 CONST_INT_P before looking at INTVAL.
3239 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
3240 instead of explicit test.
3242 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3245 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
3246 address instead of to r1 and r11.
3248 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
3250 * config.gcc: Support "knm".
3251 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
3252 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3254 * config/i386/i386.c (m_KNM): Define.
3255 (processor_target_table): Add "knm".
3257 (ix86_option_override_internal): Add "knm".
3258 (ix86_issue_rate): Add PROCESSOR_KNM.
3259 (ix86_adjust_cost): Ditto.
3260 (ia32_multipass_dfa_lookahead): Ditto.
3261 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
3262 (fold_builtin_cpu): Add M_INTEL_KNM.
3263 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
3264 (processor_type): Add PROCESSOR_KNM.
3265 * config/i386/x86-tune.def: Add m_KNM.
3266 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
3268 2017-09-20 Richard Biener <rguenther@suse.de>
3270 PR tree-optimization/80213
3271 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
3272 are allowed in empty BBs as well.
3273 (canonicalize_loop_closed_ssa): Also look for other complex
3275 (scop_detection::get_sese): Include the loop-closed PHI block
3277 (scop_detection::merge_sese): Remove code adding extra blocks.
3278 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
3279 (build_scops): Assert the final returned scop is invalid.
3281 2017-09-20 Richard Biener <rguenther@suse.de>
3283 PR tree-optimization/82264
3284 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
3286 (vn_phi_lookup): Likewise.
3287 (vn_phi_insert): Likewise.
3289 2017-09-20 Jakub Jelinek <jakub@redhat.com>
3291 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
3292 that fits into uhwi or shwi, add DW_AT_const_value regardless
3293 of early_dwarf without going through RTL, using add_AT_unsigned
3296 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
3297 (DEBUG_LTO_ABBREV_SECTION): Likewise.
3298 (DEBUG_LTO_MACINFO_SECTION): Likewise.
3299 (DEBUG_MACRO_SECTION): Likewise.
3300 (DEBUG_LTO_MACRO_SECTION): Likewise.
3301 (DEBUG_STR_DWO_SECTION): Likewise.
3302 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
3303 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
3304 (DEBUG_LTO_DWO_LINE_SECTION): Define.
3305 (DEBUG_LTO_LINE_STR_SECTION): Define.
3306 (init_sections_and_labels): Initialize debug_line_str_section
3307 variable. Initialize debug_loc_section for -gdwarf-5 to
3308 DEBUG_LOCLISTS_SECTION. Formatting fixes.
3310 2017-09-20 Richard Biener <rguenther@suse.de>
3312 * graphite-sese-to-poly.c (extract_affine): Properly handle
3313 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
3315 2017-09-20 Richard Biener <rguenther@suse.de>
3317 PR tree-optimization/81373
3318 * graphite-scop-detection.c (build_cross_bb_scalars_def):
3319 Force SESE live-out defs to be handled even if they are
3322 2017-09-19 Jeff Law <law@redhat.com>
3324 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
3325 nothing for stack adjustments with REG_STACK_CHECK.
3326 * sched-deps.c (parse_add_or_inc): Reject insns with
3327 REG_STACK_CHECK from dependency breaking.
3328 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
3329 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
3330 * reg-notes.def (STACK_CHECK): New note.
3332 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
3333 (ix86_expand_prologue): Dump stack clash info as needed.
3334 Call ix86_adjust_stack_and_probe_stack_clash as needed.
3336 * function.c (dump_stack_clash_frame_info): New function.
3337 * function.h (dump_stack_clash_frame_info): Prototype.
3338 (enum stack_clash_probes): New enum.
3340 * config/alpha/alpha.c (alpha_expand_prologue): Also check
3341 flag_stack_clash_protection.
3342 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
3343 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
3344 (arm_frame_pointer_required): Likewise.
3345 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
3346 (ia64_expand_prologue): Likewise.
3347 * config/mips/mips.c (mips_expand_prologue): Likewise.
3348 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
3349 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3350 (sparc_flat_expand_prologue): Likewise.
3351 * config/spu/spu.c (spu_expand_prologue): Likewise.
3353 * explow.c: Include "params.h".
3354 (anti_adjust_stack_and_probe_stack_clash): New function.
3355 (get_stack_check_protect): Likewise.
3356 (compute_stack_clash_protection_loop_data): Likewise.
3357 (emit_stack_clash_protection_loop_start): Likewise.
3358 (emit_stack_clash_protection_loop_end): Likewise.
3359 (allocate_dynamic_stack_space): Use get_stack_check_protect.
3360 Use anti_adjust_stack_and_probe_stack_clash.
3361 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
3362 (emit_stack_clash_protection_loop_start): Likewise.
3363 (emit_stack_clash_protection_loop_end): Likewise.
3364 * rtl.h (get_stack_check_protect): Prototype.
3365 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
3366 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
3367 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
3369 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
3371 * doc/tm.texi: Rebuilt.
3372 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3373 get_stack_check_protect.
3374 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
3375 * config/arm/arm.c (arm_expand_prologue): Likewise.
3376 (arm_frame_pointer_required): Likewise.
3377 * config/i386/i386.c (ix86_expand_prologue): Likewise.
3378 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
3379 * config/mips/mips.c (mips_expand_prologue): Likewise.
3380 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
3381 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
3382 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3383 (sparc_flat_expand_prologue): Likewise.
3385 * common.opt (-fstack-clash-protection): New option.
3386 * flag-types.h (enum stack_check_type): Note difference between
3387 -fstack-check= and -fstack-clash-protection.
3388 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
3389 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
3390 * toplev.c (process_options): Issue warnings/errors for cases
3391 not handled with -fstack-clash-protection.
3392 * doc/invoke.texi (-fstack-clash-protection): Document new option.
3393 (-fstack-check): Note additional problem with -fstack-check=generic.
3394 Note that -fstack-check is primarily for Ada and refer users
3395 to -fstack-clash-protection for stack-clash-protection.
3396 Document new params for stack clash protection.
3398 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
3400 * config/i386/i386.c (ix86_split_long_move): Do not handle
3401 address used for LEA in a special way.
3403 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3405 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
3407 2017-09-19 Martin Sebor <msebor@redhat.com>
3410 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
3411 of incompatible types.
3413 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
3415 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3416 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
3417 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3418 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
3420 2017-09-19 Richard Biener <rguenther@suse.de>
3422 PR tree-optimization/82244
3423 * tree-vrp.c (remove_range_assertions): Do not propagate
3424 a constant to abnormals but replace the assert with a copy.
3426 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
3428 PR rtl-optimization/57878
3429 PR rtl-optimization/68988
3430 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
3431 avoidance test involving non_reload_pseudos. Move frequency test
3432 below the general fragmentation avoidance test.
3434 2017-09-19 Richard Biener <rguenther@suse.de>
3436 PR tree-optimization/69728
3437 * graphite-sese-to-poly.c (schedule_error): New global.
3438 (add_loop_schedule): Handle empty domain by failing the
3440 (build_original_schedule): Handle schedule_error.
3442 2017-09-19 Richard Biener <rguenther@suse.de>
3444 * graphite-scop-detection.c (scop_detection::can_represent_loop):
3445 Do not iterate to sibling loops but only to siblings of inner
3448 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
3451 * config/m68k/m68k.md (moveq feeding equality comparison): Check
3452 that the registers are different.
3454 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
3456 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
3457 to processor_model and "amdfam17h" to arch_names_table.
3458 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
3460 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3463 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
3465 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3466 Alan Hayward <alan.hayward@arm.com>
3467 David Sherwood <david.sherwood@arm.com>
3469 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
3471 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
3473 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
3474 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
3475 vect_schedule_slp_instance.
3476 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
3477 Update call to vect_slp_analyze_node_operations. Simplify return
3479 (vect_slp_analyze_bb_1): Update call accordingly.
3480 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
3481 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
3482 (vect_schedule_slp): Update call accordingly.
3484 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3485 Alan Hayward <alan.hayward@arm.com>
3486 David Sherwood <david.sherwood@arm.com>
3488 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
3489 with types that aren't in fact scalar.
3491 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3493 * tree-vect-slp.c (vect_record_max_nunits): New function,
3495 (vect_build_slp_tree_1): ...here.
3496 (vect_build_slp_tree_2): Call it for phis too.
3498 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3500 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
3501 to vect_get_vec_def_for_operand when getting the mask operand.
3503 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3504 Alan Hayward <alan.hayward@arm.com>
3505 David Sherwood <david.sherwood@arm.com>
3507 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
3510 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3511 Alan Hayward <alan.hayward@arm.com>
3512 David Sherwood <david.sherwood@arm.com>
3514 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
3515 calculation for vector booleans.
3517 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3518 Alan Hayward <alan.hayward@arm.com>
3519 David Sherwood <david.sherwood@arm.com>
3521 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
3523 (vect_transform_stmt): ...here.
3524 (vect_analyze_stmt): Use it instead of calling
3525 vectorizable_live_operation directly.
3527 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
3529 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
3530 non-SIMT targets in acc vector loops.
3532 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
3534 * configure.ac: Add arc and check if assembler supports gdwarf2.
3535 * configure: Regenerate.
3537 2017-09-18 Richard Biener <rguenther@suse.de>
3539 PR tree-optimization/82220
3540 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
3541 epilogue niters from the min_profitable_iters compute.
3543 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3546 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
3547 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
3548 (ix86_init_pic_reg): Revert 2017-09-01 changes.
3550 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
3553 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
3554 switching to a new text section.
3556 2017-09-18 Richard Biener <rguenther@suse.de>
3558 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
3560 (build_alias_set): Reject aliases with no access function.
3562 2017-09-18 Richard Biener <rguenther@suse.de>
3564 PR tree-optimization/79622
3565 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
3567 (build_cross_bb_scalars_use): Likewise.
3569 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
3571 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
3573 2017-09-18 Alan Modra <amodra@gmail.com>
3576 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
3577 stack_pointer_rtx for count 0. Update comments. Break up
3578 large rtl expression.
3580 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
3582 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
3583 Increase to 20 bytes.
3584 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
3585 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
3586 or avx version of the stub.
3588 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
3591 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
3592 compute the minimum stack alignment. Also update preferred stack
3593 boundary for leaf functions.
3595 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
3597 PR tree-optimization/82228
3598 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
3601 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3603 * common/config/nds32/nds32-common.c
3604 (nds32_option_optimization_table): Refine formatting.
3605 (nds32_option_optimization_table): Use -fsched-pressure and
3606 -fomit-frame-pointer for specific optimization level.
3608 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3610 * config/nds32/nds32.c: Refine formatting and comments.
3611 * config/nds32/nds32.h: Likewise.
3612 * config/nds32/nds32.md: Likewise.
3613 * config/nds32/nds32-cost.c: Likewise.
3614 * config/nds32/nds32-isr.c: Likewise.
3615 * config/nds32/nds32-md-auxiliary.c: Likewise.
3616 * config/nds32/nds32-multiple.md: Likewise.
3617 * config/nds32/nds32-predicates.c: Likewise.
3619 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
3620 Jakub Jelinek <jakub@redhat.com>
3622 Add support for -std=c++2a.
3623 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
3625 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
3627 2017-09-15 Steve Ellcey <sellcey@cavium.com>
3630 * doc/extend.texi (Common Function Attributes): Add
3631 references to ARM, AArch64, and S/390 specific attributes.
3632 (Function Specific Option Pragmas): Add AArch64 and S/390
3633 to list of back ends that support the target pragma.
3635 2017-09-15 Nathan Sidwell <nathan@acm.org>
3637 * doc/standards.texi: Fix C++17 description. Update URLs for
3640 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
3642 * common.opt (Wcast-align=strict): New warning option.
3643 * doc/invoke.texi: Document -Wcast-align=strict.
3645 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
3647 * cgraph.h (cgraph_thunk_info): Add comments.
3648 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
3649 assert for VIRTUAL_* arguments stricter.
3651 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
3653 PR tree-optimization/71026
3654 * match.pd: Move RDIV patterns from fold-const.c
3655 * fold-const.c (distribute_real_division): Removed.
3656 (fold_binary_loc): Remove calls to distribute_real_divison.
3658 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3660 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
3661 c++1z and gnu++1z as deprecated. Change other references to
3662 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
3663 Change -Wc++1z-compat to -Wc++17-compat.
3664 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
3665 * dwarf2out.c (highest_c_language): Handle C++17.
3666 (gen_compile_unit_die): Likewise.
3668 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3670 PR rtl-optimization/82192
3671 * combine.c (make_extraction): Don't look through non-paradoxical
3672 SUBREGs or TRUNCATE if pos + len is or might be bigger than
3675 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3676 Alan Hayward <alan.hayward@arm.com>
3677 David Sherwood <david.sherwood@arm.com>
3679 * target.def (function_arg_offset): New hook.
3680 * targhooks.h (default_function_arg_offset): Declare.
3681 * targhooks.c (default_function_arg_offset): New function.
3682 * function.c (locate_and_pad_parm): Use
3683 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
3684 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
3685 (TARGET_FUNCTION_ARG_OFFSET): ...this.
3686 * doc/tm.texi: Regenerate.
3687 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
3688 * config/spu/spu.c (spu_function_arg_offset): New function.
3689 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
3690 * system.h (FUNCTION_ARG_OFFSET): Poison.
3692 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3693 Alan Hayard <alan.hayward@arm.com>
3694 David Sherwood <david.sherwood@arm.com>
3696 * target.def (truly_noop_truncation): New hook.
3697 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
3698 than TRULY_NOOP_TRUNCATION.
3699 * hooks.h (hook_bool_uint_uint_true): Declare.
3700 * hooks.c (hook_bool_uint_uint_true): New function.
3701 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
3702 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
3703 * doc/tm.texi: Regenerate.
3704 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
3705 rather than TRULY_NOOP_TRUNCATION in comments.
3706 (simplify_comparison): Likewise.
3707 (record_truncated_value): Likewise.
3708 * expmed.c (extract_bit_field_1): Likewise.
3709 (extract_split_bit_field): Likewise.
3710 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
3711 instead of TRULY_NOOP_TRUNCATION.
3712 * function.c (assign_parm_setup_block): Likewise.
3713 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
3714 * rtlhooks.c: Include target.h.
3715 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
3716 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
3717 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
3718 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
3719 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
3720 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
3721 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
3722 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
3723 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
3724 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
3725 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
3726 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
3727 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
3728 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
3729 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
3730 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
3731 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
3732 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
3733 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
3734 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
3735 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
3736 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
3737 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
3738 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
3739 * config/mips/mips.c (mips_truly_noop_truncation): New function.
3740 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3741 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3742 rather than TRULY_NOOP_TRUNCATION in comments.
3743 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
3744 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
3745 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
3746 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
3747 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
3748 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
3749 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
3750 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
3751 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
3752 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
3753 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
3754 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3755 rather than TRULY_NOOP_TRUNCATION in comments.
3756 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
3757 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
3758 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
3759 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
3760 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
3761 TRULY_NOOP_TRUNCATION condition.
3762 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
3763 (TRULY_NOOP_TRUNCATION): Delete.
3764 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
3765 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
3766 * config/spu/spu.c (spu_truly_noop_truncation): New function.
3767 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3768 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
3769 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
3770 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
3771 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3772 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3773 rather than TRULY_NOOP_TRUNCATION in comments.
3774 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
3775 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
3776 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
3777 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
3778 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
3779 * system.h (TRULY_NOOP_TRUNCATION): Poison.
3781 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
3784 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
3785 (*cmp_ior): Likewise.
3786 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
3787 (*ior_scc_scc_cmp): Likewise.
3788 (*and_scc_scc): Likewise.
3789 (*and_scc_scc_cmp): Likewise.
3791 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3792 Alan Hayard <alan.hayward@arm.com>
3793 David Sherwood <david.sherwood@arm.com>
3795 * target.def (can_change_mode_class): New hook.
3796 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
3797 (hard_regno_nregs): Likewise.
3798 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
3799 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
3800 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
3801 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
3802 (LOAD_EXTEND_OP): Update accordingly.
3803 * doc/tm.texi: Regenerate.
3804 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
3805 CANNOT_CHANGE_MODE_CLASS.
3806 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
3807 (REG_CAN_CHANGE_MODE_P): ...this new macro.
3808 * combine.c (simplify_set): Update accordingly.
3809 * emit-rtl.c (validate_subreg): Likewise.
3810 * recog.c (general_operand): Likewise.
3811 * regcprop.c (mode_change_ok): Likewise.
3812 * reload1.c (choose_reload_regs): Likewise.
3813 (inherit_piecemeal_p): Likewise.
3814 * rtlanal.c (simplify_subreg_regno): Likewise.
3815 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
3816 instead of CANNOT_CHANGE_MODE_CLASS.
3817 (reload_cse_simplify_operands): Likewise.
3818 * reload.c (push_reload): Use targetm.can_change_mode_class
3819 instead of CANNOT_CHANGE_MODE_CLASS.
3820 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
3821 REG_CANNOT_CHANGE_MODE_P.
3822 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3823 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
3824 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3825 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3826 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3827 (arm_can_change_mode_class): New function.
3828 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3829 than CANNOT_CHANGE_MODE_CLASS in comments.
3830 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3831 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
3832 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
3833 (ix86_can_change_mode_class): ...this new function, inverting the
3834 sense of the return value.
3835 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3836 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3837 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3838 (ia64_can_change_mode_class): New function.
3839 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3840 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
3841 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
3842 (m32c_can_change_mode_class): ...this new function, inverting the
3843 sense of the return value.
3844 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3845 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3846 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
3847 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
3848 (mips_can_change_mode_class): ...this new function, inverting the
3849 sense of the return value.
3850 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3851 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3852 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3853 (msp430_can_change_mode_class): New function.
3854 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3855 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
3856 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3857 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3858 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3859 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
3860 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3861 (pa_cannot_change_mode_class): Replace with...
3862 (pa_can_change_mode_class): ...this new function, inverting the
3863 sense of the return value.
3864 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3865 than CANNOT_CHANGE_MODE_CLASS in comments.
3866 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3867 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
3868 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3869 (pdp11_cannot_change_mode_class): Replace with...
3870 (pdp11_can_change_mode_class): ...this new function, inverting the
3871 sense of the return value.
3872 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3873 * config/powerpcspe/powerpcspe-protos.h
3874 (rs6000_cannot_change_mode_class_ptr): Delete.
3875 * config/powerpcspe/powerpcspe.c
3876 (rs6000_cannot_change_mode_class_ptr): Delete.
3877 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3878 (rs6000_option_override_internal): Assign to
3879 targetm.can_change_mode_class instead of
3880 rs6000_cannot_change_mode_class_ptr.
3881 (rs6000_cannot_change_mode_class): Replace with...
3882 (rs6000_can_change_mode_class): ...this new function, inverting the
3883 sense of the return value.
3884 (rs6000_debug_cannot_change_mode_class): Replace with...
3885 (rs6000_debug_can_change_mode_class): ...this new function.
3886 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3887 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
3888 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3889 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3890 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
3892 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
3893 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3894 (rs6000_option_override_internal): Assign to
3895 targetm.can_change_mode_class instead of
3896 rs6000_cannot_change_mode_class_ptr.
3897 (rs6000_cannot_change_mode_class): Replace with...
3898 (rs6000_can_change_mode_class): ...this new function, inverting the
3899 sense of the return value.
3900 (rs6000_debug_cannot_change_mode_class): Replace with...
3901 (rs6000_debug_can_change_mode_class): ...this new function.
3902 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3903 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
3904 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
3905 (s390_can_change_mode_class): ...this new function, inverting the
3906 sense of the return value.
3907 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3908 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3909 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
3910 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3911 (sh_cannot_change_mode_class): Replace with...
3912 (sh_can_change_mode_class): ...this new function, inverting the
3913 sense of the return value.
3914 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3915 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3916 (sparc_can_change_mode_class): New function.
3917 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3918 * config/spu/spu.c (spu_can_change_mode_class): New function.
3919 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3920 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3921 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3922 (visium_can_change_mode_class): New function.
3923 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
3925 2017-09-15 Richard Biener <rguenther@suse.de>
3927 PR tree-optimization/82217
3928 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
3929 but not undefined case.
3931 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3934 * postreload.c (reload_cse_simplify_operands): Skip
3935 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
3937 2017-09-15 Richard Biener <rguenther@suse.de>
3939 PR tree-optimization/68823
3940 * graphite-scop-detection.c (build_alias_set): If we have a
3941 possible dependence check whether we can handle them by just
3942 looking at the DRs DR_ACCESS_FNs.
3943 (build_scops): If build_alias_set fails, fail the SCOP.
3945 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3947 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
3948 to support float128 built-in functions that require the ISA 3.0
3950 (BU_FLOAT128_3_HW): Likewise.
3951 (SQRTF128): Add support for the IEEE 128-bit square root and fma
3953 (FMAF128): Likewise.
3955 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
3956 support for built-in functions that need the ISA 3.0 IEEE 128-bit
3957 floating point instructions.
3958 (rs6000_invalid_builtin): Likewise.
3959 (rs6000_builtin_mask_names): Likewise.
3960 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
3961 (RS6000_BTM_FLOAT128_HW): Likewise.
3962 (RS6000_BTM_COMMON): Likewise.
3963 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
3965 * doc/extend.texi (RS/6000 built-in functions): Document the
3966 IEEE 128-bit floating point square root and fused multiply-add
3969 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
3971 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
3972 reg (r2) isn't in the set of registers defined in the prologue.
3974 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3975 Alan Hayward <alan.hayward@arm.com>
3976 David Sherwood <david.sherwood@arm.com>
3978 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
3979 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
3980 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
3981 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
3982 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
3984 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
3985 max_vectorization_factor.
3986 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
3988 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3989 Alan Hayward <alan.hayward@arm.com>
3990 David Sherwood <david.sherwood@arm.com>
3992 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
3993 (vect_worthwhile_without_simd_p): Declare.
3994 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
3995 (vectorizable_reduction): Use it.
3996 * tree-vect-stmts.c (vectorizable_shift): Likewise.
3997 (vectorizable_operation): Likewise.
3999 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4000 Alan Hayward <alan.hayward@arm.com>
4001 David Sherwood <david.sherwood@arm.com>
4003 * tree-vectorizer.h (vect_get_num_copies): New function.
4004 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
4005 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4006 (vectorizable_induction): Likewise.
4007 (vectorizable_live_operation): Likewise.
4008 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4009 (vectorizable_bswap): Likewise.
4010 (vectorizable_call): Likewise.
4011 (vectorizable_conversion): Likewise.
4012 (vectorizable_assignment): Likewise.
4013 (vectorizable_shift): Likewise.
4014 (vectorizable_operation): Likewise.
4015 (vectorizable_store): Likewise.
4016 (vectorizable_load): Likewise.
4017 (vectorizable_condition): Likewise.
4018 (vectorizable_comparison): Likewise.
4019 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
4021 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4022 Alan Hayward <alan.hayward@arm.com>
4023 David Sherwood <david.sherwood@arm.com>
4025 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
4026 of vect_init_vector.
4028 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4029 Alan Hayward <alan.hayward@arm.com>
4030 David Sherwood <david.sherwood@arm.com>
4032 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
4033 an inline wrapper that provides a location.
4034 (gimple_build_vector): Likewise.
4035 * gimple-fold.c (gimple_build_vector_from_val): New function.
4036 (gimple_build_vector): Likewise.
4037 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
4038 functions to build the initial value. Always return a gimple value.
4039 (get_initial_defs_for_reduction): Likewise. Only compute
4041 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
4042 vect_init_vector on the results from get_initial_def(s)_for_reduction.
4043 (vectorizable_induction): Use gimple_build_vector rather than
4046 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4047 Alan Hayward <alan.hayward@arm.com>
4048 David Sherwood <david.sherwood@arm.com>
4050 * target.h (vec_perm_indices): New typedef.
4051 (auto_vec_perm_indices): Likewise.
4052 * optabs-query.h: Include target.h
4053 (can_vec_perm_p): Take a vec_perm_indices *.
4054 * optabs-query.c (can_vec_perm_p): Likewise.
4055 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
4056 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4057 * tree-vect-generic.c (lower_vec_perm): Likewise.
4058 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
4059 (vect_grouped_load_supported): Likewise.
4060 (vect_shift_permute_load_chain): Likewise.
4061 (vect_permute_store_chain): Use auto_vec_perm_indices.
4062 (vect_permute_load_chain): Likewise.
4063 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
4064 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
4065 Update uses of can_vec_perm_p.
4066 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
4067 mode with a number of elements. Take a vec_perm_indices *.
4068 (vect_create_epilog_for_reduction): Update accordingly.
4069 Use auto_vec_perm_indices.
4070 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
4071 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
4072 (vect_transform_slp_perm_load): Likewise.
4073 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
4074 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
4075 (vect_gen_perm_mask_checked): Likewise.
4076 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
4077 (vect_gen_perm_mask_checked): Likewise.
4078 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
4079 (vectorizable_store): Likewise.
4080 (vectorizable_load): Likewise.
4081 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
4082 (vectorizable_bswap): Likewise.
4084 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4085 Alan Hayward <alan.hayward@arm.com>
4086 David Sherwood <david.sherwood@arm.com>
4088 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
4089 * tree.c (build_vector): Likewise.
4090 (build_vector_from_ctor): Update accordingly.
4091 (build_vector_from_val): Likewise.
4092 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4093 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4094 * tree-vect-generic.c (add_rshift): Likewise.
4095 (expand_vector_divmod): Likewise.
4096 (optimize_vector_constructor): Likewise.
4097 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4098 (vect_transform_slp_perm_load): Likewise.
4099 (vect_schedule_slp_instance): Likewise.
4100 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
4101 (vectorizable_call): Likewise.
4102 (vect_gen_perm_mask_any): Likewise. Add elements in order.
4103 * expmed.c (make_tree): Likewise.
4104 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
4105 a vector passed to build_vector.
4106 (fold_convert_const): Likewise.
4107 (exact_inverse): Likewise.
4108 (fold_ternary_loc): Likewise.
4109 (fold_relational_const): Likewise.
4110 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
4111 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
4112 (const_unop): Likewise. Store the reduction accumulator in a
4113 variable rather than an array.
4114 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
4115 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
4116 the new vector, rather than constructing it after the input arrays.
4117 (native_interpret_vector): Use auto_vec<tree> when building
4118 a vector passed to build_vector. Add elements in order.
4119 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
4120 auto_vec<tree> when building a vector passed to build_vector.
4121 (vect_create_epilog_for_reduction): Likewise.
4122 (vectorizable_induction): Likewise.
4123 (get_initial_def_for_reduction): Likewise. Fix indentation of
4125 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
4127 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
4128 passed to build_vector.
4130 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4131 Alan Hayward <alan.hayward@arm.com>
4132 David Sherwood <david.sherwood@arm.com>
4134 * tree-core.h (tree_base::u): Add an "nelts" field.
4135 (tree_vector): Use VECTOR_CST_NELTS as the length.
4136 * tree.c (tree_size): Likewise.
4137 (make_vector): Initialize VECTOR_CST_NELTS.
4138 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
4139 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
4140 TYPE_VECTOR_SUBPARTS.
4141 * expr.c (const_vector_mask_from_tree): Consistently use "units"
4142 as the number of units, setting it from VECTOR_CST_NELTS.
4143 (const_vector_from_tree): Likewise.
4144 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
4145 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
4146 (fold_negate_expr_1): Likewise.
4147 (fold_convert_const): Likewise.
4148 (const_binop): Likewise. Differentiate the number of output and
4150 (const_unop): Likewise.
4151 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
4152 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
4153 in cases that did the opposite.
4155 2017-09-14 Richard Biener <rguenther@suse.de>
4157 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
4160 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
4162 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
4164 2017-09-14 Jakub Jelinek <jakub@redhat.com>
4167 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
4168 if and where to split a bb, except for splitting before debug insn
4169 sequences followed by non-label real insn. Delete debug insns
4170 in between basic blocks.
4172 * combine.c (make_compound_operation_int): Formatting fixes.
4174 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
4175 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4176 * config/netbsd.h (LINK_EH_SPEC): Likewise.
4177 * config/sol2.h (LINK_EH_SPEC): Likewise.
4178 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4179 * config/s390/linux.h (LINK_SPEC): Likewise.
4180 * config/freebsd.h (LINK_EH_SPEC): Likewise.
4181 * config/openbsd.h (LINK_EH_SPEC): Likewise.
4182 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4183 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
4184 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
4185 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4186 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
4187 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
4189 2017-09-13 Jakub Jelinek <jakub@redhat.com>
4191 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
4193 (ENDFILE_LINUX_SPEC): Likewise.
4194 (LINK_EH_SPEC): Likewise.
4195 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
4196 (LINK_OS_LINUX_SPEC32): Likewise.
4197 (LINK_OS_LINUX_SPEC64): Likewise.
4198 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
4199 (LINK_OS_LINUX_SPEC): Likewise.
4201 2017-09-13 Martin Liska <mliska@suse.cz>
4204 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
4205 CASE_HIGH is NULL_TREE.
4207 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4208 Alan Hayward <alan.hayward@arm.com>
4209 David Sherwood <david.sherwood@arm.com>
4211 * target.def (secondary_memory_needed): New hook.
4212 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
4213 instead of SECONDARY_MEMORY_NEEDED.
4214 (secondary_memory_needed_mode): Likewise.
4215 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
4216 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
4217 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
4218 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
4219 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
4220 * doc/tm.texi: Regenerate.
4221 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
4222 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
4223 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4224 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
4225 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
4226 * config/i386/i386.c (inline_secondary_memory_needed): Put the
4227 mode argument first and change the reg_class arguments to reg_class_t.
4228 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
4229 Make static. Update the call to inline_secondary_memory_needed.
4230 (ix86_register_move_cost): Update the call to
4231 inline_secondary_memory_needed.
4232 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4233 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
4235 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
4236 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4238 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
4239 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
4240 * config/mips/mips.c (mips_secondary_memory_needed): Make static
4241 and match hook interface. Add comment from mips.h.
4242 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4243 * config/mmix/mmix.md (truncdfsf2): Refer to
4244 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4246 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
4247 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
4248 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4249 (pa_secondary_memory_needed): New function.
4250 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
4251 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
4252 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4253 (pdp11_secondary_memory_needed): Make static and match hook interface.
4254 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
4255 * config/powerpcspe/powerpcspe-protos.h
4256 (rs6000_secondary_memory_needed_ptr): Delete.
4257 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
4259 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4260 (rs6000_option_override_internal): Assign to
4261 targetm.secondary_memory_needed rather than
4262 rs6000_secondary_memory_needed_ptr.
4263 (rs6000_secondary_memory_needed): Match hook interface.
4264 (rs6000_debug_secondary_memory_needed): Likewise.
4265 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
4266 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
4267 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
4268 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4269 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
4270 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
4272 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
4273 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4274 (rs6000_option_override_internal): Assign to
4275 targetm.secondary_memory_needed rather than
4276 rs6000_secondary_memory_needed_ptr.
4277 (rs6000_secondary_memory_needed): Match hook interface.
4278 (rs6000_debug_secondary_memory_needed): Likewise.
4279 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
4280 * config/s390/s390.c (s390_secondary_memory_needed): New function.
4281 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4282 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
4283 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4284 (sparc_secondary_memory_needed): New function.
4285 * lra-constraints.c (check_and_process_move): Refer to
4286 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4288 (curr_insn_transform): Likewise.
4289 (process_alt_operands): Use targetm.secondary_memory_needed
4290 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4291 (check_secondary_memory_needed_p): Likewise.
4292 (choose_split_class): Likewise.
4293 * reload.c: Unconditionally include code that was previously
4294 conditional on SECONDARY_MEMORY_NEEDED.
4295 (push_secondary_reload): Use targetm.secondary_memory_needed
4296 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4297 (push_reload): Likewise.
4298 * reload1.c: Unconditionally include code that was previously
4299 conditional on SECONDARY_MEMORY_NEEDED.
4300 (choose_reload_regs): Use targetm.secondary_memory_needed
4301 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4302 (gen_reload): Likewise.
4303 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
4305 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4306 Alan Hayward <alan.hayward@arm.com>
4307 David Sherwood <david.sherwood@arm.com>
4309 * target.def (secondary_memory_needed_mode): New hook:
4310 * targhooks.c (default_secondary_memory_needed_mode): Declare.
4311 * targhooks.h (default_secondary_memory_needed_mode): New function.
4312 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
4313 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
4314 * doc/tm.texi: Regenerate.
4315 * lra-constraints.c (check_and_process_move): Use
4316 targetm.secondary_memory_needed_mode instead of
4317 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
4318 (curr_insn_transform): Likewise.
4319 * reload.c (get_secondary_mem): Likewise.
4320 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4321 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
4323 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4324 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4325 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
4326 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4327 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
4329 * config/powerpcspe/powerpcspe-protos.h
4330 (rs6000_secondary_memory_needed_mode): Delete.
4331 * config/powerpcspe/powerpcspe.c
4332 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4333 (rs6000_secondary_memory_needed_mode): Make static.
4334 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4335 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
4337 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4339 (rs6000_secondary_memory_needed_mode): Make static.
4340 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4341 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
4342 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4343 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4344 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4346 (sparc_secondary_memory_needed_mode): New function.
4347 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
4349 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
4351 * config/aarch64/constraints.md (Umq): New constraint.
4352 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
4354 (mov<mode>): Update condition.
4356 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4358 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
4359 when bitposition is the same.
4361 2017-09-13 Richard Biener <rguenther@suse.de>
4363 * dwarf2out.c (output_die_symbol): Remove.
4364 (output_die): Do not output a DIEs symbol.
4366 2017-09-13 Richard Biener <rguenther@suse.de>
4369 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
4370 default-def to avoid breaking iterator update with the weird
4371 interaction with cgraph_update_edges_for_call_stmt_node.
4373 2017-09-13 Richard Biener <rguenther@suse.de>
4375 * tree-cfg.c (verify_gimple_assign_binary): Add verification
4376 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
4377 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
4378 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
4380 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
4382 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
4383 Disable pc relative literal load irrespective of
4384 TARGET_FIX_ERR_A53_84341 for default.
4386 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
4388 * config/sparc/sparc.c (output_return): Output the source location of
4389 the insn in the delay slot, if any.
4390 (output_sibcall): Likewise.
4392 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
4395 * common.opt (-static-pie): New alias.
4396 (shared): Negate static-pie.
4397 (-no-pie): Update help text.
4399 (static-pie): New option.
4400 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
4401 -static-pie support.
4402 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
4403 (LINK_EH_SPEC): Likewise.
4404 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4405 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4406 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4407 * gcc.c (LINK_COMMAND_SPEC): Likewise.
4408 (init_gcc_specs): Likewise.
4409 (init_spec): Likewise.
4410 (display_help): Update help message for -pie.
4411 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
4414 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
4416 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
4417 (movdi_aarch64): Likewise.
4418 (movti_aarch64): Likewise.
4420 2017-09-12 Simon Wright <simon@pushface.org>
4423 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
4424 calculation of the minor version, always output as 0.
4426 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4429 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4430 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
4431 on it early, rather than manual conversion late. For
4432 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
4433 instead of performing manual conversion.
4435 2017-09-12 Carl Love <cel@us.ibm.com>
4437 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
4438 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
4440 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4441 VMULOSW): Add definitions.
4442 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4443 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4444 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4445 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4446 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4448 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4450 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
4452 (movti_aarch64): Likewise.
4453 (movdf_aarch64): Likewise.
4454 (movtf_aarch64): Likewise.
4455 (load_pairdi): Likewise.
4456 (store_pairdi): Likewise.
4457 (load_pairdf): Likewise.
4458 (store_pairdf): Likewise.
4459 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
4460 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
4461 (ldr_got_small_<mode>): Likewise.
4462 (ldr_got_small_28k_<mode>): Likewise.
4463 (ldr_got_tiny): Likewise.
4464 * config/aarch64/iterators.md (ldst_sz): New.
4465 (ldpstp_sz): Likewise.
4466 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
4468 (thunderx_load): Split load_8 to load_16.
4469 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
4471 (thunderx2t99_storepair_basic): Split store_8 to store_16.
4472 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
4473 (xgene1_store_pair): Split store_8 to store_16.
4474 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
4475 (falkor_st_0_st_sd): Split store_8 to store_16.
4477 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4479 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
4480 and store1/2/3/4 to store_4/8/12/16.
4481 * config/aarch64/aarch64.md: Update for rename.
4482 * config/arm/arm.md: Likewise.: Likewise.
4483 * config/arm/arm.c: Likewise.
4484 * config/arm/thumb1.md: Likewise.
4485 * config/arm/thumb2.md: Likewise.
4486 * config/arm/vfp.md: Likewise.
4487 * config/arm/arm-generic.md: Likewise.
4488 * config/arm/arm1020e.md: Likewise.
4489 * config/arm/arm1026ejs.md: Likewise.
4490 * config/arm/arm1136jfs.md: Likewise.
4491 * config/arm/arm926ejs.md: Likewise.
4492 * config/arm/cortex-a15.md: Likewise.
4493 * config/arm/cortex-a17.md: Likewise.
4494 * config/arm/cortex-a5.md: Likewise.
4495 * config/arm/cortex-a53.md: Likewise.
4496 * config/arm/cortex-a57.md: Likewise.
4497 * config/arm/cortex-a7.md: Likewise.
4498 * config/arm/cortex-a8.md: Likewise.
4499 * config/arm/cortex-a9.md: Likewise.
4500 * config/arm/cortex-m4.md: Likewise.
4501 * config/arm/cortex-m7.md: Likewise.
4502 * config/arm/cortex-r4.md: Likewise.
4503 * config/arm/exynos-m1.md: Likewise.
4504 * config/arm/fa526.md: Likewise.
4505 * config/arm/fa606te.md: Likewise.
4506 * config/arm/fa626te.md: Likewise.
4507 * config/arm/fa726te.md: Likewise.
4508 * config/arm/fmp626.md: Likewise.
4509 * config/arm/iwmmxt.md: Likewise.
4510 * config/arm/ldmstm.md: Likewise.
4511 * config/arm/marvell-pj4.md: Likewise.
4512 * config/arm/xgene1.md: Likewise.
4513 * config/aarch64/thunderx.md: Likewise.
4514 * config/aarch64/thunderx2t99.md: Likewise.
4515 * config/aarch64/falkor.md: Likewise.
4517 2017-09-12 Martin Liska <mliska@suse.cz>
4519 * attribs.c (private_lookup_attribute): New function.
4520 * attribs.h (private_lookup_attribute): Declared here.
4521 (lookup_attribute): Called from this place.
4523 2017-09-12 Richard Biener <rguenther@suse.de>
4525 PR tree-optimization/82157
4526 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
4527 stmts with side-effects.
4529 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4530 Alan Hayward <alan.hayward@arm.com>
4531 David Sherwood <david.sherwood@arm.com>
4533 * target.def (hard_regno_nregs): New hook.
4534 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
4535 * targhooks.h (default_hard_regno_nregs): Declare.
4536 * targhooks.c (default_hard_regno_nregs): New function.
4537 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
4538 (TARGET_HARD_REGNO_NREGS): ...this hook.
4539 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
4540 (CLASS_MAX_NREGS): Likewise.
4541 * doc/tm.texi: Regenerate.
4542 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
4543 instead of HARD_REGNO_NREGS.
4544 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
4545 HARD_REGNO_NREGS in the comment.
4546 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
4547 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
4548 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
4549 Return an unsigned int.
4550 (TARGET_HARD_REGNO_NREGS): Redefine.
4551 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
4552 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
4553 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4554 (arc_hard_regno_nregs): New function.
4555 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
4556 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
4557 (arm_hard_regno_nregs): New function.
4558 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
4559 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
4560 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
4561 (TARGET_HARD_REGNO_NREGS): Redefine.
4562 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
4563 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
4564 (HARD_REGNO_NREGS): Delete.
4565 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
4566 (cr16_hard_regno_nregs): New function.
4567 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
4568 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
4569 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
4570 (cris_hard_regno_nregs): New function.
4571 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
4572 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
4573 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4574 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
4575 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
4576 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
4577 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
4578 (frv_hard_regno_nregs): Make static. Take and return an
4580 (frv_class_max_nregs): Remove outdated copy of documentation.
4581 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
4582 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
4583 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
4584 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
4585 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
4586 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
4587 (TARGET_HARD_REGNO_NREGS): Redefine.
4588 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
4589 (CLASS_MAX_NREGS): Update comment.
4590 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
4591 (ia64_hard_regno_nregs): New function.
4592 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
4593 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
4594 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
4595 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
4596 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
4598 (m32c_hard_regno_nregs): Likewise. Make static.
4599 (TARGET_HARD_REGNO_NREGS): Redefine.
4600 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
4601 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
4602 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
4603 (m68k_hard_regno_nregs): New function.
4604 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
4605 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
4606 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
4607 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
4608 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
4609 Take and return an unsigned int.
4610 (TARGET_HARD_REGNO_NREGS): Redefine.
4611 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
4612 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4613 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
4614 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
4615 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
4616 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
4617 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
4618 (msp430_hard_regno_nregs): Make static. Take and return an
4620 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
4621 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
4622 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
4623 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
4624 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
4625 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
4626 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
4627 (TARGET_HARD_REGNO_NREGS): Redefine.
4628 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
4629 (PA_HARD_REGNO_NREGS): ...this.
4630 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
4631 (PA_HARD_REGNO_NREGS): ...this.
4632 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4633 (pa_hard_regno_nregs): New function.
4634 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
4635 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
4636 (pdp11_hard_regno_nregs): New function.
4637 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
4638 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
4639 (rs6000_hard_regno_nregs_hook): New function.
4640 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
4641 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4642 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
4643 Take and return an unsigned int. Move earlier in file.
4644 (TARGET_HARD_REGNO_NREGS): Redefine.
4645 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
4646 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
4647 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
4648 (rl78_hard_regno_nregs): Make static. Take and return an
4650 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
4651 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
4652 (rs6000_hard_regno_nregs_hook): New function.
4653 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
4654 * config/rx/rx.c (rx_hard_regno_nregs): New function.
4655 (TARGET_HARD_REGNO_NREGS): Redefine.
4656 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
4657 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
4658 instead of HARD_REGNO_NREGS.
4659 (s390_hard_regno_nregs): New function.
4660 (s390_hard_regno_mode_ok): Add comment from s390.h.
4661 (TARGET_HARD_REGNO_NREGS): Redefine.
4662 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
4663 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
4664 (sh_hard_regno_nregs): New function.
4665 (sh_pass_in_reg_p): Use it.
4666 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
4667 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4668 (sparc_hard_regno_nregs): New function.
4669 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
4670 * config/spu/spu.c (spu_hard_regno_nregs): New function.
4671 (spu_function_arg_advance): Use it, supplying a valid register number.
4672 (TARGET_HARD_REGNO_NREGS): Redefine.
4673 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
4674 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
4675 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
4676 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
4677 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
4678 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
4679 (CLASS_MAX_NREGS): Remove copy of old documentation.
4680 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
4681 (visium_hard_regno_nregs): New function.
4682 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
4683 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
4684 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4685 xtensa_hard_regno_nregs): New function.
4686 * system.h (HARD_REGNO_NREGS): Poison.
4688 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4690 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
4691 hard_regno_nregs instead of HARD_REGNO_NREGS.
4692 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
4693 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
4694 (c6x_expand_epilogue): Likewise.
4695 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
4696 (frv_read_iacc_argument): Likewise.
4697 * config/sh/sh.c: Include regs.h.
4698 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4699 (regs_used): Likewise.
4700 (output_stack_adjust): Likewise.
4701 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
4702 * expmed.c: Include regs.h.
4703 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4704 * ree.c: Include regs.h.
4705 (combine_reaching_defs): Use hard_regno_nregs instead of
4707 (add_removable_extension): Likewise.
4709 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4711 * regs.h (hard_regno_nregs): Turn into a function.
4712 (end_hard_regno): Update accordingly.
4713 * caller-save.c (setup_save_areas): Likewise.
4714 (save_call_clobbered_regs): Likewise.
4715 (replace_reg_with_saved_mem): Likewise.
4716 (insert_restore): Likewise.
4717 (insert_save): Likewise.
4718 * combine.c (can_change_dest_mode): Likewise.
4719 (move_deaths): Likewise.
4720 (distribute_notes): Likewise.
4721 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
4722 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
4723 (rs6000_split_multireg_move): Likewise.
4724 (rs6000_register_move_cost): Likewise.
4725 (rs6000_memory_move_cost): Likewise.
4726 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
4727 (rs6000_split_multireg_move): Likewise.
4728 (rs6000_register_move_cost): Likewise.
4729 (rs6000_memory_move_cost): Likewise.
4730 * cselib.c (cselib_reset_table): Likewise.
4731 (cselib_lookup_1): Likewise.
4732 * emit-rtl.c (set_mode_and_regno): Likewise.
4733 * function.c (aggregate_value_p): Likewise.
4734 * ira-color.c (setup_profitable_hard_regs): Likewise.
4735 (check_hard_reg_p): Likewise.
4736 (calculate_saved_nregs): Likewise.
4737 (assign_hard_reg): Likewise.
4738 (improve_allocation): Likewise.
4739 (calculate_spill_cost): Likewise.
4740 * ira-emit.c (modify_move_list): Likewise.
4741 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
4742 (ira_hard_reg_in_set_p): Likewise.
4743 * ira.c (setup_reg_mode_hard_regset): Likewise.
4744 (clarify_prohibited_class_mode_regs): Likewise.
4745 (check_allocation): Likewise.
4746 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4747 (lra_setup_reg_renumber): Likewise.
4748 (setup_try_hard_regno_pseudos): Likewise.
4749 (spill_for): Likewise.
4750 (assign_hard_regno): Likewise.
4751 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
4752 * lra-constraints.c (in_class_p): Likewise.
4753 (lra_constraint_offset): Likewise.
4754 (simplify_operand_subreg): Likewise.
4755 (lra_constraints): Likewise.
4756 (split_reg): Likewise.
4757 (split_if_necessary): Likewise.
4758 (invariant_p): Likewise.
4759 (inherit_in_ebb): Likewise.
4760 * lra-lives.c (process_bb_lives): Likewise.
4761 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
4762 (get_hard_regs): Likewise.
4763 (do_remat): Likewise.
4764 * lra-spills.c (assign_spill_hard_regs): Likewise.
4765 * mode-switching.c (create_pre_exit): Likewise.
4766 * postreload.c (reload_combine_recognize_pattern): Likewise.
4767 * recog.c (peep2_find_free_register): Likewise.
4768 * regcprop.c (kill_value_regno): Likewise.
4769 (set_value_regno): Likewise.
4770 (copy_value): Likewise.
4771 (maybe_mode_change): Likewise.
4772 (find_oldest_value_reg): Likewise.
4773 (copyprop_hardreg_forward_1): Likewise.
4774 * regrename.c (check_new_reg_p): Likewise.
4775 (regrename_do_replace): Likewise.
4776 * reload.c (push_reload): Likewise.
4777 (combine_reloads): Likewise.
4778 (find_dummy_reload): Likewise.
4779 (operands_match_p): Likewise.
4780 (find_reloads): Likewise.
4781 (find_equiv_reg): Likewise.
4782 (reload_adjust_reg_for_mode): Likewise.
4783 * reload1.c (count_pseudo): Likewise.
4784 (count_spilled_pseudo): Likewise.
4785 (find_reg): Likewise.
4786 (clear_reload_reg_in_use): Likewise.
4787 (free_for_value_p): Likewise.
4788 (allocate_reload_reg): Likewise.
4789 (choose_reload_regs): Likewise.
4790 (reload_adjust_reg_for_temp): Likewise.
4791 (emit_reload_insns): Likewise.
4792 (delete_output_reload): Likewise.
4793 * rtlanal.c (subreg_get_info): Likewise.
4794 * sched-deps.c (sched_analyze_reg): Likewise.
4795 * sel-sched.c (init_regs_for_mode): Likewise.
4796 (mark_unavailable_hard_regs): Likewise.
4797 (choose_best_reg_1): Likewise.
4798 (verify_target_availability): Likewise.
4799 * valtrack.c (dead_debug_insert_temp): Likewise.
4800 * var-tracking.c (track_loc_p): Likewise.
4801 (emit_note_insn_var_location): Likewise.
4802 * varasm.c (make_decl_rtl): Likewise.
4803 * reginfo.c (choose_hard_reg_mode): Likewise.
4804 (init_reg_modes_target): Refer directly to
4805 this_target_regs->x_hard_regno_nregs.
4807 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4809 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
4810 instead of hard_regno_nregs.
4812 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4814 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
4815 end_hard_regno instead of hard_regno_nregs.
4816 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
4817 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
4818 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
4819 * ira-color.c (improve_allocation): Likewise.
4820 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4821 * lra-lives.c (mark_regno_live): Likewise.
4822 (mark_regno_dead): Likewise.
4823 * lra-remat.c (operand_to_remat): Likewise.
4824 * lra.c (collect_non_operand_hard_regs): Likewise.
4825 * postreload.c (reload_combine_note_store): Likewise.
4826 (move2add_valid_value_p): Likewise.
4827 * reload.c (regno_clobbered_p): Likewise.
4829 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4831 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
4833 * config/v850/v850.c (v850_reorg): Likewise.
4834 * reload.c (refers_to_regno_for_reload_p): Likewise.
4835 (find_equiv_reg): Likewise.
4836 * reload1.c (reload_reg_reaches_end_p): Likewise.
4838 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4840 * caller-save.c (add_used_regs): Use REG_NREGS instead of
4842 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
4843 * config/arm/arm.c (output_move_neon): Likewise.
4844 (arm_attr_length_move_neon): Likewise.
4845 (neon_split_vcombine): Likewise.
4846 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
4847 (c6x_mark_reg_written): Likewise.
4848 (c6x_dwarf_register_span): Likewise.
4849 * config/i386/i386.c (ix86_save_reg): Likewise.
4850 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
4851 (rws_access_reg): Likewise.
4852 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
4853 * mode-switching.c (create_pre_exit): Likewise.
4854 * ree.c (combine_reaching_defs): Likewise.
4855 (add_removable_extension): Likewise.
4856 * regcprop.c (find_oldest_value_reg): Likewise.
4857 (copyprop_hardreg_forward_1): Likewise.
4858 * reload.c (reload_inner_reg_of_subreg): Likewise.
4859 (push_reload): Likewise.
4860 (combine_reloads): Likewise.
4861 (find_dummy_reload): Likewise.
4862 (reload_adjust_reg_for_mode): Likewise.
4863 * reload1.c (find_reload_regs): Likewise.
4864 (forget_old_reloads_1): Likewise.
4865 (reload_reg_free_for_value_p): Likewise.
4866 (reload_adjust_reg_for_temp): Likewise.
4867 (emit_reload_insns): Likewise.
4868 (delete_output_reload): Likewise.
4869 * sel-sched.c (choose_best_reg_1): Likewise.
4870 (choose_best_pseudo_reg): Likewise.
4872 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4873 Alan Hayward <alan.hayward@arm.com>
4874 David Sherwood <david.sherwood@arm.com>
4876 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
4877 * target.def (slow_unaligned_access): New hook.
4878 * targhooks.h (default_slow_unaligned_access): Declare.
4879 * targhooks.c (default_slow_unaligned_access): New function.
4880 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
4881 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
4882 * doc/tm.texi: Regenerate.
4883 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
4884 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
4885 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
4887 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
4888 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
4890 (rs6000_slow_unaligned_access): New function.
4891 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4892 (expand_block_compare): Likewise.
4893 (expand_strn_compare): Likewise.
4894 (rs6000_rtx_costs): Likewise.
4895 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
4896 (riscv_slow_unaligned_access): Likewise.
4897 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
4898 (riscv_slow_unaligned_access_p): ...this and make static.
4899 (riscv_option_override): Update accordingly.
4900 (riscv_slow_unaligned_access): New function.
4901 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4902 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
4903 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4904 (rs6000_slow_unaligned_access): New function.
4905 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4906 (rs6000_rtx_costs): Likewise.
4907 * config/rs6000/rs6000-string.c (expand_block_compare)
4908 (expand_strn_compare): Use targetm.slow_unaligned_access instead
4909 of SLOW_UNALIGNED_ACCESS.
4910 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
4911 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
4912 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
4913 of SLOW_UNALIGNED_ACCESS.
4914 * expmed.c (simple_mem_bitfield_p): Likewise.
4915 * expr.c (alignment_for_piecewise_move): Likewise.
4916 (emit_group_load_1): Likewise.
4917 (emit_group_store): Likewise.
4918 (copy_blkmode_from_reg): Likewise.
4919 (emit_push_insn): Likewise.
4920 (expand_assignment): Likewise.
4921 (store_field): Likewise.
4922 (expand_expr_real_1): Likewise.
4923 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
4924 * lra-constraints.c (simplify_operand_subreg): Likewise.
4925 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4926 * gimple-ssa-store-merging.c: Likewise in block comment at start
4928 * tree-ssa-strlen.c: Include target.h.
4929 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
4930 of SLOW_UNALIGNED_ACCESS.
4931 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
4933 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4935 PR rtl-optimization/82185
4936 * expmed.c (emit_store_flag_int): Only test tem if it has been
4939 2017-09-12 Richard Biener <rguenther@suse.de>
4942 * match.pd ((FTYPE) N CMP CST): Fix typo.
4944 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
4946 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
4948 (mips_near_type_p): Add 'short_call' attribute as a synonym
4950 * doc/extend.texi (short_call): Document new function attribute.
4952 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4955 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
4956 assertion check that in the condition.
4957 (get_atomic_generic_size): Likewise. Before testing if parameter
4958 has pointer type, if it has array type, call for C++
4959 default_conversion to perform array-to-pointer conversion.
4961 2017-09-12 Richard Biener <rguenther@suse.de>
4963 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
4964 for operations we cannot scalarize.
4966 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
4968 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
4969 vectors heap vectors. Clean up comments.
4970 Make visited_bbs a reference.
4971 (profitable_jump_thread_path): Make GC
4972 vectors heap vectors. Clean up comments.
4974 (convert_and_register_jump_thread_path): Make GC vectors heap
4976 (check_subpath_and_update_thread_path): Same. Clean up comments.
4977 Make visited_bbs a reference.
4978 (handle_phi): Abstract common code to to
4979 register_jump_thread_path_if_profitable.
4980 Rename VAR_BB to DEF_BB.
4982 Make GC vectors heap vectors.
4983 Make visited_bbs a reference.
4984 (handle_assignment): Same.
4985 (register_jump_thread_path_if_profitable): New.
4986 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
4988 Make GC vectors heap vectors. Clean up comments.
4989 Make visited_bbs a reference.
4990 (find_jump_threads_backwards): Make visited_bbs live in the stack.
4991 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
4994 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
4997 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
4998 words of E_DImode object are reachable by xtensa_uimm8x4 access.
5000 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
5002 Revert r251800 and r251799.
5004 2017-09-11 Martin Jambor <mjambor@suse.cz>
5007 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
5008 arguments in advance.
5009 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
5010 use it to find predecessor edges.
5011 (naive_outof_ssa): Collect vector of predecessors.
5013 2017-09-08 Jason Merrill <jason@redhat.com>
5015 PR c++/70029 - ICE with ref-qualifier and -flto
5016 * langhooks.h (struct lang_hooks_for_types): Add
5017 copy_lang_qualifiers.
5018 * attribs.c (build_type_attribute_qual_variant): Use it.
5019 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
5021 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
5022 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
5024 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
5027 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
5028 (*mulsi3_sp64): New instruction.
5029 (mulsi3): New expander.
5031 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
5033 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
5035 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5037 * sancov.c: Include memmodel.h.
5039 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
5042 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
5045 2017-09-07 Carl Love <cel@us.ibm.com>
5047 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
5048 the sldi instruction.
5050 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
5052 * sancov.c: Include tm_p.h.
5054 2017-09-07 Jakub Jelinek <jakub@redhat.com>
5057 * output.h (switch_to_other_text_partition): New declaration.
5058 * varasm.c (switch_to_other_text_partition): New function.
5059 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
5060 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
5061 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
5062 to the other text partition before emitting LCL label and switch back
5063 after emitting the word after it.
5065 2017-09-07 Richard Biener <rguenther@suse.de>
5067 * passes.def (pass_split_crit_edges): Remove instance before PRE.
5068 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
5069 critical edges here, after loop init.
5070 (pass_data_pre): Remove PROP_no_crit_edges flags.
5071 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
5072 for valueization of call args to avoid leaking VN_TOP.
5073 (visit_use): Assert we do not visit default defs.
5074 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
5075 Use error_mark_node to more easily detect leaking VN_TOP.
5076 All default-defs are varying, not VN_TOP. Mark them visited.
5077 (run_scc_vn): Make code match comment.
5079 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5081 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
5082 OPTION_MASK_FLOAT128_KEYWORD.
5083 (POWERPC_MASKS): Likewise.
5084 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
5085 support for the -mfloat128-type option, and make -mfloat128
5086 default on PowerPC Linux systems. Define or undefine
5087 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
5088 Define __float128 to be __ieee128 if IEEE 128-bit support is
5089 enabled, or undefine it.
5090 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
5091 Delete defining __FLOAT128_TYPE__.
5092 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
5093 -mfloat128-type option and make -mfloat128 default on PowerPC
5095 (TARGET_FLOAT128_TYPE): Likewise.
5096 (-mfloat128-type): Likewise.
5097 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5098 Delete the -mfloat128-type option and make -mfloat128 default on
5099 PowerPC Linux systems. Always use __ieee128 to be the keyword for
5100 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
5101 128-bit floating point is enabled. Change tests from using
5102 -mfloat128-type to -mfloat128.
5103 (rs6000_mangle_type): Use the correct mangling for the __float128
5104 type even if normal long double is restricted to 64-bits.
5105 (floatn_mode): Enable the _Float128 type by default on VSX Linux
5107 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
5108 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
5109 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
5111 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
5112 documentation for -mfloat128.
5114 2017-09-06 Olivier Hainque <hainque@adacore.com>
5116 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
5118 2017-09-06 Wish Wu <wishwu007@gmail.com>
5119 Jakub Jelinek <jakub@redhat.com>
5121 * asan.c (initialize_sanitizer_builtins): Add
5122 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
5123 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
5124 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
5125 BT_FN_VOID_UINT64_PTR variables.
5126 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
5127 (BT_FN_VOID_UINT16_UINT16): Likewise.
5128 (BT_FN_VOID_UINT32_UINT32): Likewise.
5129 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
5130 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
5131 (BT_FN_VOID_UINT64_PTR): Likewise.
5132 * common.opt (flag_sanitize_coverage): New variable.
5133 (fsanitize-coverage=trace-pc): Remove.
5134 (fsanitize-coverage=): Add.
5135 * flag-types.h (enum sanitize_coverage_code): New enum.
5136 * fold-const.c (fold_range_test): Disable non-short-circuit
5137 optimization if flag_sanitize_coverage.
5138 (fold_truth_andor): Likewise.
5139 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
5140 * opts.c (COVERAGE_SANITIZER_OPT): Define.
5141 (coverage_sanitizer_opts): New array.
5142 (get_closest_sanitizer_option): Add OPTS argument, handle also
5143 OPT_fsanitize_coverage_.
5144 (parse_sanitizer_options): Adjusted to also handle
5145 OPT_fsanitize_coverage_.
5146 (common_handle_option): Add OPT_fsanitize_coverage_.
5147 * sancov.c (instrument_comparison, instrument_switch): New function.
5148 (sancov_pass): Add trace-cmp support.
5149 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
5150 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
5151 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
5152 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
5153 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
5154 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
5155 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
5156 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
5157 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
5158 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
5160 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5162 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5163 error. Only quit immediately if parsing is complete.
5164 (BEGIN): Initialize fatal_err and parse_done.
5165 (begin fpu, end fpu): Check number of arguments.
5166 (begin arch, end arch): Likewise.
5167 (begin cpu, end cpu): Likewise.
5168 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5169 (optalias): Likewise.
5171 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5173 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5174 * config/arm/arm-isa.h: Delete. Move definitions to ...
5175 * arm-cpus.in: ... here. Use new feature and fgroup values.
5176 * config/arm/arm.c (arm_option_override): Use lower case for feature
5178 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5179 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5180 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5182 (print_isa_bits_for): New function.
5183 (gen_isa): New function.
5184 (gen_comm_data): Use print_isa_bits_for.
5185 (define feature): New keyword.
5186 (define fgroup): New keyword.
5187 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
5188 (arm-isa.h): Add rule to generate file.
5189 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5190 case for feature bit names.
5192 2017-09-06 Richard Biener <rguenther@suse.de>
5194 * tree-ssa-pre.c (NECESSARY): Remove.
5195 (create_expression_by_pieces): Do not touch pass-local flags.
5196 (insert_into_preds_of_block): Likewise.
5197 (do_pre_regular_insertion): Likewise.
5198 (eliminate_insert): Likewise.
5199 (eliminate_dom_walker::before_dom_children): Likewise.
5200 (fini_eliminate): Do not look at inserted_exprs.
5201 (mark_operand_necessary): Remove.
5202 (remove_dead_inserted_code): Replace with simple work-list
5203 algorithm based on inserted_exprs and SSA uses.
5204 (pass_pre::execute): Re-order fini_eliminate and
5205 remove_dead_inserted_code.
5207 2017-09-06 Olivier Hainque <hainque@adacore.com>
5209 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
5210 for VxWorks 7. Adjust surrounding comments.
5212 2017-09-06 Richard Biener <rguenther@suse.de>
5214 * gimple-ssa-strength-reduction.c
5215 (find_candidates_dom_walker::before_dom_children): Also allow
5218 2017-09-06 Richard Biener <rguenther@suse.de>
5220 PR tree-optimization/82108
5221 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
5222 for gap in the non-permutation SLP case.
5224 2017-09-06 Martin Jambor <mjambor@suse.cz>
5226 PR tree-optimization/82078
5227 * tree-sra.c (sort_and_splice_var_accesses): Move call to
5228 add_access_to_work_queue...
5229 (build_accesses_from_assign): ...here.
5230 (propagate_all_subaccesses): Make sure racc is the group
5231 representative, if there is one.
5233 2017-09-06 Jakub Jelinek <jakub@redhat.com>
5236 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
5239 2017-09-06 Richard Biener <rguenther@suse.de>
5241 * gimple-ssa-strength-reduction.c
5242 (find_candidates_dom_walker::before_doom_children): Use a
5243 type and not a mode check.
5245 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5248 * config/arm/predicates.md (arm_general_adddi_operand): Create new
5250 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
5252 2017-09-05 Jeff Law <law@redhat.com>
5254 PR tree-optimization/64910
5255 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
5256 cases where we have 3 or more operands.
5258 2017-09-05 Jakub Jelinek <jakub@redhat.com>
5261 * omp-low.c (lower_omp_for): Recompute tree invariant if
5262 gimple_omp_for_initial/final is ADDR_EXPR.
5265 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
5266 into gimple val before gimplification fo the COND_EXPR.
5268 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
5270 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
5271 REGION_COPY argument.
5272 (thread_through_all_blocks): Remove unused argument to
5273 duplicate_thread_path.
5275 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5276 Alan Hayward <alan.hayward@arm.com>
5277 David Sherwood <david.sherwood@arm.com>
5279 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
5280 Take a scalar_mode rather than a machine_mode.
5281 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5282 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
5283 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5284 (aarch64_gen_adjusted_ldpstp): Likewise.
5285 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
5287 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5288 Alan Hayward <alan.hayward@arm.com>
5289 David Sherwood <david.sherwood@arm.com>
5291 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
5292 Take a scalar_int_mode instead of a machine_mode.
5293 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5294 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5295 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5296 (aarch64_simd_attr_length_rglist): Delete.
5297 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
5298 a scalar_int_mode instead of a machine_mode.
5299 (aarch64_add_offset): Likewise.
5300 (aarch64_internal_mov_immediate): Likewise
5301 (aarch64_add_constant_internal): Likewise.
5302 (aarch64_add_constant): Likewise.
5303 (aarch64_movw_imm): Likewise.
5304 (aarch64_rtx_arith_op_extract_p): Likewise.
5305 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5306 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5307 Remove assert that the mode isn't a vector.
5308 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5309 (aarch64_expand_mov_immediate): Update calls after above changes.
5310 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
5311 (aarch64_and_bitmask_imm): Check for scalar integer modes.
5312 (aarch64_move_imm): Likewise.
5313 (aarch64_can_const_movi_rtx_p): Likewise.
5314 (aarch64_strip_extend): Likewise.
5315 (aarch64_extr_rtx_p): Likewise.
5316 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
5317 a CONST_INT when the mode parameter is VOIDmode.
5318 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
5320 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5322 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
5323 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5324 (bitwise_type_for_mode): Update accordingly.
5326 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5328 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
5329 * stor-layout.c (mode_for_size_tree): Likewise.
5330 (mode_for_array): Update accordingly.
5331 (layout_decl): Likewise.
5332 (compute_record_mode): Likewise. Only set the mode once.
5334 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5336 * target.def (get_mask_mode): Change return type to opt_mode.
5338 * doc/tm.texi: Regenerate.
5339 * targhooks.h (default_get_mask_mode): Return an opt_mode.
5340 * targhooks.c (default_get_mask_mode): Likewise.
5341 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5342 * optabs-query.c (can_vec_mask_load_store_p): Update use of
5343 targetm.get_mask_mode.
5344 * tree.c (build_truth_vector_type): Likewise.
5346 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5348 * machmode.h (mode_for_vector): Return an opt_mode.
5349 * stor-layout.c (mode_for_vector): Likewise.
5350 (mode_for_int_vector): Update accordingly.
5351 (layout_type): Likewise.
5352 * config/i386/i386.c (emit_memmov): Likewise.
5353 (ix86_expand_set_or_movmem): Likewise.
5354 (ix86_expand_vector_init): Likewise.
5355 (ix86_get_mask_mode): Likewise.
5356 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
5358 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
5359 * expmed.c (extract_bit_field_1): Likewise.
5360 * expr.c (expand_expr_real_2): Likewise.
5361 * optabs-query.c (can_vec_perm_p): Likewise.
5362 (can_vec_mask_load_store_p): Likewise.
5363 * optabs.c (expand_vec_perm): Likewise.
5364 * targhooks.c (default_get_mask_mode): Likewise.
5365 * tree-vect-stmts.c (vectorizable_store): Likewise.
5366 (vectorizable_load): Likewise.
5367 (get_vectype_for_scalar_type_and_size): Likewise.
5369 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5371 * machmode.h (mode_for_int_vector): New function.
5372 * stor-layout.c (mode_for_int_vector): Likewise.
5373 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
5374 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
5375 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5376 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5377 (s390_expand_vcond): Likewise.
5379 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5381 * machmode.h (opt_machine_mode): New type.
5382 (opt_mode<T>): Allow construction from anything that can be
5384 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
5385 (mode_for_size): Return an opt_machine_mode.
5386 * stor-layout.c (mode_for_size): Likewise.
5387 (mode_for_size_tree): Update call accordingly.
5388 (bitwise_mode_for_mode): Likewise.
5389 (make_fract_type): Likewise.
5390 (make_accum_type): Likewise.
5391 * caller-save.c (replace_reg_with_saved_mem): Update call
5393 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5394 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5395 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5396 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5397 * expmed.c (extract_bit_field_1): Likewise.
5398 * reload.c (get_secondary_mem): Likewise.
5399 * varasm.c (assemble_integer): Likewise.
5400 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
5403 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5405 * machmode.h (decimal_float_mode_for_size): New function.
5406 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
5407 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
5408 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
5409 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
5410 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
5411 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
5413 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5415 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
5416 (get_builtin_sync_mode): Likewise.
5417 (expand_ifn_atomic_compare_exchange): Likewise.
5418 (expand_builtin_atomic_clear): Likewise.
5419 (expand_builtin_atomic_test_and_set): Likewise.
5420 (fold_builtin_atomic_always_lock_free): Likewise.
5421 * calls.c (compute_argument_addresses): Likewise.
5422 (emit_library_call_value_1): Likewise.
5423 (store_one_arg): Likewise.
5424 * combine.c (combine_instructions): Likewise.
5425 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
5426 * config/arm/arm.c (arm_function_value): Likewise.
5427 (aapcs_allocate_return_reg): Likewise.
5428 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
5429 * config/i386/i386.c (construct_container): Likewise.
5430 (ix86_gimplify_va_arg): Likewise.
5431 (ix86_expand_sse_cmp): Likewise.
5432 (emit_memmov): Likewise.
5433 (emit_memset): Likewise.
5434 (expand_small_movmem_or_setmem): Likewise.
5435 (ix86_expand_pextr): Likewise.
5436 (ix86_expand_pinsr): Likewise.
5437 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
5438 * config/microblaze/microblaze.c (microblaze_block_move_straight):
5440 * config/mips/mips.c (mips_function_value_1) Likewise.
5441 (mips_block_move_straight): Likewise.
5442 (mips_expand_ins_as_unaligned_store): Likewise.
5443 * config/powerpcspe/powerpcspe.c
5444 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5445 (rs6000_darwin64_record_arg_flush): Likewise.
5446 * config/rs6000/rs6000.c
5447 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5448 (rs6000_darwin64_record_arg_flush): Likewise.
5449 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
5450 (sparc_function_value_1): Likewise.
5451 * config/spu/spu.c (adjust_operand): Likewise.
5452 (spu_emit_branch_or_set): Likewise.
5453 (arith_immediate_p): Likewise.
5454 * emit-rtl.c (gen_lowpart_common): Likewise.
5455 * expr.c (expand_expr_real_1): Likewise.
5456 * function.c (assign_parm_setup_block): Likewise.
5457 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5458 * reload1.c (alter_reg): Likewise.
5459 * stor-layout.c (mode_for_vector): Likewise.
5460 (layout_type): Likewise.
5462 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5464 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
5465 (spu_convert_move): Likewise.
5466 * lower-subreg.c (resolve_simple_move): Likewise.
5468 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5471 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
5472 define_insn to a define_expand.
5473 (altivec_vsum2sws_direct): New define_insn.
5474 (altivec_vsumsws): Convert from a define_insn to a define_expand.
5476 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5478 * config/arm/arm.c (arm_option_params_internal): Improve setting of
5481 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
5486 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
5487 realign stack if stack alignment needed is less than incoming
5490 2017-09-05 Marek Polacek <polacek@redhat.com>
5493 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
5496 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5498 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
5500 2017-09-05 Richard Biener <rguenther@suse.de>
5502 PR tree-optimization/82084
5503 * fold-const.c (can_native_encode_string_p): Handle wide characters.
5505 2017-09-05 Richard Biener <rguenther@suse.de>
5507 PR tree-optimization/82102
5508 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
5510 2017-09-05 Martin Liska <mliska@suse.cz>
5512 PR tree-optimization/82032
5513 * tree-cfg.c (generate_range_test): New function.
5514 * tree-cfg.h (generate_range_test): Declared here.
5515 * tree-cfgcleanup.c (convert_single_case_switch): New function.
5516 (cleanup_control_expr_graph): Use it.
5517 * tree-switch-conversion.c (try_switch_expansion): Remove
5519 (emit_case_nodes): Use generate_range_test.
5521 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5524 * config/i386/i386.md (*<btsc><mode>_mask): Add
5525 TARGET_USE_BT to insn constraint.
5526 (*btr<mode>_mask): Ditto.
5528 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
5530 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
5531 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
5533 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5536 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
5537 TARGET_NEON and TARGET_IWMMXT.
5538 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
5539 TARGET_NEON and TARGET_IWMMXT.
5540 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
5542 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5544 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
5545 (ix86_rewrite_tls_address): Ditto.
5546 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
5547 (ix86_rewrite_tls_address_1): Ditto.
5548 (ix86_rewrite_tls_address): Ditto.
5549 * config/i386/predicates.md (tls_address_pattern): New predicate.
5550 * config/i386/i386.md (TLS address splitter): New splitter.
5552 2017-09-04 Richard Biener <rguenther@suse.de>
5554 PR tree-optimization/82084
5555 * fold-const.h (can_native_encode_string_p): Declare.
5556 * fold-const.c (can_native_encode_string_p): Factor out from ...
5557 (native_encode_string): ... here.
5558 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
5559 vectorizing stores from constants we later cannot handle.
5561 2017-09-04 Marek Polacek <polacek@redhat.com>
5564 * doc/invoke.texi: Update -Wtautological-compare documentation.
5566 2017-09-04 Jeff Law <law@redhat.com>
5568 PR tree-optimization/64910
5569 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
5570 swap the first and last operand if the last is a constant.
5572 2017-09-04 Marek Polacek <polacek@redhat.com>
5575 * convert.c (do_narrow): When sanitizing signed integer overflows,
5576 bail out for signed types.
5577 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
5579 2017-09-04 Richard Biener <rguenther@suse.de>
5581 PR tree-optimization/82060
5582 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
5583 Move devirtualization after stmt folding and before EH/AB/noreturn
5584 cleanup to get the stmt refs canonicalized. Use a bool instead
5585 of gimple_modified_p since that doesn't work for NOPs. Schedule
5586 NOPs generated by folding for removal.
5588 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5589 Alan Hayward <alan.hayward@arm.com>
5590 David Sherwood <david.sherwood@arm.com>
5592 * coretypes.h (pad_direction): New enum.
5593 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
5594 (FUNCTION_ARG_PADDING): Likewise.
5595 * target.def (function_arg_padding): New hook.
5596 * targhooks.h (default_function_arg_padding): Declare.
5597 * targhooks.c (default_function_arg_padding): New function.
5598 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
5599 (TARGET_FUNCTION_ARG_PADDING): ...this.
5600 * doc/tm.texi: Regenerate.
5601 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
5602 instead of direction.
5603 (compute_argument_addresses): Likewise.
5604 (load_register_parameters): Likewise.
5605 (emit_library_call_value_1): Likewise.
5606 (store_one_arg): Use targetm.calls.function_arg_padding instead
5607 of FUNCTION_ARG_PADDING.
5608 (must_pass_in_stack_var_size_or_pad): Likewise.
5609 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
5610 (emit_group_store): Likewise.
5611 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
5612 instead of FUNCTION_ARG_PADDING.
5613 (emit_push_insn): Likewise, and propagate enum change throughout
5615 * function.h (direction): Delete.
5616 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
5618 * function.c (assign_parm_find_stack_rtl): Likewise.
5619 (assign_parm_setup_block_p): Likewise.
5620 (assign_parm_setup_block): Likewise.
5621 (gimplify_parameters): Likewise.
5622 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
5623 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
5625 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
5626 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5627 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
5628 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
5629 (aarch64_function_arg_padding): ...this new function.
5630 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
5631 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5632 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
5633 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5634 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
5635 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5636 (arm_pad_arg_upward): Replace with...
5637 (arm_function_arg_padding): ...this new function.
5638 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
5640 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
5641 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
5642 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5643 (ia64_hpux_function_arg_padding): Replace with...
5644 (ia64_function_arg_padding): ...this new function. Use pad_direction
5645 instead of direction. Check for TARGET_HPUX.
5646 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
5647 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5648 (iq2000_function_arg_padding): New function.
5649 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
5650 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
5651 (mips_function_arg_padding): ...this new function.
5652 (mips_pad_reg_upward): Update accordingly.
5653 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5654 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
5655 targetm.calls.function_arg_padding.
5656 (FUNCTION_ARG_PADDING): Delete.
5657 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5658 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
5659 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5660 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
5661 (nios2_block_reg_padding): Return pad_direction instead of direction.
5662 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
5663 instead of direction.
5664 (nios2_function_arg_padding): Likewise. Make static.
5665 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5666 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
5667 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
5668 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
5669 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5670 (pa_function_arg_padding): Make static. Return pad_direction instead
5672 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
5673 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5674 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
5675 instead of direction. Use targetm.calls.function_arg_padding.
5676 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
5677 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5678 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
5679 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
5680 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
5682 (function_arg_padding): Rename to...
5683 (rs6000_function_arg_padding): ...this. Make static. Return
5684 pad_direction instead of direction.
5685 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5686 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
5687 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5688 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
5689 instead of direction. Use targetm.calls.function_arg_padding.
5690 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
5691 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5692 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
5693 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
5694 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5695 (function_arg_padding): Rename to...
5696 (rs6000_function_arg_padding): ...this. Make static. Return
5697 pad_direction instead of direction.
5698 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5699 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
5700 * config/s390/s390.c (s390_function_arg_padding): New function.
5701 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5702 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
5703 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
5704 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5705 (function_arg_padding): Rename to...
5706 (sparc_function_arg_padding): ...this. Make static. Return
5707 pad_direction instead of direction.
5708 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
5709 * config/spu/spu.c (spu_function_arg_padding): New function.
5710 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5711 * system.h (FUNCTION_ARG_PADDING): Poison.
5713 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5714 Alan Hayward <alan.hayward@arm.com>
5715 David Sherwood <david.sherwood@arm.com>
5717 * target.def (modes_tieable_p): New hook.
5718 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
5719 (TARGET_MODES_TIEABLE_P): ...this.
5720 * doc/tm.texi.in: Regenerate.
5721 * hooks.h (hook_bool_mode_mode_true): Declare.
5722 * hooks.c (hook_bool_mode_mode_true): New function.
5723 * combine.c (subst): Use targetm.modes_tieable_p instead of
5725 * dse.c (find_shift_sequence): Likewise.
5726 * expmed.c (extract_low_bits): Likewise.
5727 * lower-subreg.c: Include target.h.
5728 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
5730 * rtlanal.c (rtx_cost): Likewise.
5731 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
5732 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
5733 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
5734 (TARGET_MODES_TIEABLE_P): Redefine.
5735 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
5736 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
5737 (TARGET_MODES_TIEABLE_P): Redefine.
5738 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
5739 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
5740 (arc_modes_tieable_p): New function.
5741 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
5742 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
5743 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
5744 (arm_modes_tieable_p): Make static.
5745 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
5746 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
5747 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
5748 (TARGET_MODES_TIEABLE_P): Redefine.
5749 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
5750 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
5751 (TARGET_MODES_TIEABLE_P): Redefine.
5752 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
5753 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
5754 (cr16_modes_tieable_p): New function.
5755 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
5756 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
5757 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
5758 (TRULY_NOOP_TRUNCATION): Update comment.
5759 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
5760 (TRULY_NOOP_TRUNCATION): Update comment.
5761 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
5762 (frv_modes_tieable_p): New function.
5763 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
5764 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
5765 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
5766 (TARGET_MODES_TIEABLE_P): Redefine.
5767 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
5768 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
5769 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
5770 (TARGET_MODES_TIEABLE_P): Redefine.
5771 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
5772 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
5773 (ia64_modes_tieable_p): New function.
5774 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
5775 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
5776 (iq2000_modes_tieable_p): New function.
5777 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
5778 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
5779 (lm32_modes_tieable_p): New function.
5780 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
5781 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
5782 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
5783 (TARGET_MODES_TIEABLE_P): Redefine.
5784 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
5785 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
5786 (m32r_modes_tieable_p): New function.
5787 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
5788 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
5789 (m68k_modes_tieable_p): New function.
5790 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
5791 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
5792 (mcore_modes_tieable_p): New function.
5793 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
5794 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
5796 (TARGET_MODES_TIEABLE_P): Redefine.
5797 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
5798 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
5799 * config/mips/mips.c (mips_modes_tieable_p): Make static.
5800 (TARGET_MODES_TIEABLE_P): Redefine.
5801 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
5802 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
5803 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
5804 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
5805 (mn10300_modes_tieable_p): ...this and make static.
5806 (TARGET_MODES_TIEABLE_P): Redefine.
5807 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
5808 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
5809 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
5810 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
5811 (msp430_modes_tieable_p): Make static.
5812 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
5813 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
5814 (TARGET_MODES_TIEABLE_P): Redefine.
5815 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
5816 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
5817 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
5818 (TARGET_MODES_TIEABLE_P): Redefine.
5819 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
5820 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
5821 * config/pa/pa.c (pa_modes_tieable_p): Make static.
5822 (TARGET_MODES_TIEABLE_P): Redefine.
5823 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
5824 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
5825 (pdp11_modes_tieable_p): New function.
5826 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
5827 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
5828 (rs6000_modes_tieable_p): New function.
5829 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
5830 * config/powerpcspe/powerpcspe.md: Update comment.
5831 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
5832 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
5833 (TARGET_MODES_TIEABLE_P): Redefine.
5834 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
5835 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
5836 (rl78_modes_tieable_p): New function.
5837 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
5838 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
5839 (rs6000_modes_tieable_p): New function.
5840 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
5841 * config/rs6000/rs6000.md: Update comment.
5842 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
5843 * config/rx/rx.c (rx_modes_tieable_p): New function.
5844 (TARGET_MODES_TIEABLE_P): Redefine.
5845 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
5846 * config/s390/s390.c (s390_modes_tieable_p): New function.
5847 (TARGET_MODES_TIEABLE_P): Redefine.
5848 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
5849 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
5850 (sh_modes_tieable_p): New function.
5851 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
5852 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
5853 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
5854 (sparc_modes_tieable_p): Make static.
5855 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
5856 * config/spu/spu.c (spu_modes_tieable_p): New function.
5857 (TARGET_MODES_TIEABLE_P): Redefine.
5858 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
5859 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
5860 (TARGET_MODES_TIEABLE_P): Redefine.
5861 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
5862 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
5863 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
5864 * config/v850/v850.c (v850_modes_tieable_p): New function.
5865 (TARGET_MODES_TIEABLE_P): Redefine.
5866 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
5867 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
5868 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
5869 (visium_modes_tieable_p): New function.
5870 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
5871 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
5872 (xtensa_modes_tieable_p): New function.
5873 * system.h (MODES_TIEABLE_P): Poison.
5875 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5876 Alan Hayward <alan.hayward@arm.com>
5877 David Sherwood <david.sherwood@arm.com>
5879 * target.def (hard_regno_mode_ok): New hook.
5880 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
5881 (TARGET_HARD_REGNO_MODE_OK): ...this.
5882 * doc/tm.texi.in: Regenerate.
5883 * hooks.h (hook_bool_uint_mode_true): Declare.
5884 * hooks.c (hook_bool_uint_mode_true): New function.
5885 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
5887 * genpreds.c (write_insn_preds_c): Add an include of target.h.
5888 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
5889 instead of HARD_REGNO_MODE_OK.
5890 * caller-save.c: Include target.h.
5891 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
5893 * combine.c (can_combine_p): Likewise.
5894 (combinable_i3pat): Likewise.
5895 (can_change_dest_mode): Likewise.
5896 * expr.c (init_expr_target): Likewise.
5897 (convert_move): Likewise.
5898 (convert_modes): Likewise.
5899 * ira.c (setup_prohibited_class_mode_regs): Likewise.
5900 (setup_prohibited_mode_move_regs): Likewise.
5901 * ira.h (target_ira): Likewise.
5902 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5903 * lra-constraints.c (process_alt_operands): Likewise.
5904 (split_reg): Likewise.
5905 * recog.c (peep2_find_free_register): Likewise.
5906 * ree.c (combine_reaching_defs): Likewise.
5907 * regcprop.c (maybe_mode_change): Likewise.
5908 * reginfo.c (init_reg_sets_1): Likewise.
5909 (choose_hard_reg_mode): Likewise.
5910 (simplifiable_subregs): Likewise.
5911 * regrename.c (check_new_reg_p): Likewise.
5912 * reload.c (find_valid_class): Likewise.
5913 (find_valid_class_1): Likewise.
5914 (reload_inner_reg_of_subreg): Likewise.
5915 (push_reload): Likewise.
5916 (combine_reloads): Likewise.
5917 (find_dummy_reload): Likewise.
5918 (find_reloads): Likewise.
5919 * reload1.c (find_reg): Likewise.
5920 (set_reload_reg): Likewise.
5921 (allocate_reload_reg): Likewise.
5922 (choose_reload_regs): Likewise.
5923 (reload_adjust_reg_for_temp): Likewise.
5924 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
5925 (simplify_subreg_regno): Likewise.
5926 * sel-sched.c (init_regs_for_mode): Likewise.
5927 * varasm.c (make_decl_rtl): Likewise.
5928 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
5929 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
5931 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
5932 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
5933 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5934 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
5935 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
5936 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5937 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
5938 (arc_mode_class): Delete.
5939 (HARD_REGNO_MODE_OK): Delete.
5940 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5941 (arc_hard_regno_mode_ok): Rename old array to...
5942 (arc_hard_regno_mode_ok_modes): ...this.
5943 (arc_conditional_register_usage): Update accordingly.
5944 (arc_mode_class): Make static.
5945 (arc_hard_regno_mode_ok): New function.
5946 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
5947 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
5948 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5949 (arm_hard_regno_mode_ok): Make static.
5950 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
5952 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
5953 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
5954 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
5956 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5957 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
5958 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
5959 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
5960 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
5961 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5962 * config/bfin/predicates.md (valid_reg_operand): Use
5963 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
5964 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
5965 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
5966 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5967 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
5968 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
5969 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5970 (cr16_hard_regno_mode_ok): Make static and return a bool.
5971 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
5972 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5973 (cris_hard_regno_mode_ok): New function.
5974 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
5975 (epiphany_mode_class): Delete.
5976 (HARD_REGNO_MODE_OK): Delete.
5977 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
5978 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5979 (hard_regno_mode_ok): Rename to...
5980 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
5981 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
5982 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
5984 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
5985 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
5986 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5987 (frv_hard_regno_mode_ok): Make static and return a bool.
5988 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
5990 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
5991 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
5992 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
5993 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
5995 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5996 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
5997 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
5998 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
6000 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6001 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
6002 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6003 (ia64_hard_regno_mode_ok): New function.
6004 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
6005 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6006 (iq2000_hard_regno_mode_ok): New function.
6007 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
6008 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6009 (lm32_hard_regno_mode_ok): New function.
6010 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
6011 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
6012 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
6013 instead of HARD_REGNO_MODE_OK.
6014 (m32c_hard_regno_ok): Rename to...
6015 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
6016 (m32c_cannot_change_mode_class): Update accordingly.
6017 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6018 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
6019 (m32r_mode_class): Delete.
6020 (HARD_REGNO_MODE_OK): Delete.
6021 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6022 (m32r_hard_regno_mode_ok): Rename to...
6023 (m32r_hard_regno_modes): ...this.
6024 (m32r_mode_class): Make static.
6025 (m32r_hard_regno_mode_ok): New function.
6026 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
6027 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
6028 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6029 (m68k_hard_regno_mode_ok): Make static.
6030 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
6031 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6032 (mcore_hard_regno_mode_ok): New function.
6033 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
6034 (HARD_REGNO_MODE_OK): Delete.
6035 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
6037 (microblaze_hard_regno_mode_ok_p): ...this and make static.
6038 (microblaze_hard_regno_mode_ok): New function.
6039 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6040 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
6041 (mips_hard_regno_mode_ok): Delete.
6042 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
6043 (mips_hard_regno_mode_ok_p): ...this and make static.
6044 (mips_hard_regno_mode_ok_p): Rename to...
6045 (mips_hard_regno_mode_ok_uncached): ...this.
6046 (mips_hard_regno_mode_ok): New function.
6047 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
6048 of HARD_REGNO_MODE_OK.
6049 (mips_option_override): Update after above name changes.
6050 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6051 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
6052 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
6053 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
6054 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
6055 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6056 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
6057 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
6058 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
6059 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6060 (msp430_hard_regno_mode_ok): Make static and return a bool.
6061 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
6062 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
6063 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
6065 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6066 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
6067 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
6068 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
6069 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
6070 (PA_HARD_REGNO_MODE_OK): ...this
6071 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
6072 (PA_HARD_REGNO_MODE_OK): ...this.
6073 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6074 (pa_hard_regno_mode_ok): New function.
6075 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
6076 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6077 (pdp11_hard_regno_mode_ok): New function.
6078 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
6079 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
6081 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
6083 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6084 (rs6000_hard_regno_mode_ok): Rename to...
6085 (rs6000_hard_regno_mode_ok_uncached): ...this.
6086 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6087 (rs6000_hard_regno_mode_ok): New function.
6088 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
6089 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
6090 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
6091 (riscv_hard_regno_mode_ok): ...this and make static.
6092 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6093 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
6094 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
6095 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6096 (rl78_hard_regno_mode_ok): Make static and return bool.
6097 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
6098 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
6100 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
6101 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6102 (rs6000_hard_regno_mode_ok): Rename to...
6103 (rs6000_hard_regno_mode_ok_uncached): ...this.
6104 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6105 (rs6000_hard_regno_mode_ok): New function.
6106 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
6107 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
6108 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6109 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
6110 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
6111 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
6112 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6113 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
6114 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
6115 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6116 (sh_hard_regno_mode_ok): Make static.
6117 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
6118 instead of HARD_REGNO_MODE_OK.
6119 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
6120 (sparc_mode_class): Delete.
6121 (HARD_REGNO_MODE_OK): Delete.
6122 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6123 (hard_regno_mode_classes): Make static.
6124 (sparc_mode_class): Likewise.
6125 (sparc_hard_regno_mode_ok): New function.
6126 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
6127 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
6128 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
6130 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6131 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
6132 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
6133 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
6134 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
6135 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6136 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
6137 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
6138 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6139 (visium_hard_regno_mode_ok): New function.
6140 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
6141 instead of HARD_REGNO_MODE_OK.
6142 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
6143 (HARD_REGNO_MODE_OK): Delete.
6144 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
6145 (xtensa_hard_regno_mode_ok_p): ...this and make static.
6146 (xtensa_option_override): Update accordingly.
6147 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6148 (xtensa_hard_regno_mode_ok): New function.
6149 * system.h (HARD_REGNO_MODE_OK): Poison.
6151 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6152 Alan Hayward <alan.hayward@arm.com>
6153 David Sherwood <david.sherwood@arm.com>
6155 * target.def (hard_regno_call_part_clobbered): New hook.
6156 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
6157 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
6158 * doc/tm.texi: Regenerate.
6159 * hooks.h (hook_bool_uint_mode_false): Declare.
6160 * hooks.c (hook_bool_uint_mode_false): New function.
6161 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6162 * cselib.c (cselib_process_insn): Use
6163 targetm.hard_regno_call_part_clobbered instead of
6164 HARD_REGNO_CALL_PART_CLOBBERED.
6165 * ira-conflicts.c (ira_build_conflicts): Likewise.
6166 * ira-costs.c (ira_tune_allocno_costs): Likewise.
6167 * lra-constraints.c (need_for_call_save_p): Likewise.
6168 * lra-lives.c: Include target.h.
6169 (check_pseudos_live_through_calls): Use
6170 targetm.hard_regno_call_part_clobbered instead of
6171 HARD_REGNO_CALL_PART_CLOBBERED.
6172 * regcprop.c: Include target.h.
6173 (copyprop_hardreg_forward_1): Use
6174 targetm.hard_regno_call_part_clobbered instead of
6175 HARD_REGNO_CALL_PART_CLOBBERED.
6176 * reginfo.c (choose_hard_reg_mode): Likewise.
6177 * regrename.c (check_new_reg_p): Likewise.
6178 * reload.c (find_equiv_reg): Likewise.
6179 * reload1.c (emit_reload_insns): Likewise.
6180 * sched-deps.c (deps_analyze_insn): Likewise.
6181 * sel-sched.c (init_regs_for_mode): Likewise.
6182 (mark_unavailable_hard_regs): Likewise.
6183 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
6184 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6185 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
6187 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6188 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6189 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
6191 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
6193 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6194 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6195 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
6197 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6198 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6199 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
6201 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6202 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
6204 * config/powerpcspe/powerpcspe.c
6205 (rs6000_hard_regno_call_part_clobbered): New function.
6206 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6207 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6208 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
6210 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6211 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6212 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
6214 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6215 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6216 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
6218 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6219 Alan Hayward <alan.hayward@arm.com>
6220 David Sherwood <david.sherwood@arm.com>
6222 * rtl.h (subreg_memory_offset): Declare.
6223 * emit-rtl.c (subreg_memory_offset): New function.
6224 * expmed.c (store_bit_field_1): Use it.
6225 * expr.c (undefined_operand_subword_p): Likewise.
6226 * simplify-rtx.c (simplify_subreg): Likewise.
6228 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
6230 PR rtl-optimization/57448
6233 * optabs.c (expand_atomic_load): Place compiler memory barriers if
6234 using atomic_load pattern.
6235 (expand_atomic_store): Likewise.
6237 2017-09-04 Jakub Jelinek <jakub@redhat.com>
6240 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
6241 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
6242 handling. Use replace_call_with_value with NULL instead of
6243 gsi_replace, unlink_stmt_vdef and release_defs.
6245 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
6248 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
6250 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6253 * rtl.h (emit_library_call_value_1): Declare.
6254 (emit_library_call): Replace declaration with a series of overloads.
6255 Remove the parameter count argument.
6256 (emit_library_call_value): Likewise.
6257 * calls.c (emit_library_call_value_1): Make global. Replace varargs
6258 with an "rtx_mode_t *".
6259 (emit_library_call_value): Delete.
6260 (emit_library_call): Likewise.
6261 * asan.c (asan_emit_stack_protection): Update calls accordingly.
6262 (asan_emit_allocas_unpoison): Likewise.
6263 * builtins.c (expand_builtin_powi): Likewise.
6264 (expand_asan_emit_allocas_unpoison): Likewise.
6265 * cfgexpand.c (expand_main_function): Likewise.
6266 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
6267 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
6268 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
6269 * config/arm/arm.c (arm_trampoline_init): Likewise.
6270 (arm_call_tls_get_addr): Likewise.
6271 (arm_expand_divmod_libfunc): Likewise.
6272 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
6273 (smulsi3_highpart): Likewise.
6274 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
6275 (c6x_expand_compare): Likewise.
6276 (c6x_expand_movmem): Likewise.
6277 * config/frv/frv.c (frv_trampoline_init): Likewise.
6278 * config/i386/i386.c (ix86_trampoline_init): Likewise.
6279 (ix86_expand_divmod_libfunc): Likewise.
6280 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
6281 (ia64_expand_compare): Likewise.
6282 (ia64_profile_hook): Likewise.
6283 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
6284 (nonlocal_goto): Likewise.
6285 (restore_stack_nonlocal): Likewise.
6286 * config/m32r/m32r.c (block_move_call): Likewise.
6287 (m32r_trampoline_init): Likewise.
6288 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
6289 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
6290 (m68k_call_m68k_read_tp): Likewise.
6291 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
6292 (microblaze_expand_divide): Likewise.
6293 * config/mips/mips.h (mips_args): Likewise.
6294 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
6295 (MIPS_ICACHE_SYNC): Likewise.
6296 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
6297 (nios2_trampoline_init): Likewise.
6298 * config/pa/pa.c (hppa_tls_call): Likewise.
6299 (pa_trampoline_init): Likewise.
6300 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
6301 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
6302 (expand_strn_compare): Likewise.
6303 (rs6000_generate_compare): Likewise.
6304 (rs6000_expand_float128_convert): Likewise.
6305 (output_profile_hook): Likewise.
6306 (rs6000_trampoline_init): Likewise.
6307 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
6308 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
6309 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
6310 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
6311 (rs6000_generate_compare): Likewise.
6312 (rs6000_expand_float128_convert): Likewise.
6313 (output_profile_hook): Likewise.
6314 (rs6000_trampoline_init): Likewise.
6315 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
6316 * config/sh/sh.c (sh_trampoline_init): Likewise.
6317 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
6318 (sparc_emit_float_lib_cmp): Likewise.
6319 (sparc32_initialize_trampoline): Likewise.
6320 (sparc64_initialize_trampoline): Likewise.
6321 (sparc_profile_hook): Likewise.
6322 * config/spu/spu.c (ea_load_store): Likewise.
6323 * config/spu/spu.md (floatunssidf2): Likewise.
6324 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
6325 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
6326 * config/visium/visium.c (expand_block_move_4): Likewise.
6327 (expand_block_move_2): Likewise.
6328 (expand_block_move_1): Likewise.
6329 (expand_block_set_4): Likewise.
6330 (expand_block_set_2): Likewise.
6331 (expand_block_set_1): Likewise.
6332 (visium_trampoline_init): Likewise.
6333 (visium_profile_hook): Likewise.
6334 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
6335 (xtensa_setup_frame_addresses): Likewise.
6336 (xtensa_trampoline_init): Likewise.
6337 * except.c (sjlj_emit_function_enter): Likewise.
6338 (sjlj_emit_function_exit): Likewise.
6339 * explow.c (allocate_dynamic_stack_space): Likewise.
6340 (probe_stack_range): Likewise.
6341 * expr.c (convert_mode_scalar): Likewise.
6342 * optabs.c (expand_binop): Likewise.
6343 (expand_twoval_binop_libfunc): Likewise.
6344 (expand_unop): Likewise.
6345 (prepare_cmp_insn): Likewise.
6346 (prepare_float_lib_cmp): Likewise.
6347 (expand_float): Likewise.
6348 (expand_fix): Likewise.
6349 (expand_fixed_convert): Likewise.
6350 (maybe_emit_sync_lock_test_and_set): Likewise.
6351 (expand_atomic_compare_and_swap): Likewise.
6352 (expand_mem_thread_fence): Likewise.
6353 (expand_atomic_fetch_op): Likewise.
6355 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
6357 * doc/generic.texi (OpenACC): Adjust URL.
6358 * doc/invoke.texi (C Dialect Options): Ditto.
6360 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
6362 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
6363 predicate for operand 1. Add (m,<S>) constraint.
6364 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
6365 Prevent memory operand 1 with register operand 2.
6367 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
6369 PR rtl-optimization/82024
6370 * combine.c (try_combine): If the combination result is a PARALLEL,
6371 and we only need to retain the SET in there that would be placed
6372 at I2, check that we can place that at I3 instead, before doing so.
6374 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6377 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
6379 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
6380 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
6383 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
6384 Jeff Law <law@redhat.com>
6386 * varasm.c (bss_initializer_p): Do not put constants into .bss
6387 (categorize_decl_for_section): Handle bss_initializer_p returning
6388 false when DECL_INITIAL is NULL.
6390 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6393 * config/s390/s390.c (s390_can_inline_p): New function.
6395 2017-09-01 Jeff Law <law@redhat.com>
6397 PR tree-optimization/82052
6398 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
6399 Always initialize the returned slot after a hash table miss
6400 when INSERT is true.
6402 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
6404 * config/s390/s390.md (mem_signal_fence): Remove.
6405 * doc/md.texi (mem_signal_fence): Remove.
6406 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
6408 * target-insns.def (mem_signal_fence): Remove.
6410 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6413 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
6416 * asan.c (create_odr_indicator): Strip name encoding from assembler
6417 name before appending it after __odr_asan_.
6419 2017-09-01 Martin Liska <mliska@suse.cz>
6421 PR tree-optimization/82059
6422 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
6423 frequency only when an edge is redirected.
6425 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6427 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
6428 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
6429 (arc_conditional_register_usage): Remove ARC600 lp_count
6431 (arc_file_start): Emit Tag_ARC_CPU_variation.
6432 (arc_can_use_doloop_p): New conditions to use ZOLs.
6433 (hwloop_fail): New function.
6434 (hwloop_optimize): Likewise.
6435 (hwloop_pattern_reg): Likewise.
6436 (arc_doloop_hooks): New struct, to be used with reorg_loops.
6437 (arc_reorg_loops): New function, calls reorg_loops.
6438 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
6439 (arc600_corereg_hazard): Remove ZOL checking, case handled by
6441 (arc_loop_hazard): Remove function, functionality moved into
6443 (arc_hazard): Remove arc_loop_hazard call.
6444 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
6445 into hwloop_optimize.
6446 (arc_label_align): Remove ZOL handling.
6447 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
6448 * config/arc/arc.md (doloop_begin): Remove pattern.
6449 (doloop_begin_i): Likewise.
6450 (doloop_end_i): Likewise.
6451 (doloop_fallback): Likewise.
6452 (doloop_fallback_m): Likewise.
6453 (doloop_end): Reimplement expand.
6454 (arc_lp): New pattern for LP instruction.
6455 (loop_end): New pattern.
6456 (loop_fail): Likewise.
6457 (decrement_and_branch_until_zero): Likewise.
6458 * config/arc/arc.opt (mlpc-width): New option.
6459 * doc/invoke.texi (mlpc-width): Document option.
6461 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6463 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
6464 (arc_ccfsm_advance): Fix checking for delay slots.
6465 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
6467 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6469 * config/arc/arc.md (movqi_insn): Add stores to save constant long
6471 (movhi_insn): Update store instruction constraint which are saving
6472 6-bit short immediates.
6473 (movsi_insn): Consider also short scaled load operations.
6474 (zero_extendhisi2_i): Use Usd constraint instead of T.
6475 (extendhisi2_i): Add q constraint.
6476 (arc_clzsi2): Add type and length attributes.
6477 (arc_ctzsi2): Likewise.
6478 * config/arc/constraints.md (Usc): Update constraint, the
6479 assembler can parse two relocations for a single instruction.
6481 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6483 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
6484 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
6486 2017-08-31 Olivier Hainque <hainque@adacore.com>
6488 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
6489 match as powerpc-wrs-vxworks*.
6491 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
6493 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
6494 register constraint for by-element operand.
6495 (aarch64_mls_elt_merge<mode>): Likewise.
6497 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6499 * config/arc/arc.c (arc_can_follow_jump): Check for short
6502 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6504 * config.gcc: Use g.opt for arc.
6505 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
6506 functionality moved to ...
6507 (legitimate_scaled_address_p): New function, ...here.
6508 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
6509 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
6510 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
6512 (arc_override_options): Handle G option.
6513 (arc_output_pic_addr_const): Correct function definition.
6514 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
6515 (arc_decl_anon_ns_mem_p): Delete.
6516 (arc_in_small_data_p): Overhaul this function to take into
6517 consideration the value given via G option.
6518 (arc_rewrite_small_data_1): Renamed and corrected old
6519 arc_rewrite_small_data function.
6520 (arc_rewrite_small_data): New function.
6521 (small_data_pattern): Don't use pic_offset_table_rtx.
6522 * config/arc/arc.h (CC1_SPEC): Recognize G option.
6523 * config/arc/simdext.md (movmisalignv2hi): Use
6524 prepare_move_operands function.
6526 (movmisalign*): Likewise.
6527 * doc/invoke.texi (ARC options): Document -G option.
6529 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6531 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
6533 * config/arc/arc.c (arc_print_operand): Output scalled address for
6534 sdata whenever is possible.
6535 (arc_in_small_data_p): Allow sdata for 64bit datum when double
6536 load/stores are available.
6537 (compact_sda_memory_operand): Check for the alignment required by
6538 code density instructions.
6539 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
6541 * config/arc/constraints.md (Usd): Update constraint.
6542 (Us0): New constraint.
6543 (Usc): Update constraint.
6545 2017-08-31 Richard Biener <rguenther@suse.de>
6548 * dwarf2out.c (dwarf2out_early_global_decl): Process each
6551 2017-08-31 Tamar Christina <tamar.christina@arm.com>
6553 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
6554 Resize type_signature.
6556 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6557 Alan Hayward <alan.hayward@arm.com>
6558 David Sherwood <david.sherwood@arm.com>
6560 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
6561 subregs whose inner modes can be stored in GPRs.
6562 (aarch64_classify_index): Likewise.
6564 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6565 Alan Hayward <alan.hayward@arm.com>
6566 David Sherwood <david.sherwood@arm.com>
6568 * config/aarch64/iterators.md (V_cmp_result): Rename to...
6569 (V_INT_EQUIV): ...this.
6570 (v_cmp_result): Rename to...
6571 (v_int_equiv): ...this.
6572 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
6573 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
6574 (copysign<mode>3): Likewise.
6575 (aarch64_simd_bsl<mode>_internal): Likewise.
6576 (aarch64_simd_bsl<mode>): Likewise.
6577 (vec_cmp<mode><mode>): Likewise.
6578 (vcond<mode><mode>): Likewise.
6579 (vcond<v_cmp_mixed><mode>): Likewise.
6580 (vcondu<mode><v_cmp_mixed>): Likewise.
6581 (aarch64_cm<optab><mode>): Likewise.
6582 (aarch64_cmtst<mode>): Likewise.
6583 (aarch64_fac<optab><mode>): Likewise.
6584 (vec_perm_const<mode>): Likewise.
6585 (vcond_mask_<mode><v_cmp_result>): Rename to...
6586 (vcond_mask_<mode><v_int_equiv>): ...this.
6587 (vec_cmp<mode><v_cmp_result>): Rename to...
6588 (vec_cmp<mode><v_int_equiv>): ...this.
6590 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6591 Alan Hayward <alan.hayward@arm.com>
6592 David Sherwood <david.sherwood@arm.com>
6594 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
6596 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
6597 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
6598 (UNSPEC_LD4_DREG): New unspecs.
6599 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
6600 (aarch64_ld2<mode>_dreg_be): Replace with...
6601 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
6603 (aarch64_ld3<mode>_dreg_le)
6604 (aarch64_ld3<mode>_dreg_be): Replace with...
6605 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
6607 (aarch64_ld4<mode>_dreg_le)
6608 (aarch64_ld4<mode>_dreg_be): Replace with...
6609 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
6612 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6614 PR tree-optimization/81987
6615 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
6616 insert an initializer in a location not dominated by the stride
6619 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6621 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
6622 on the entire header of the finally block in the fallthru case.
6624 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6626 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
6628 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
6630 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
6631 rs6000_emit_move_from_cr and call renamed function.
6632 (rs6000_emit_prologue): Call renamed functions.
6633 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
6634 movesi_from_cr, remove volatile CRs.
6636 2017-08-30 Jon Beniston <jon@beniston.com>
6637 Richard Biener <rguenther@suse.de>
6639 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
6640 of VECTOR_MODE_P check.
6641 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
6642 element vector types.
6644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6646 * df.h (df_read_modify_subreg_p): Remove in favor of...
6647 * rtl.h (read_modify_subreg_p): ...this new function. Take a
6648 const_rtx instead of an rtx.
6649 * cprop.c (local_cprop_find_used_regs): Update accordingly.
6650 * df-problems.c (df_word_lr_mark_ref): Likewise.
6651 * ira-lives.c (mark_pseudo_reg_live): Likewise.
6652 (mark_pseudo_reg_dead): Likewise.
6653 (mark_ref_dead): Likewise.
6654 * reginfo.c (init_subregs_of_mode): Likewise.
6655 * sched-deps.c (sched_analyze_1): Likewise.
6656 * df-scan.c (df_def_record_1): Likewise.
6657 (df_uses_record): Likewise.
6658 (df_read_modify_subreg_p): Remove in favor of...
6659 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
6660 const_rtx instead of an rtx.
6662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6663 Alan Hayward <alan.hayward@arm.com>
6664 David Sherwood <david.sherwood@arm.com>
6666 * rtl.h (partial_subreg_p): New function.
6667 * caller-save.c (save_call_clobbered_regs): Use it.
6668 * calls.c (expand_call): Likewise.
6669 * combine.c (combinable_i3pat): Likewise.
6670 (simplify_set): Likewise.
6671 (make_extraction): Likewise.
6672 (make_compound_operation_int): Likewise.
6673 (gen_lowpart_or_truncate): Likewise.
6674 (force_to_mode): Likewise.
6675 (make_field_assignment): Likewise.
6676 (reg_truncated_to_mode): Likewise.
6677 (record_truncated_value): Likewise.
6678 (move_deaths): Likewise.
6679 * cse.c (record_jump_cond): Likewise.
6680 (cse_insn): Likewise.
6681 * cselib.c (cselib_lookup_1): Likewise.
6682 * expmed.c (extract_bit_field_using_extv): Likewise.
6683 * function.c (assign_parm_setup_reg): Likewise.
6684 * ifcvt.c (noce_convert_multiple_sets): Likewise.
6685 * ira-build.c (create_insn_allocnos): Likewise.
6686 * lra-coalesce.c (merge_pseudos): Likewise.
6687 * lra-constraints.c (match_reload): Likewise.
6688 (simplify_operand_subreg): Likewise.
6689 (curr_insn_transform): Likewise.
6690 * lra-lives.c (process_bb_lives): Likewise.
6691 * lra.c (new_insn_reg): Likewise.
6692 (lra_substitute_pseudo): Likewise.
6693 * regcprop.c (mode_change_ok): Likewise.
6694 (maybe_mode_change): Likewise.
6695 (copyprop_hardreg_forward_1): Likewise.
6696 * reload.c (push_reload): Likewise.
6697 (find_reloads): Likewise.
6698 (find_reloads_subreg_address): Likewise.
6699 * reload1.c (alter_reg): Likewise.
6700 (eliminate_regs_1): Likewise.
6701 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6703 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
6705 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
6706 back to if statements, including unpack.
6708 2017-08-30 Martin Liska <mliska@suse.cz>
6711 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
6713 (func_checker::compare_asm_inputs_outputs): ... this function.
6714 (func_checker::compare_gimple_asm): Use the function to compare
6715 also ASM constrains.
6716 * ipa-icf-gimple.h: Rename the function.
6718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6719 Alan Hayward <alan.hayward@arm.com>
6720 David Sherwood <david.sherwood@arm.com>
6722 * coretypes.h (complex_mode): New type.
6723 * gdbhooks.py (build_pretty_printer): Handle it.
6724 * machmode.h (complex_mode): New class.
6725 (complex_mode::includes_p): New function.
6726 (is_complex_int_mode): Likewise.
6727 (is_complex_float_mode): Likewise.
6728 * genmodes.c (get_mode_class): Handle complex mode classes.
6729 * function.c (expand_function_end): Use is_complex_int_mode.
6731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6732 Alan Hayward <alan.hayward@arm.com>
6733 David Sherwood <david.sherwood@arm.com>
6735 * coretypes.h (scalar_mode_pod): New typedef.
6736 * gdbhooks.py (build_pretty_printer): Handle it.
6737 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
6738 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
6739 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
6740 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
6743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6744 Alan Hayward <alan.hayward@arm.com>
6745 David Sherwood <david.sherwood@arm.com>
6747 * machmode.h (mode_for_vector): Take a scalar_mode instead
6749 * stor-layout.c (mode_for_vector): Likewise.
6750 * explow.c (promote_mode): Use as_a <scalar_mode>.
6751 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
6753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6754 Alan Hayward <alan.hayward@arm.com>
6755 David Sherwood <david.sherwood@arm.com>
6757 * target.def (preferred_simd_mode): Take a scalar_mode
6758 instead of a machine_mode.
6759 * targhooks.h (default_preferred_simd_mode): Likewise.
6760 * targhooks.c (default_preferred_simd_mode): Likewise.
6761 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
6762 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
6763 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
6764 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
6765 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
6766 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
6767 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
6768 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
6770 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
6771 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
6772 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
6773 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
6774 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
6775 * doc/tm.texi: Regenerate.
6776 * optabs-query.c (can_vec_mask_load_store_p): Return false for
6779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6780 Alan Hayward <alan.hayward@arm.com>
6781 David Sherwood <david.sherwood@arm.com>
6783 * target.def (scalar_mode_supported_p): Take a scalar_mode
6784 instead of a machine_mode.
6785 * targhooks.h (default_scalar_mode_supported_p): Likewise.
6786 * targhooks.c (default_scalar_mode_supported_p): Likewise.
6787 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
6788 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6789 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
6790 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
6791 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
6792 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
6793 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
6794 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
6795 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
6796 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
6797 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
6798 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
6800 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
6801 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
6802 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6803 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6804 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
6806 * doc/tm.texi: Regenerate.
6808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6809 Alan Hayward <alan.hayward@arm.com>
6810 David Sherwood <david.sherwood@arm.com>
6812 * coretypes.h (opt_scalar_mode): New typedef.
6813 * gdbhooks.py (build_pretty_printers): Handle it.
6814 * machmode.h (mode_iterator::get_2xwider): Add overload for
6816 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
6818 * expr.c (convert_mode_scalar): Likewise.
6819 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6820 * optabs.c (expand_float): Likewise.
6821 (expand_fix): Likewise.
6822 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6825 Alan Hayward <alan.hayward@arm.com>
6826 David Sherwood <david.sherwood@arm.com>
6828 * optabs.c (expand_float): Explicitly check for scalars before
6829 using a branching expansion.
6830 (expand_fix): Likewise.
6832 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6833 Alan Hayward <alan.hayward@arm.com>
6834 David Sherwood <david.sherwood@arm.com>
6836 * expr.c (convert_mode): Split scalar handling out into...
6837 (convert_mode_scalar): ...this new function. Treat the modes
6840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6841 Alan Hayward <alan.hayward@arm.com>
6842 David Sherwood <david.sherwood@arm.com>
6844 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
6846 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
6848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6849 Alan Hayward <alan.hayward@arm.com>
6850 David Sherwood <david.sherwood@arm.com>
6852 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
6853 rather than a machine_mode.
6854 (fixed_from_string): Likewise.
6855 (fixed_convert): Likewise.
6856 (fixed_convert_from_int): Likewise.
6857 (fixed_convert_from_real): Likewise.
6858 (real_convert_from_fixed): Likewise.
6859 * fixed-value.c (fixed_from_double_int): Likewise.
6860 (fixed_from_string): Likewise.
6861 (fixed_convert): Likewise.
6862 (fixed_convert_from_int): Likewise.
6863 (fixed_convert_from_real): Likewise.
6864 (real_convert_from_fixed): Likewise.
6865 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
6867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6868 Alan Hayward <alan.hayward@arm.com>
6869 David Sherwood <david.sherwood@arm.com>
6871 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
6872 of separate mode class checks. Do not allow vector modes here.
6873 (immed_wide_int_const): Use as_a <scalar_mode>.
6874 * explow.c (trunc_int_for_mode): Likewise.
6875 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
6876 (wi::shwi): Likewise.
6877 (wi::min_value): Likewise.
6878 (wi::max_value): Likewise.
6879 * dwarf2out.c (loc_descriptor): Likewise.
6880 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
6883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6884 Alan Hayward <alan.hayward@arm.com>
6885 David Sherwood <david.sherwood@arm.com>
6887 * tree.h (SCALAR_TYPE_MODE): New macro.
6888 * expr.c (expand_expr_addr_expr_1): Use it.
6889 (expand_expr_real_2): Likewise.
6890 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
6891 (fold_convert_const_fixed_from_int): Likewise.
6892 (fold_convert_const_fixed_from_real): Likewise.
6893 (native_encode_fixed): Likewise
6894 (native_encode_complex): Likewise
6895 (native_encode_vector): Likewise.
6896 (native_interpret_fixed): Likewise.
6897 (native_interpret_real): Likewise.
6898 (native_interpret_complex): Likewise.
6899 (native_interpret_vector): Likewise.
6900 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
6901 (simd_clone_adjust_argument_types): Likewise.
6902 (simd_clone_init_simd_arrays): Likewise.
6903 (simd_clone_adjust): Likewise.
6904 * stor-layout.c (layout_type): Likewise.
6905 * tree.c (build_minus_one_cst): Likewise.
6906 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6907 * tree-inline.c (estimate_move_cost): Likewise.
6908 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
6909 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
6910 (vectorizable_reduction): Likewise.
6911 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
6912 (vect_recog_mixed_size_cond_pattern): Likewise.
6913 (check_bool_pattern): Likewise.
6914 (adjust_bool_pattern): Likewise.
6915 (search_type_for_mask_1): Likewise.
6916 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
6917 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6918 (vectorizable_load): Likewise.
6919 (vectorizable_store): Likewise.
6920 * ubsan.c (ubsan_encode_value): Likewise.
6921 * varasm.c (output_constant): Likewise.
6923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6924 Alan Hayward <alan.hayward@arm.com>
6925 David Sherwood <david.sherwood@arm.com>
6927 * coretypes.h (scalar_mode): New class.
6928 * machmode.h (scalar_mode): Likewise.
6929 (scalar_mode::includes_p): New function.
6930 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
6931 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
6932 * genmodes.c (get_mode_class): Handle remaining scalar modes.
6933 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
6934 * expmed.c (store_bit_field_1): Likewise.
6935 (extract_bit_field_1): Likewise.
6936 * expr.c (write_complex_part): Likewise.
6937 (read_complex_part): Likewise.
6938 (emit_move_complex_push): Likewise.
6939 (expand_expr_real_2): Likewise.
6940 * function.c (assign_parm_setup_reg): Likewise.
6941 (assign_parms_unsplit_complex): Likewise.
6942 * optabs.c (expand_binop): Likewise.
6943 * rtlanal.c (subreg_get_info): Likewise.
6944 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6945 * varasm.c (output_constant_pool_2): Likewise.
6947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6948 Alan Hayward <alan.hayward@arm.com>
6949 David Sherwood <david.sherwood@arm.com>
6951 * expmed.c (extract_high_half): Use scalar_int_mode and remove
6953 (expmed_mult_highpart_optab): Likewise.
6954 (expmed_mult_highpart): Likewise.
6956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6957 Alan Hayward <alan.hayward@arm.com>
6958 David Sherwood <david.sherwood@arm.com>
6960 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
6961 instead of a machine_mode.
6962 (builtin_memset_read_str): Likewise.
6963 * builtins.c (c_readstr): Likewise.
6964 (builtin_memcpy_read_str): Likewise.
6965 (builtin_strncpy_read_str): Likewise.
6966 (builtin_memset_read_str): Likewise.
6967 (builtin_memset_gen_str): Likewise.
6968 (expand_builtin_signbit): Use scalar_int_mode for local variables.
6969 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
6970 instead of a machine_mode.
6971 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
6973 (make_extraction): Likewise.
6974 (try_widen_shift_mode): Take and return scalar_int_modes instead
6976 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
6977 a scalar_int_mode instead of a machine_mode.
6978 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
6979 (avr_addr_space_pointer_mode): Likewise.
6980 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
6981 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
6982 (msp430_unwind_word_mode): Likewise.
6983 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
6984 (spu_addr_space_pointer_mode): Likewise.
6985 (spu_addr_space_address_mode): Likewise.
6986 (spu_libgcc_cmp_return_mode): Likewise.
6987 (spu_libgcc_shift_count_mode): Likewise.
6988 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
6989 (rl78_addr_space_pointer_mode): Likewise.
6990 (fl78_unwind_word_mode): Likewise.
6991 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
6993 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
6994 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
6995 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
6996 (mips_valid_pointer_mode): Likewise.
6997 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
6998 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
6999 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
7001 (ft32_addr_space_address_mode): Likewise.
7002 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
7003 scalar_int_mode instead of a machine_mode.
7004 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
7006 (m32c_addr_space_address_mode): Likewise.
7007 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
7008 (rs6000_eh_return_filter_mode): Likewise.
7009 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
7010 (rs6000_eh_return_filter_mode): Likewise.
7011 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
7012 (s390_libgcc_shift_count_mode): Likewise.
7013 (s390_unwind_word_mode): Likewise.
7014 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
7016 * target.def (mode_rep_extended): Likewise.
7017 (valid_pointer_mode): Likewise.
7018 (addr_space.valid_pointer_mode): Likewise.
7019 (eh_return_filter_mode): Return a scalar_int_mode rather than
7021 (libgcc_cmp_return_mode): Likewise.
7022 (libgcc_shift_count_mode): Likewise.
7023 (unwind_word_mode): Likewise.
7024 (addr_space.pointer_mode): Likewise.
7025 (addr_space.address_mode): Likewise.
7026 * doc/tm.texi: Regenerate.
7027 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
7029 (do_jump): Use scalar_int_mode for local variables.
7030 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
7031 rather than a machine_mode.
7032 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
7033 (scompare_loc_descriptor_wide): Likewise.
7034 (scompare_loc_descriptor_narrow): Likewise.
7035 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
7037 * except.c (sjlj_emit_dispatch_table): Likewise.
7038 (expand_builtin_eh_copy_values): Likewise.
7039 * explow.c (convert_memory_address_addr_space_1): Likewise.
7040 Take a scalar_int_mode rather than a machine_mode.
7041 (convert_memory_address_addr_space): Take a scalar_int_mode rather
7042 than a machine_mode.
7043 (memory_address_addr_space): Use scalar_int_mode for local variables.
7044 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
7045 rather than a machine_mode.
7046 * expmed.c (mask_rtx): Likewise.
7047 (init_expmed_one_conv): Likewise.
7048 (expand_mult_highpart_adjust): Likewise.
7049 (extract_high_half): Likewise.
7050 (expmed_mult_highpart_optab): Likewise.
7051 (expmed_mult_highpart): Likewise.
7052 (expand_smod_pow2): Likewise.
7053 (expand_sdiv_pow2): Likewise.
7054 (emit_store_flag_int): Likewise.
7055 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
7057 (extract_low_bits): Likewise.
7058 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
7060 * expr.c (pieces_addr::adjust): Likewise.
7061 (can_store_by_pieces): Likewise.
7062 (store_by_pieces): Likewise.
7063 (clear_by_pieces_1): Likewise.
7064 (expand_expr_addr_expr_1): Likewise.
7065 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
7066 (expand_expr_real_1): Likewise.
7067 (try_casesi): Likewise.
7068 * final.c (shorten_branches): Likewise.
7069 * fold-const.c (fold_convert_const_int_from_fixed): Change the
7070 type of "mode" to machine_mode.
7071 * internal-fn.c (expand_arith_overflow_result_store): Take a
7072 scalar_int_mode rather than a machine_mode.
7073 (expand_mul_overflow): Use scalar_int_mode for local variables.
7074 * loop-doloop.c (doloop_modify): Likewise.
7075 (doloop_optimize): Likewise.
7076 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
7077 than a machine_mode.
7078 (expand_doubleword_shift_condmove): Likewise.
7079 (expand_doubleword_shift): Likewise.
7080 (expand_doubleword_clz): Likewise.
7081 (expand_doubleword_popcount): Likewise.
7082 (expand_doubleword_parity): Likewise.
7083 (expand_absneg_bit): Use scalar_int_mode for local variables.
7084 (prepare_float_lib_cmp): Likewise.
7085 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
7086 rather than a machine_mode.
7087 (convert_memory_address_addr_space): Likewise.
7088 (get_mode_bounds): Likewise.
7089 (get_address_mode): Return a scalar_int_mode rather than a
7091 * rtlanal.c (get_address_mode): Likewise.
7092 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
7093 than a machine_mode.
7094 * targhooks.c (default_mode_rep_extended): Likewise.
7095 (default_valid_pointer_mode): Likewise.
7096 (default_addr_space_valid_pointer_mode): Likewise.
7097 (default_eh_return_filter_mode): Return a scalar_int_mode rather
7098 than a machine_mode.
7099 (default_libgcc_cmp_return_mode): Likewise.
7100 (default_libgcc_shift_count_mode): Likewise.
7101 (default_unwind_word_mode): Likewise.
7102 (default_addr_space_pointer_mode): Likewise.
7103 (default_addr_space_address_mode): Likewise.
7104 * targhooks.h (default_eh_return_filter_mode): Likewise.
7105 (default_libgcc_cmp_return_mode): Likewise.
7106 (default_libgcc_shift_count_mode): Likewise.
7107 (default_unwind_word_mode): Likewise.
7108 (default_addr_space_pointer_mode): Likewise.
7109 (default_addr_space_address_mode): Likewise.
7110 (default_mode_rep_extended): Take a scalar_int_mode rather than
7112 (default_valid_pointer_mode): Likewise.
7113 (default_addr_space_valid_pointer_mode): Likewise.
7114 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
7116 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
7117 rather than a machine_mode.
7118 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
7119 for local variables.
7120 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7121 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
7122 than a machine_mode.
7124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7125 Alan Hayward <alan.hayward@arm.com>
7126 David Sherwood <david.sherwood@arm.com>
7128 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
7129 the mode argument to scalar_int_mode.
7130 (do_jump_by_parts_zero_rtx): Likewise.
7131 (do_jump_by_parts_equality_rtx): Likewise.
7132 (do_jump_by_parts_greater): Take a mode argument.
7133 (do_jump_by_parts_equality): Likewise.
7134 (do_jump_1): Update calls accordingly.
7136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7137 Alan Hayward <alan.hayward@arm.com>
7138 David Sherwood <david.sherwood@arm.com>
7140 * is-a.h (safe_dyn_cast): New function.
7141 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
7142 (jump_table_for_label): Likewise.
7143 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
7144 instead of an rtx_insn *.
7145 (shorten_branches): Use dyn_cast instead of LABEL_P and
7146 JUMP_TABLE_DATA_P. Use jump_table_for_label and
7147 rtx_jump_table_data::get_data_mode.
7148 (final_scan_insn): Likewise.
7150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7151 Alan Hayward <alan.hayward@arm.com>
7152 David Sherwood <david.sherwood@arm.com>
7154 * combine.c (try_combine): Use is_a <scalar_int_mode> when
7155 trying to combine a full-register integer set with a subreg
7158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7159 Alan Hayward <alan.hayward@arm.com>
7160 David Sherwood <david.sherwood@arm.com>
7162 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
7163 that is always either address_mode or pointer_mode.
7165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7166 Alan Hayward <alan.hayward@arm.com>
7167 David Sherwood <david.sherwood@arm.com>
7169 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
7170 when the two are known to be equal.
7172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7173 Alan Hayward <alan.hayward@arm.com>
7174 David Sherwood <david.sherwood@arm.com>
7176 * simplify-rtx.c (simplify_const_unary_operation): Use
7177 is_a <scalar_int_mode> instead of checking for a nonzero
7178 precision. Forcibly convert op_mode to a scalar_int_mode
7179 in that case. More clearly differentiate the operand and
7180 result modes and use the former when deciding what the value
7181 of a count-bits operation should be. Use is_int_mode instead
7182 of checking for a MODE_INT. Remove redundant check for whether
7183 this mode has a zero precision.
7185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7186 Alan Hayward <alan.hayward@arm.com>
7187 David Sherwood <david.sherwood@arm.com>
7189 * optabs.c (widen_leading): Change the type of the mode argument
7190 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
7191 (widen_bswap): Likewise.
7192 (expand_parity): Likewise.
7193 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
7194 (expand_ffs): Likewise.
7195 (epand_unop): Check for scalar integer modes before calling the
7198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7199 Alan Hayward <alan.hayward@arm.com>
7200 David Sherwood <david.sherwood@arm.com>
7202 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
7204 (expand_expr_real_1): Update call accordingly.
7206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7207 Alan Hayward <alan.hayward@arm.com>
7208 David Sherwood <david.sherwood@arm.com>
7210 * expmed.c (store_bit_field_using_insv): Add op0_mode and
7211 value_mode arguments. Use scalar_int_mode internally.
7212 (store_bit_field_1): Rename the new integer mode from imode
7213 to op0_mode and use it instead of GET_MODE (op0). Update calls
7214 to store_split_bit_field, store_bit_field_using_insv and
7215 store_fixed_bit_field.
7216 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
7217 Use scalar_int_mode internally. Use a bit count rather than a mode
7218 when calculating the largest bit size for get_best_mode.
7219 Update calls to store_split_bit_field and store_fixed_bit_field_1.
7220 (store_fixed_bit_field_1): Add mode and value_mode arguments.
7221 Remove assertion that OP0 has a scalar integer mode.
7222 (store_split_bit_field): Add op0_mode and value_mode arguments.
7223 Update calls to extract_fixed_bit_field.
7224 (extract_bit_field_using_extv): Add an op0_mode argument.
7225 Use scalar_int_mode internally.
7226 (extract_bit_field_1): Rename the new integer mode from imode to
7227 op0_mode and use it instead of GET_MODE (op0). Update calls to
7228 extract_split_bit_field, extract_bit_field_using_extv and
7229 extract_fixed_bit_field.
7230 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
7231 to extract_split_bit_field and extract_fixed_bit_field_1.
7232 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
7233 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
7235 (extract_split_bit_field): Add an op0_mode argument. Update call
7236 to extract_fixed_bit_field.
7238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7239 Alan Hayward <alan.hayward@arm.com>
7240 David Sherwood <david.sherwood@arm.com>
7242 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
7243 * explow.c (hard_function_value): Likewise.
7244 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
7245 convert_to_mode call outside the loop.
7246 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
7247 for the mode iterator. Require the mode specified by max_pieces
7249 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
7251 (copy_blkmode_to_reg): Likewise.
7252 (set_storage_via_setmem): Likewise.
7253 * optabs.c (prepare_cmp_insn): Likewise.
7254 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7255 * stor-layout.c (finish_bitfield_representative): Likewise.
7257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7258 Alan Hayward <alan.hayward@arm.com>
7259 David Sherwood <david.sherwood@arm.com>
7261 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
7262 * expr.c (convert_move): Use them.
7263 (convert_modes): Likewise.
7264 (store_expr_with_bounds): Likewise.
7266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7267 Alan Hayward <alan.hayward@arm.com>
7268 David Sherwood <david.sherwood@arm.com>
7270 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
7271 parameter for the mode of "x". Remove the "known_x", "known_mode"
7272 and "known_ret" arguments. Change the type of the mode argument
7274 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
7275 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
7276 (reg_num_sign_bit_copies_for_combine): Likewise.
7277 * rtlanal.c (nonzero_bits1): Likewise.
7278 (num_sign_bit_copies1): Likewise.
7279 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
7280 (reg_num_sign_bit_copies_general): Likewise.
7281 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
7282 (reg_nonzero_bits_general): Likewise.
7284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7285 Alan Hayward <alan.hayward@arm.com>
7286 David Sherwood <david.sherwood@arm.com>
7288 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
7289 than in subroutines. Return 1 for non-integer modes.
7290 (cached_num_sign_bit_copies): Change the type of the mode parameter
7292 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
7293 classes. Handle CONST_INT_P first and then check whether X also
7294 has a scalar integer mode. Check the same thing for inner registers
7295 of a SUBREG and for values that are being extended or truncated.
7297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7298 Alan Hayward <alan.hayward@arm.com>
7299 David Sherwood <david.sherwood@arm.com>
7301 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
7302 in subroutines. Return the mode mask for non-integer modes.
7303 (cached_nonzero_bits): Change the type of the mode parameter
7305 (nonzero_bits1): Likewise. Remove early exit for other mode
7306 classes. Handle CONST_INT_P first and then check whether X
7307 also has a scalar integer mode.
7309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7310 Alan Hayward <alan.hayward@arm.com>
7311 David Sherwood <david.sherwood@arm.com>
7313 * expr.c (widest_int_mode_for_size): Make the comment match the code.
7314 Return a scalar_int_mode and assert that the size is greater than
7316 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
7317 (op_by_pieces_d::op_by_pieces_d): Likewise.
7318 (op_by_pieces_d::run): Likewise.
7319 (can_store_by_pieces): Likewise.
7321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7322 Alan Hayward <alan.hayward@arm.com>
7323 David Sherwood <david.sherwood@arm.com>
7325 * combine.c (extract_left_shift): Add a mode argument and update
7327 (make_compound_operation_int): Change the type of the mode parameter
7328 to scalar_int_mode and update the call to extract_left_shift.
7330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7331 Alan Hayward <alan.hayward@arm.com>
7332 David Sherwood <david.sherwood@arm.com>
7334 * combine.c (simplify_and_const_int): Change the type of the mode
7335 parameter to scalar_int_mode.
7336 (simplify_and_const_int_1): Likewise. Update recursive call.
7338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7339 Alan Hayward <alan.hayward@arm.com>
7340 David Sherwood <david.sherwood@arm.com>
7342 * combine.c (simplify_compare_const): Check that the mode is a
7343 scalar_int_mode (rather than VOIDmode) before testing its
7345 (simplify_comparison): Move COMPARISON_P handling out of the
7346 loop and restrict the latter part of the loop to scalar_int_modes.
7347 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
7348 and when considering SUBREG_REGs. Use is_int_mode instead of
7349 checking GET_MODE_CLASS against MODE_INT.
7351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7352 Alan Hayward <alan.hayward@arm.com>
7353 David Sherwood <david.sherwood@arm.com>
7355 * combine.c (try_widen_shift_mode): Move check for equal modes to...
7356 (simplify_shift_const_1): ...here. Use scalar_int_mode for
7357 shift_unit_mode and for modes involved in scalar shifts.
7359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7360 Alan Hayward <alan.hayward@arm.com>
7361 David Sherwood <david.sherwood@arm.com>
7363 * combine.c (force_int_to_mode): New function, split out from...
7364 (force_to_mode): ...here. Keep xmode up-to-date and use it
7365 instead of GET_MODE (x).
7367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7368 Alan Hayward <alan.hayward@arm.com>
7369 David Sherwood <david.sherwood@arm.com>
7371 * optabs-query.h (extraction_insn::struct_mode): Change type to
7372 opt_scalar_int_mode and update comment.
7373 (extraction_insn::field_mode): Change type to scalar_int_mode.
7374 (extraction_insn::pos_mode): Likewise.
7375 * combine.c (make_extraction): Update accordingly.
7376 * optabs-query.c (get_traditional_extraction_insn): Likewise.
7377 (get_optab_extraction_insn): Likewise.
7378 * recog.c (simplify_while_replacing): Likewise.
7379 * expmed.c (narrow_bit_field_mem): Change the type of the mode
7380 parameter to opt_scalar_int_mode.
7382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7383 Alan Hayward <alan.hayward@arm.com>
7384 David Sherwood <david.sherwood@arm.com>
7386 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
7387 to a scalar_int_mode instead of a machine_mode.
7388 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
7389 (get_best_mode): Return a boolean and use a pointer argument to store
7390 the selected mode. Replace the limit mode parameter with a bit limit.
7391 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
7392 for the values returned by bit_field_mode_iterator::next_mode.
7393 (store_bit_field): Update call to get_best_mode.
7394 (store_fixed_bit_field): Likewise.
7395 (extract_fixed_bit_field): Likewise.
7396 * expr.c (optimize_bitfield_assignment_op): Likewise.
7397 * fold-const.c (optimize_bit_field_compare): Likewise.
7398 (fold_truth_andor_1): Likewise.
7399 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
7400 Update for new type of m_mode.
7401 (get_best_mode): As above.
7403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7404 Alan Hayward <alan.hayward@arm.com>
7405 David Sherwood <david.sherwood@arm.com>
7407 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
7408 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
7409 (store_bit_field): Check is_a <scalar_int_mode> before calling
7410 strict_volatile_bitfield_p.
7411 (extract_bit_field): Likewise.
7413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7414 Alan Hayward <alan.hayward@arm.com>
7415 David Sherwood <david.sherwood@arm.com>
7417 * target.def (cstore_mode): Return a scalar_int_mode.
7418 * doc/tm.texi: Regenerate.
7419 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
7420 * targhooks.h (default_cstore_mode): Likewise.
7421 * targhooks.c (default_cstore_mode): Likewise, using a forced
7423 * expmed.c (emit_cstore): Expect the target of the cstore to be
7426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7427 Alan Hayward <alan.hayward@arm.com>
7428 David Sherwood <david.sherwood@arm.com>
7430 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
7432 (niter_desc): Likewise mode.
7433 (iv_analyze): Add a mode parameter.
7435 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
7436 and change its type to scalar_int_mode.
7437 * loop-iv.c: Update commentary at head of file.
7438 (iv_constant): Pass the mode paraeter before the rtx it describes
7439 and change its type to scalar_int_mode. Remove VOIDmode handling.
7440 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
7441 (iv_extend): Likewise.
7442 (shorten_into_mode): Likewise.
7443 (iv_add): Use scalar_int_mode.
7444 (iv_mult): Likewise.
7445 (iv_shift): Likewise.
7446 (canonicalize_iv_subregs): Likewise.
7447 (get_biv_step_1): Pass the outer_mode parameter before the rtx
7448 it describes and change its mode to scalar_int_mode. Also change
7449 the type of the returned inner_mode to scalar_int_mode.
7450 (get_biv_step): Likewise, turning outer_mode from a pointer
7451 into a direct parameter. Update call to get_biv_step_1.
7452 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
7453 iv_constant and get_biv_step.
7454 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
7455 and change its type to scalar_int_mode. Don't initialise iv->mode
7456 to VOIDmode and remove later checks for its still being VOIDmode.
7457 Update calls to iv_analyze_op and iv_analyze_expr. Check
7458 is_a <scalar_int_mode> when changing the mode under consideration.
7459 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
7460 Update call to iv_analyze_expr.
7461 (iv_analyze_op): Add a mode parameter. Reject subregs whose
7462 inner register is not also a scalar_int_mode. Update call to
7464 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
7465 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
7466 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
7467 separate mode class checks. Update calls to iv_analyze. Remove
7468 fix-up of VOIDmodes after iv_analyze_biv.
7469 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
7470 don't have a scalar_int_mode. Update call to biv_p.
7472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7473 Alan Hayward <alan.hayward@arm.com>
7474 David Sherwood <david.sherwood@arm.com>
7476 * cfgexpand.c (convert_debug_memory_address): Use
7477 as_a <scalar_int_mode>.
7478 * combine.c (expand_compound_operation): Likewise.
7479 (make_extraction): Likewise.
7480 (change_zero_ext): Likewise.
7481 (simplify_comparison): Likewise.
7482 * cse.c (cse_insn): Likewise.
7483 * dwarf2out.c (minmax_loc_descriptor): Likewise.
7484 (mem_loc_descriptor): Likewise.
7485 (loc_descriptor): Likewise.
7486 * expmed.c (init_expmed_one_mode): Likewise.
7487 (synth_mult): Likewise.
7488 (emit_store_flag_1): Likewise.
7489 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
7490 of a comparison with size.
7491 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
7492 (reduce_to_bit_field_precision): Likewise.
7493 * function.c (expand_function_end): Likewise.
7494 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
7495 * loop-doloop.c (doloop_modify): Likewise.
7496 * optabs.c (expand_binop): Likewise.
7497 (expand_unop): Likewise.
7498 (expand_copysign_absneg): Likewise.
7499 (prepare_cmp_insn): Likewise.
7500 (maybe_legitimize_operand): Likewise.
7501 * recog.c (const_scalar_int_operand): Likewise.
7502 * rtlanal.c (get_address_mode): Likewise.
7503 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7504 (simplify_cond_clz_ctz): Likewise.
7505 * tree-nested.c (get_nl_goto_field): Likewise.
7506 * tree.c (build_vector_type_for_mode): Likewise.
7507 * var-tracking.c (use_narrower_mode): Likewise.
7509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7510 Alan Hayward <alan.hayward@arm.com>
7511 David Sherwood <david.sherwood@arm.com>
7513 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
7514 * builtins.c (expand_builtin_signbit): Use it.
7515 * cfgexpand.c (expand_debug_expr): Likewise.
7516 * dojump.c (do_jump): Likewise.
7517 (do_compare_and_jump): Likewise.
7518 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
7519 * expmed.c (make_tree): Likewise.
7520 * expr.c (expand_expr_real_2): Likewise.
7521 (expand_expr_real_1): Likewise.
7522 (try_casesi): Likewise.
7523 * fold-const-call.c (fold_const_call_ss): Likewise.
7524 * fold-const.c (unextend): Likewise.
7525 (extract_muldiv_1): Likewise.
7526 (fold_single_bit_test): Likewise.
7527 (native_encode_int): Likewise.
7528 (native_encode_string): Likewise.
7529 (native_interpret_int): Likewise.
7530 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
7531 * internal-fn.c (expand_addsub_overflow): Likewise.
7532 (expand_neg_overflow): Likewise.
7533 (expand_mul_overflow): Likewise.
7534 (expand_arith_overflow): Likewise.
7535 * match.pd: Likewise.
7536 * stor-layout.c (layout_type): Likewise.
7537 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7538 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
7539 * tree-ssanames.c (get_range_info): Likewise.
7540 * tree-switch-conversion.c (array_value_type) Likewise.
7541 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
7542 (vect_recog_divmod_pattern): Likewise.
7543 (vect_recog_mixed_size_cond_pattern): Likewise.
7544 * tree-vrp.c (extract_range_basic): Likewise.
7545 (simplify_float_conversion_using_ranges): Likewise.
7546 * tree.c (int_fits_type_p): Likewise.
7547 * ubsan.c (instrument_bool_enum_load): Likewise.
7548 * varasm.c (mergeable_string_section): Likewise.
7549 (narrowing_initializer_constant_valid_p): Likewise.
7550 (output_constant): Likewise.
7552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7553 Alan Hayward <alan.hayward@arm.com>
7554 David Sherwood <david.sherwood@arm.com>
7556 * machmode.h (NARROWEST_INT_MODE): New macro.
7557 * expr.c (alignment_for_piecewise_move): Use it instead of
7558 GET_CLASS_NARROWEST_MODE (MODE_INT).
7559 (push_block): Likewise.
7560 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
7562 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7565 Alan Hayward <alan.hayward@arm.com>
7566 David Sherwood <david.sherwood@arm.com>
7568 * postreload.c (move2add_valid_value_p): Change the type of the
7569 mode parameter to scalar_int_mode.
7570 (move2add_use_add2_insn): Add a mode parameter and use it instead
7572 (move2add_use_add3_insn): Likewise.
7573 (reload_cse_move2add): Update accordingly.
7575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7576 Alan Hayward <alan.hayward@arm.com>
7577 David Sherwood <david.sherwood@arm.com>
7579 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
7581 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
7582 * optabs.c (expand_unop): Likewise.
7584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7585 Alan Hayward <alan.hayward@arm.com>
7586 David Sherwood <david.sherwood@arm.com>
7588 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
7589 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
7590 (popcount_loc_descriptor): Likewise.
7591 (bswap_loc_descriptor): Likewise.
7592 (rotate_loc_descriptor): Likewise.
7593 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
7594 calling the functions above.
7596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7597 Alan Hayward <alan.hayward@arm.com>
7598 David Sherwood <david.sherwood@arm.com>
7600 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
7602 (try_combine): Likewise.
7603 (simplify_if_then_else): Likewise.
7604 * cse.c (cse_insn): Likewise.
7605 * dwarf2out.c (mem_loc_descriptor): Likewise.
7606 * emit-rtl.c (gen_lowpart_common): Likewise.
7607 * simplify-rtx.c (simplify_truncation): Likewise.
7608 (simplify_binary_operation_1): Likewise.
7609 (simplify_const_relational_operation): Likewise.
7610 (simplify_ternary_operation): Likewise.
7611 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
7613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7614 Alan Hayward <alan.hayward@arm.com>
7615 David Sherwood <david.sherwood@arm.com>
7617 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
7618 * reload.c (push_reload): Likewise.
7619 (find_reloads): Likewise.
7621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7622 Alan Hayward <alan.hayward@arm.com>
7623 David Sherwood <david.sherwood@arm.com>
7625 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
7626 (make_compound_operation_int): Likewise.
7627 (change_zero_ext): Likewise.
7628 * expr.c (convert_move): Likewise.
7629 (convert_modes): Likewise.
7630 * fwprop.c (forward_propagate_subreg): Likewise.
7631 * loop-iv.c (get_biv_step_1): Likewise.
7632 * optabs.c (widen_operand): Likewise.
7633 * postreload.c (move2add_valid_value_p): Likewise.
7634 * recog.c (simplify_while_replacing): Likewise.
7635 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7636 (simplify_binary_operation_1): Likewise. Remove redundant
7637 mode equality check.
7639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7640 Alan Hayward <alan.hayward@arm.com>
7641 David Sherwood <david.sherwood@arm.com>
7643 * combine.c (combine_simplify_rtx): Add checks for
7644 is_a <scalar_int_mode>.
7645 (simplify_if_then_else): Likewise.
7646 (make_field_assignment): Likewise.
7647 (simplify_comparison): Likewise.
7648 * ifcvt.c (noce_try_bitop): Likewise.
7649 * loop-invariant.c (canonicalize_address_mult): Likewise.
7650 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7653 Alan Hayward <alan.hayward@arm.com>
7654 David Sherwood <david.sherwood@arm.com>
7656 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
7657 is_a <scalar_int_mode> instead of != BLKmode.
7659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7660 Alan Hayward <alan.hayward@arm.com>
7661 David Sherwood <david.sherwood@arm.com>
7663 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
7664 instead of != VOIDmode.
7665 * combine.c (if_then_else_cond): Likewise.
7666 (change_zero_ext): Likewise.
7667 * dwarf2out.c (mem_loc_descriptor): Likewise.
7668 (loc_descriptor): Likewise.
7669 * rtlanal.c (canonicalize_condition): Likewise.
7670 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
7672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7673 Alan Hayward <alan.hayward@arm.com>
7674 David Sherwood <david.sherwood@arm.com>
7676 * simplify-rtx.c (simplify_binary_operation_1): Use
7677 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
7679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7680 Alan Hayward <alan.hayward@arm.com>
7681 David Sherwood <david.sherwood@arm.com>
7683 * wide-int.h (int_traits<unsigned char>) New class.
7684 (int_traits<unsigned short>) Likewise.
7685 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
7686 Use GET_MODE_UNIT_PRECISION and remove redundant test for
7688 * combine.c (set_nonzero_bits_and_sign_copies): Use
7689 is_a <scalar_int_mode>.
7690 (find_split_point): Likewise.
7691 (combine_simplify_rtx): Likewise.
7692 (simplify_logical): Likewise.
7693 (expand_compound_operation): Likewise.
7694 (expand_field_assignment): Likewise.
7695 (make_compound_operation): Likewise.
7696 (extended_count): Likewise.
7697 (change_zero_ext): Likewise.
7698 (simplify_comparison): Likewise.
7699 * dwarf2out.c (scompare_loc_descriptor): Likewise.
7700 (ucompare_loc_descriptor): Likewise.
7701 (minmax_loc_descriptor): Likewise.
7702 (mem_loc_descriptor): Likewise.
7703 (loc_descriptor): Likewise.
7704 * expmed.c (init_expmed_one_mode): Likewise.
7705 * lra-constraints.c (lra_constraint_offset): Likewise.
7706 * optabs.c (prepare_libcall_arg): Likewise.
7707 * postreload.c (move2add_note_store): Likewise.
7708 * reload.c (operands_match_p): Likewise.
7709 * rtl.h (load_extend_op): Likewise.
7710 * rtlhooks.c (gen_lowpart_general): Likewise.
7711 * simplify-rtx.c (simplify_truncation): Likewise.
7712 (simplify_unary_operation_1): Likewise.
7713 (simplify_binary_operation_1): Likewise.
7714 (simplify_const_binary_operation): Likewise.
7715 (simplify_const_relational_operation): Likewise.
7716 (simplify_subreg): Likewise.
7717 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7718 * var-tracking.c (adjust_mems): Likewise.
7719 (prepare_call_arguments): Likewise.
7721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7722 Alan Hayward <alan.hayward@arm.com>
7723 David Sherwood <david.sherwood@arm.com>
7725 * machmode.h (is_int_mode): New fuction.
7726 * combine.c (find_split_point): Use it.
7727 (combine_simplify_rtx): Likewise.
7728 (simplify_if_then_else): Likewise.
7729 (simplify_set): Likewise.
7730 (simplify_shift_const_1): Likewise.
7731 (simplify_comparison): Likewise.
7732 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
7733 * cse.c (notreg_cost): Likewise.
7734 (cse_insn): Likewise.
7735 * cselib.c (cselib_lookup_1): Likewise.
7736 * dojump.c (do_jump_1): Likewise.
7737 (do_compare_rtx_and_jump): Likewise.
7738 * dse.c (get_call_args): Likewise.
7739 * dwarf2out.c (rtl_for_decl_init): Likewise.
7740 (native_encode_initializer): Likewise.
7741 * expmed.c (emit_store_flag_1): Likewise.
7742 (emit_store_flag): Likewise.
7743 * expr.c (convert_modes): Likewise.
7744 (store_field): Likewise.
7745 (expand_expr_real_1): Likewise.
7746 * fold-const.c (fold_read_from_constant_string): Likewise.
7747 * gimple-ssa-sprintf.c (get_format_string): Likewise.
7748 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
7749 * optabs.c (expand_binop): Likewise.
7750 (expand_unop): Likewise.
7751 (expand_abs_nojump): Likewise.
7752 (expand_one_cmpl_abs_nojump): Likewise.
7753 * simplify-rtx.c (mode_signbit_p): Likewise.
7754 (val_signbit_p): Likewise.
7755 (val_signbit_known_set_p): Likewise.
7756 (val_signbit_known_clear_p): Likewise.
7757 (simplify_relational_operation_1): Likewise.
7758 * tree.c (vector_type_mode): Likewise.
7760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7761 Alan Hayward <alan.hayward@arm.com>
7762 David Sherwood <david.sherwood@arm.com>
7764 * machmode.h (smallest_mode_for_size): Fix formatting.
7765 (smallest_int_mode_for_size): New function.
7766 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
7767 instead of smallest_mode_for_size.
7768 * combine.c (make_extraction): Likewise.
7769 * config/arc/arc.c (arc_expand_movmem): Likewise.
7770 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
7771 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7772 * config/s390/s390.c (s390_expand_insv): Likewise.
7773 * config/sparc/sparc.c (assign_int_registers): Likewise.
7774 * config/spu/spu.c (spu_function_value): Likewise.
7775 (spu_function_arg): Likewise.
7776 * coverage.c (get_gcov_type): Likewise.
7777 (get_gcov_unsigned_t): Likewise.
7778 * dse.c (find_shift_sequence): Likewise.
7779 * expmed.c (store_bit_field_1): Likewise.
7780 * expr.c (convert_move): Likewise.
7781 (store_field): Likewise.
7782 * internal-fn.c (expand_arith_overflow): Likewise.
7783 * optabs-query.c (get_best_extraction_insn): Likewise.
7784 * optabs.c (expand_twoval_binop_libfunc): Likewise.
7785 * stor-layout.c (layout_type): Likewise.
7786 (initialize_sizetypes): Likewise.
7787 * targhooks.c (default_get_mask_mode): Likewise.
7788 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
7790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7791 Alan Hayward <alan.hayward@arm.com>
7792 David Sherwood <david.sherwood@arm.com>
7794 * machmode.h (opt_mode::else_blk): New function.
7795 (int_mode_for_mode): Declare.
7796 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
7797 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
7799 * cfgexpand.c (expand_debug_expr): Likewise.
7800 * combine.c (gen_lowpart_or_truncate): Likewise.
7801 (gen_lowpart_for_combine): Likewise.
7802 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
7803 * config/avr/avr.c (avr_to_int_mode): Likewise.
7804 (avr_out_plus_1): Likewise.
7805 (avr_out_plus): Likewise.
7806 (avr_out_round): Likewise.
7807 * config/i386/i386.c (ix86_split_to_parts): Likewise.
7808 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
7809 (s390_expand_vcond): Likewise.
7810 * config/spu/spu.c (spu_split_immediate): Likewise.
7811 (spu_expand_mov): Likewise.
7812 * dse.c (get_stored_val): Likewise.
7813 * expmed.c (store_bit_field_1): Likewise.
7814 (convert_extracted_bit_field): Use int_mode_for_mode instead of
7816 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
7817 (extract_low_bits): Likewise.
7818 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
7819 handling rather than repeating the check.
7820 (emit_group_store): Likewise.
7821 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
7822 * optabs.c (expand_absneg_bit): Likewise.
7823 (expand_copysign_absneg): Likewise.
7824 (expand_copysign_bit): Likewise.
7825 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
7826 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
7827 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
7828 * var-tracking.c (prepare_call_arguments): Likewise.
7829 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
7830 int_mode_for_mode instead of mode_for_size.
7831 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
7833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7834 Alan Hayward <alan.hayward@arm.com>
7835 David Sherwood <david.sherwood@arm.com>
7837 * machmode.h (int_mode_for_size): New function.
7838 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
7839 instead of mode_for_size.
7840 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
7842 * combine.c (expand_field_assignment): Use int_mode_for_size
7843 instead of mode_for_size.
7844 (make_extraction): Likewise.
7845 (simplify_shift_const_1): Likewise.
7846 (simplify_comparison): Likewise.
7847 * dojump.c (do_jump): Likewise.
7848 * dwarf2out.c (mem_loc_descriptor): Likewise.
7849 * emit-rtl.c (init_derived_machine_modes): Likewise.
7850 * expmed.c (flip_storage_order): Likewise.
7851 (convert_extracted_bit_field): Likewise.
7852 * expr.c (copy_blkmode_from_reg): Likewise.
7853 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
7854 * internal-fn.c (expand_mul_overflow): Likewise.
7855 * lower-subreg.c (simple_move): Likewise.
7856 * optabs-libfuncs.c (init_optabs): Likewise.
7857 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7858 * tree.c (vector_type_mode): Likewise.
7859 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
7860 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
7861 * tree-vect-generic.c (expand_vector_parallel): Likewise.
7862 * tree-vect-stmts.c (vectorizable_load): Likewise.
7863 (vectorizable_store): Likewise.
7865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7866 Alan Hayward <alan.hayward@arm.com>
7867 David Sherwood <david.sherwood@arm.com>
7869 * coretypes.h (pod_mode): New type.
7870 (scalar_int_mode_pod): New typedef.
7871 * machmode.h (pod_mode): New class.
7872 (int_n_data_t::m): Change type to scalar_int_mode_pod.
7873 * genmodes.c (emit_mode_int_n): Update accordingly.
7874 * lower-subreg.h (target_lower_subreg): Change type to
7875 scalar_int_mode_pod.
7876 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
7877 scalar_int_mode_pod.
7879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7880 Alan Hayward <alan.hayward@arm.com>
7881 David Sherwood <david.sherwood@arm.com>
7883 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
7884 machine_mode to scalar_int_mode.
7885 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
7886 (rs6000_option_override_internal): Remove cast to int.
7887 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
7888 machine_mode to scalar_int_mode.
7889 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
7890 (rs6000_option_override_internal): Remove cast to int.
7891 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
7892 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
7894 * config/s390/s390.c (s390_expand_builtin): Likewise.
7895 * coretypes.h (scalar_int_mode): New type.
7896 (opt_scalar_int_mode): New typedef.
7897 * machmode.h (scalar_int_mode): New class.
7898 (scalar_int_mode::includes_p): New function.
7899 (byte_mode): Change type to scalar_int_mode.
7900 (word_mode): Likewise.
7901 (ptr_mode): Likewise.
7902 * emit-rtl.c (byte_mode): Likewise.
7903 (word_mode): Likewise.
7904 (ptr_mode): Likewise.
7905 (init_derived_machine_modes): Update accordingly.
7906 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
7907 and MODE_PARTIAL_INT.
7908 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
7909 opt_scalar_int_mode.
7911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7912 Alan Hayward <alan.hayward@arm.com>
7913 David Sherwood <david.sherwood@arm.com>
7915 * target.def (libgcc_floating_mode_supported_p): Take a
7917 * doc/tm.texi: Regenerate.
7918 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
7920 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7921 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
7924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7925 Alan Hayward <alan.hayward@arm.com>
7926 David Sherwood <david.sherwood@arm.com>
7928 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
7929 * doc/tm.texi: Regenerate.
7930 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
7931 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
7932 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
7933 * targhooks.h (default_floatn_mode): Likewise.
7934 * targhooks.c (default_floatn_mode): Likewise.
7935 * tree.c (build_common_tree_nodes): Update accordingly.
7937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7938 Alan Hayward <alan.hayward@arm.com>
7939 David Sherwood <david.sherwood@arm.com>
7941 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
7942 (mode_iterator::iterate_p): Likewise.
7943 (mode_iterator::get_wider): Likewise.
7944 * expr.c (init_expr_target): Use opt_scalar_float_mode.
7946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7947 Alan Hayward <alan.hayward@arm.com>
7948 David Sherwood <david.sherwood@arm.com>
7950 * coretypes.h (opt_scalar_float_mode): New typedef.
7951 * machmode.h (float_mode_for_size): New function.
7952 * emit-rtl.c (double_mode): Delete.
7953 (init_emit_once): Use float_mode_for_size.
7954 * stor-layout.c (layout_type): Likewise.
7955 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
7957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7958 Alan Hayward <alan.hayward@arm.com>
7959 David Sherwood <david.sherwood@arm.com>
7961 * output.h (assemble_real): Take a scalar_float_mode.
7962 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
7963 * config/arm/arm.md (consttable_4): Likewise.
7964 (consttable_8): Likewise.
7965 (consttable_16): Likewise.
7966 * config/mips/mips.md (consttable_float): Likewise.
7967 * config/s390/s390.c (s390_output_pool_entry): Likewise.
7968 * varasm.c (assemble_real): Take a scalar_float_mode.
7969 (output_constant_pool_2): Update accordingly.
7970 (output_constant): Likewise.
7972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7973 Alan Hayward <alan.hayward@arm.com>
7974 David Sherwood <david.sherwood@arm.com>
7976 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
7977 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
7978 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
7979 (native_encode_real): Likewise.
7980 (native_interpret_real): Likewise.
7981 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
7982 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7985 Alan Hayward <alan.hayward@arm.com>
7986 David Sherwood <david.sherwood@arm.com>
7988 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
7989 <scalar_float_mode>. Simplify.
7990 (gen_extend_conv_libfunc): Likewise.
7992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7993 Alan Hayward <alan.hayward@arm.com>
7994 David Sherwood <david.sherwood@arm.com>
7996 * coretypes.h (scalar_float_mode): New type.
7997 * machmode.h (mode_traits::from_int): Use machine_mode if
7998 USE_ENUM_MODES is defined.
7999 (is_a): New function.
8001 (dyn_cast): Likewise.
8002 (scalar_float_mode): New class.
8003 (scalar_float_mode::includes_p): New function.
8004 (is_float_mode): Likewise.
8005 * gdbhooks.py (MachineModePrinter): New class.
8006 (build_pretty_printer): Use it for scalar_float_mode.
8007 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
8008 (format_helper::format_helper): Turn into a template.
8009 * genmodes.c (get_mode_class): New function.
8010 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
8011 or machine_mode if none.
8012 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
8013 as_a <scalar_float_mode>.
8014 * dwarf2out.c (mem_loc_descriptor): Likewise.
8015 (insert_float): Likewise.
8016 (add_const_value_attribute): Likewise.
8017 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8018 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
8019 (expand_unop): Update accordingly.
8020 (expand_abs_nojump): Likewise.
8021 (expand_copysign_absneg): Take a scalar_float_mode.
8022 (expand_copysign_bit): Likewise.
8023 (expand_copysign): Update accordingly.
8025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8026 Alan Hayward <alan.hayward@arm.com>
8027 David Sherwood <david.sherwood@arm.com>
8029 * coretypes.h (opt_mode): New class.
8030 * machmode.h (opt_mode): Likewise.
8031 (opt_mode::else_void): New function.
8032 (opt_mode::require): Likewise.
8033 (opt_mode::exists): Likewise.
8034 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
8035 (GET_MODE_2XWIDER_MODE): Likewise.
8036 (mode_iterator::get_wider): Update accordingly.
8037 (mode_iterator::get_2xwider): Likewise.
8038 (mode_iterator::get_known_wider): Likewise, turning into a template.
8039 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
8040 forcing a wider mode to exist.
8041 * config/cr16/cr16.h (LONG_REG_P): Likewise.
8042 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8043 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
8044 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
8045 * lower-subreg.c (init_lower_subreg): Likewise.
8046 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
8047 on the final iteration.
8048 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
8049 a wider mode exists before asking for a move pattern.
8050 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
8051 forcing a wider mode to exist.
8052 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
8053 returning false if no such mode exists.
8054 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
8055 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
8056 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
8057 Avoid checking for a MODE_INT if we already know the mode is not a
8059 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
8060 forcing a wider mode to exist.
8061 (expmed_mult_highpart_optab): Likewise.
8062 (expmed_mult_highpart): Likewise.
8063 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
8065 * lto-streamer-in.c (lto_input_mode_table): Likewise.
8066 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
8067 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8068 * internal-fn.c (expand_mul_overflow): Update use of
8069 GET_MODE_2XWIDER_MODE.
8070 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8071 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
8072 GET_MODE_WIDER_MODE.
8073 (convert_plusminus_to_widen): Likewise.
8074 * tree-switch-conversion.c (array_value_type): Likewise.
8075 * var-tracking.c (emit_note_insn_var_location): Likewise.
8076 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8077 Return false inside rather than outside the loop if no wider mode
8079 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
8080 and GET_MODE_2XWIDER_MODE
8081 (can_compare_p): Use else_void.
8082 * gdbhooks.py (OptMachineModePrinter): New class.
8083 (build_pretty_printer): Use it for opt_mode.
8085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8086 Alan Hayward <alan.hayward@arm.com>
8087 David Sherwood <david.sherwood@arm.com>
8089 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
8090 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
8092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8093 Alan Hayward <alan.hayward@arm.com>
8094 David Sherwood <david.sherwood@arm.com>
8096 * machmode.h (mode_traits): New structure.
8097 (get_narrowest_mode): New function.
8098 (mode_iterator::start): Likewise.
8099 (mode_iterator::iterate_p): Likewise.
8100 (mode_iterator::get_wider): Likewise.
8101 (mode_iterator::get_known_wider): Likewise.
8102 (mode_iterator::get_2xwider): Likewise.
8103 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
8104 (FOR_EACH_MODE): Likewise.
8105 (FOR_EACH_MODE_FROM): Likewise.
8106 (FOR_EACH_MODE_UNTIL): Likewise.
8107 (FOR_EACH_WIDER_MODE): Likewise.
8108 (FOR_EACH_2XWIDER_MODE): Likewise.
8109 * builtins.c (expand_builtin_strlen): Use new mode iterators.
8110 * combine.c (simplify_comparison): Likewise
8111 * config/i386/i386.c (type_natural_mode): Likewise.
8112 * cse.c (cse_insn): Likewise.
8113 * dse.c (find_shift_sequence): Likewise.
8114 * emit-rtl.c (init_derived_machine_modes): Likewise.
8115 (init_emit_once): Likewise.
8116 * explow.c (hard_function_value): Likewise.
8117 * expmed.c (extract_fixed_bit_field_1): Likewise.
8118 (extract_bit_field_1): Likewise.
8119 (expand_divmod): Likewise.
8120 (emit_store_flag_1): Likewise.
8121 * expr.c (init_expr_target): Likewise.
8122 (convert_move): Likewise.
8123 (alignment_for_piecewise_move): Likewise.
8124 (widest_int_mode_for_size): Likewise.
8125 (emit_block_move_via_movmem): Likewise.
8126 (copy_blkmode_to_reg): Likewise.
8127 (set_storage_via_setmem): Likewise.
8128 (compress_float_constant): Likewise.
8129 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8130 * optabs-query.c (get_best_extraction_insn): Likewise.
8131 * optabs.c (expand_binop): Likewise.
8132 (expand_twoval_unop): Likewise.
8133 (expand_twoval_binop): Likewise.
8134 (widen_leading): Likewise.
8135 (widen_bswap): Likewise.
8136 (expand_parity): Likewise.
8137 (expand_unop): Likewise.
8138 (prepare_cmp_insn): Likewise.
8139 (prepare_float_lib_cmp): Likewise.
8140 (expand_float): Likewise.
8141 (expand_fix): Likewise.
8142 (expand_sfix_optab): Likewise.
8143 * postreload.c (move2add_use_add2_insn): Likewise.
8144 * reg-stack.c (reg_to_stack): Likewise.
8145 * reginfo.c (choose_hard_reg_mode): Likewise.
8146 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8147 * stor-layout.c (mode_for_size): Likewise.
8148 (smallest_mode_for_size): Likewise.
8149 (mode_for_vector): Likewise.
8150 (finish_bitfield_representative): Likewise.
8151 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
8152 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
8153 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8154 * var-tracking.c (prepare_call_arguments): Likewise.
8156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8157 Alan Hayward <alan.hayward@arm.com>
8158 David Sherwood <david.sherwood@arm.com>
8160 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
8161 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
8162 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
8163 * machmode.h (mode_size): Move earlier in file.
8164 (mode_precision): Likewise.
8165 (mode_inner): Likewise.
8166 (mode_nunits): Likewise.
8167 (mode_unit_size): Likewise.
8168 (unit_unit_precision): Likewise.
8169 (mode_wider): Likewise.
8170 (mode_2xwider): Likewise.
8171 (machine_mode): New class.
8172 (mode_to_bytes): New function.
8173 (mode_to_bits): Likewise.
8174 (mode_to_precision): Likewise.
8175 (mode_to_inner): Likewise.
8176 (mode_to_unit_size): Likewise.
8177 (mode_to_unit_precision): Likewise.
8178 (mode_to_nunits): Likewise.
8179 (GET_MODE_SIZE): Use mode_to_bytes.
8180 (GET_MODE_BITSIZE): Use mode_to_bits.
8181 (GET_MODE_PRECISION): Use mode_to_precision.
8182 (GET_MODE_INNER): Use mode_to_inner.
8183 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
8184 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
8185 (GET_MODE_NUNITS): Use mode_to_nunits.
8186 * system.h (ALWAYS_INLINE): New macro.
8187 * config/powerpcspe/powerpcspe-c.c
8188 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
8189 int for arg1_mode and arg2_mode.
8191 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8192 Alan Hayward <alan.hayward@arm.com>
8193 David Sherwood <david.sherwood@arm.com>
8195 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8196 Prefix mode names with E_ in case statements.
8197 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8198 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
8199 (aarch64_split_simd_move): Likewise.
8200 (aarch64_gen_storewb_pair): Likewise.
8201 (aarch64_gen_loadwb_pair): Likewise.
8202 (aarch64_gen_store_pair): Likewise.
8203 (aarch64_gen_load_pair): Likewise.
8204 (aarch64_get_condition_code_1): Likewise.
8205 (aarch64_constant_pool_reload_icode): Likewise.
8206 (get_rsqrte_type): Likewise.
8207 (get_rsqrts_type): Likewise.
8208 (get_recpe_type): Likewise.
8209 (get_recps_type): Likewise.
8210 (aarch64_gimplify_va_arg_expr): Likewise.
8211 (aarch64_simd_container_mode): Likewise.
8212 (aarch64_emit_load_exclusive): Likewise.
8213 (aarch64_emit_store_exclusive): Likewise.
8214 (aarch64_expand_compare_and_swap): Likewise.
8215 (aarch64_gen_atomic_cas): Likewise.
8216 (aarch64_emit_bic): Likewise.
8217 (aarch64_emit_atomic_swap): Likewise.
8218 (aarch64_emit_atomic_load_op): Likewise.
8219 (aarch64_evpc_trn): Likewise.
8220 (aarch64_evpc_uzp): Likewise.
8221 (aarch64_evpc_zip): Likewise.
8222 (aarch64_evpc_ext): Likewise.
8223 (aarch64_evpc_rev): Likewise.
8224 (aarch64_evpc_dup): Likewise.
8225 (aarch64_gen_ccmp_first): Likewise.
8226 (aarch64_gen_ccmp_next): Likewise.
8227 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8228 (alpha_emit_xfloating_libcall): Likewise.
8229 (emit_insxl): Likewise.
8230 (alpha_arg_type): Likewise.
8231 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
8232 (arc_preferred_simd_mode): Likewise.
8233 (arc_secondary_reload): Likewise.
8234 (get_arc_condition_code): Likewise.
8235 (arc_print_operand): Likewise.
8236 (arc_legitimate_constant_p): Likewise.
8237 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8238 * config/arc/arc.md (casesi_load): Likewise.
8239 (casesi_compact_jump): Likewise.
8240 * config/arc/predicates.md (proper_comparison_operator): Likewise.
8241 (cc_use_register): Likewise.
8242 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8243 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8244 (arm_init_iwmmxt_builtins): Likewise.
8245 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
8246 (neon_expand_vector_init): Likewise.
8247 (arm_attr_length_move_neon): Likewise.
8248 (maybe_get_arm_condition_code): Likewise.
8249 (arm_emit_vector_const): Likewise.
8250 (arm_preferred_simd_mode): Likewise.
8251 (arm_output_iwmmxt_tinsr): Likewise.
8252 (thumb1_output_casesi): Likewise.
8253 (thumb2_output_casesi): Likewise.
8254 (arm_emit_load_exclusive): Likewise.
8255 (arm_emit_store_exclusive): Likewise.
8256 (arm_expand_compare_and_swap): Likewise.
8257 (arm_evpc_neon_vuzp): Likewise.
8258 (arm_evpc_neon_vzip): Likewise.
8259 (arm_evpc_neon_vrev): Likewise.
8260 (arm_evpc_neon_vtrn): Likewise.
8261 (arm_evpc_neon_vext): Likewise.
8262 (arm_validize_comparison): Likewise.
8263 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
8264 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
8265 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
8266 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
8267 (c6x_preferred_simd_mode): Likewise.
8268 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
8269 (epiphany_rtx_costs): Likewise.
8270 * config/epiphany/predicates.md (proper_comparison_operator):
8272 * config/frv/frv.c (condexec_memory_operand): Likewise.
8273 (frv_emit_move): Likewise.
8274 (output_move_single): Likewise.
8275 (output_condmove_single): Likewise.
8276 (frv_hard_regno_mode_ok): Likewise.
8277 (frv_matching_accg_mode): Likewise.
8278 * config/h8300/h8300.c (split_adds_subs): Likewise.
8279 (h8300_rtx_costs): Likewise.
8280 (h8300_print_operand): Likewise.
8281 (compute_mov_length): Likewise.
8282 (output_logical_op): Likewise.
8283 (compute_logical_op_length): Likewise.
8284 (compute_logical_op_cc): Likewise.
8285 (h8300_shift_needs_scratch_p): Likewise.
8286 (output_a_shift): Likewise.
8287 (compute_a_shift_length): Likewise.
8288 (compute_a_shift_cc): Likewise.
8289 (expand_a_rotate): Likewise.
8290 (output_a_rotate): Likewise.
8291 * config/i386/i386.c (classify_argument): Likewise.
8292 (function_arg_advance_32): Likewise.
8293 (function_arg_32): Likewise.
8294 (function_arg_64): Likewise.
8295 (function_value_64): Likewise.
8296 (ix86_gimplify_va_arg): Likewise.
8297 (ix86_legitimate_constant_p): Likewise.
8298 (put_condition_code): Likewise.
8299 (split_double_mode): Likewise.
8300 (ix86_avx256_split_vector_move_misalign): Likewise.
8301 (ix86_expand_vector_logical_operator): Likewise.
8302 (ix86_split_idivmod): Likewise.
8303 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
8304 (ix86_build_const_vector): Likewise.
8305 (ix86_build_signbit_mask): Likewise.
8306 (ix86_match_ccmode): Likewise.
8307 (ix86_cc_modes_compatible): Likewise.
8308 (ix86_expand_branch): Likewise.
8309 (ix86_expand_sse_cmp): Likewise.
8310 (ix86_expand_sse_movcc): Likewise.
8311 (ix86_expand_int_sse_cmp): Likewise.
8312 (ix86_expand_vec_perm_vpermi2): Likewise.
8313 (ix86_expand_vec_perm): Likewise.
8314 (ix86_expand_sse_unpack): Likewise.
8315 (ix86_expand_int_addcc): Likewise.
8316 (ix86_split_to_parts): Likewise.
8317 (ix86_vectorize_builtin_gather): Likewise.
8318 (ix86_vectorize_builtin_scatter): Likewise.
8319 (avx_vpermilp_parallel): Likewise.
8320 (inline_memory_move_cost): Likewise.
8321 (ix86_tieable_integer_mode_p): Likewise.
8322 (x86_maybe_negate_const_int): Likewise.
8323 (ix86_expand_vector_init_duplicate): Likewise.
8324 (ix86_expand_vector_init_one_nonzero): Likewise.
8325 (ix86_expand_vector_init_one_var): Likewise.
8326 (ix86_expand_vector_init_concat): Likewise.
8327 (ix86_expand_vector_init_interleave): Likewise.
8328 (ix86_expand_vector_init_general): Likewise.
8329 (ix86_expand_vector_set): Likewise.
8330 (ix86_expand_vector_extract): Likewise.
8331 (emit_reduc_half): Likewise.
8332 (ix86_emit_i387_round): Likewise.
8333 (ix86_mangle_type): Likewise.
8334 (ix86_expand_round_sse4): Likewise.
8335 (expand_vec_perm_blend): Likewise.
8336 (canonicalize_vector_int_perm): Likewise.
8337 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
8338 (expand_vec_perm_1): Likewise.
8339 (expand_vec_perm_interleave3): Likewise.
8340 (expand_vec_perm_even_odd_pack): Likewise.
8341 (expand_vec_perm_even_odd_1): Likewise.
8342 (expand_vec_perm_broadcast_1): Likewise.
8343 (ix86_vectorize_vec_perm_const_ok): Likewise.
8344 (ix86_expand_vecop_qihi): Likewise.
8345 (ix86_expand_mul_widen_hilo): Likewise.
8346 (ix86_expand_sse2_abs): Likewise.
8347 (ix86_expand_pextr): Likewise.
8348 (ix86_expand_pinsr): Likewise.
8349 (ix86_preferred_simd_mode): Likewise.
8350 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
8351 * config/i386/sse.md (*andnot<mode>3): Likewise.
8352 (<mask_codefor><code><mode>3<mask_name>): Likewise.
8353 (*<code><mode>3): Likewise.
8354 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
8355 (ia64_expand_atomic_op): Likewise.
8356 (ia64_arg_type): Likewise.
8357 (ia64_mode_to_int): Likewise.
8358 (ia64_scalar_mode_supported_p): Likewise.
8359 (ia64_vector_mode_supported_p): Likewise.
8360 (expand_vec_perm_broadcast): Likewise.
8361 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
8362 (iq2000_function_arg_advance): Likewise.
8363 (iq2000_function_arg): Likewise.
8364 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
8365 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
8366 (m68k_libcall_value): Likewise.
8367 (m68k_function_value): Likewise.
8368 (sched_attr_op_type): Likewise.
8369 * config/mcore/mcore.c (mcore_output_move): Likewise.
8370 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
8372 (microblaze_function_arg): Likewise.
8373 * config/mips/mips.c (mips16_build_call_stub): Likewise.
8374 (mips_print_operand): Likewise.
8375 (mips_mode_ok_for_mov_fmt_p): Likewise.
8376 (mips_vector_mode_supported_p): Likewise.
8377 (mips_preferred_simd_mode): Likewise.
8378 (mips_expand_vpc_loongson_even_odd): Likewise.
8379 (mips_expand_vec_unpack): Likewise.
8380 (mips_expand_vi_broadcast): Likewise.
8381 (mips_expand_vector_init): Likewise.
8382 (mips_expand_vec_reduc): Likewise.
8383 (mips_expand_msa_cmp): Likewise.
8384 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
8385 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
8386 (cc_flags_for_mode): Likewise.
8387 * config/msp430/msp430.c (msp430_print_operand): Likewise.
8388 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
8389 (nds32_output_casesi_pc_relative): Likewise.
8390 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8391 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
8392 (nvptx_gen_unpack): Likewise.
8393 (nvptx_gen_pack): Likewise.
8394 (nvptx_gen_shuffle): Likewise.
8395 (nvptx_gen_wcast): Likewise.
8396 (nvptx_preferred_simd_mode): Likewise.
8397 * config/pa/pa.c (pa_secondary_reload): Likewise.
8398 * config/pa/predicates.md (base14_operand): Likewise.
8399 * config/powerpcspe/powerpcspe-c.c
8400 (altivec_resolve_overloaded_builtin): Likewise.
8401 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
8403 (rs6000_preferred_simd_mode): Likewise.
8404 (output_vec_const_move): Likewise.
8405 (rs6000_expand_vector_extract): Likewise.
8406 (rs6000_split_vec_extract_var): Likewise.
8407 (reg_offset_addressing_ok_p): Likewise.
8408 (rs6000_legitimate_offset_address_p): Likewise.
8409 (rs6000_legitimize_address): Likewise.
8410 (rs6000_emit_set_const): Likewise.
8411 (rs6000_const_vec): Likewise.
8412 (rs6000_emit_move): Likewise.
8413 (spe_build_register_parallel): Likewise.
8414 (rs6000_darwin64_record_arg_recurse): Likewise.
8415 (swap_selector_for_mode): Likewise.
8416 (spe_init_builtins): Likewise.
8417 (paired_init_builtins): Likewise.
8418 (altivec_init_builtins): Likewise.
8419 (do_load_for_compare): Likewise.
8420 (rs6000_generate_compare): Likewise.
8421 (rs6000_expand_float128_convert): Likewise.
8422 (emit_load_locked): Likewise.
8423 (emit_store_conditional): Likewise.
8424 (rs6000_output_function_epilogue): Likewise.
8425 (rs6000_handle_altivec_attribute): Likewise.
8426 (rs6000_function_value): Likewise.
8427 (emit_fusion_gpr_load): Likewise.
8428 (emit_fusion_p9_load): Likewise.
8429 (emit_fusion_p9_store): Likewise.
8430 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
8431 (fusion_gpr_mem_load): Likewise.
8432 (fusion_addis_mem_combo_load): Likewise.
8433 (fusion_addis_mem_combo_store): Likewise.
8434 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
8435 (fusion_gpr_mem_load): Likewise.
8436 (fusion_addis_mem_combo_load): Likewise.
8437 (fusion_addis_mem_combo_store): Likewise.
8438 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8440 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
8441 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8442 (rs6000_preferred_simd_mode): Likewise.
8443 (output_vec_const_move): Likewise.
8444 (rs6000_expand_vector_extract): Likewise.
8445 (rs6000_split_vec_extract_var): Likewise.
8446 (reg_offset_addressing_ok_p): Likewise.
8447 (rs6000_legitimate_offset_address_p): Likewise.
8448 (rs6000_legitimize_address): Likewise.
8449 (rs6000_emit_set_const): Likewise.
8450 (rs6000_const_vec): Likewise.
8451 (rs6000_emit_move): Likewise.
8452 (rs6000_darwin64_record_arg_recurse): Likewise.
8453 (swap_selector_for_mode): Likewise.
8454 (paired_init_builtins): Likewise.
8455 (altivec_init_builtins): Likewise.
8456 (rs6000_expand_float128_convert): Likewise.
8457 (emit_load_locked): Likewise.
8458 (emit_store_conditional): Likewise.
8459 (rs6000_output_function_epilogue): Likewise.
8460 (rs6000_handle_altivec_attribute): Likewise.
8461 (rs6000_function_value): Likewise.
8462 (emit_fusion_gpr_load): Likewise.
8463 (emit_fusion_p9_load): Likewise.
8464 (emit_fusion_p9_store): Likewise.
8465 * config/rx/rx.c (rx_gen_move_template): Likewise.
8466 (flags_from_mode): Likewise.
8467 * config/s390/predicates.md (s390_alc_comparison): Likewise.
8468 (s390_slb_comparison): Likewise.
8469 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
8470 (s390_vector_mode_supported_p): Likewise.
8471 (s390_cc_modes_compatible): Likewise.
8472 (s390_match_ccmode_set): Likewise.
8473 (s390_canonicalize_comparison): Likewise.
8474 (s390_emit_compare_and_swap): Likewise.
8475 (s390_branch_condition_mask): Likewise.
8476 (s390_rtx_costs): Likewise.
8477 (s390_secondary_reload): Likewise.
8478 (__SECONDARY_RELOAD_CASE): Likewise.
8479 (s390_expand_cs): Likewise.
8480 (s390_preferred_simd_mode): Likewise.
8481 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
8482 * config/sh/sh.c (sh_print_operand): Likewise.
8483 (dump_table): Likewise.
8484 (sh_secondary_reload): Likewise.
8485 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8486 * config/sh/sh.md (casesi_worker_1): Likewise.
8487 (casesi_worker_2): Likewise.
8488 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
8489 (fcc_comparison_operator): Likewise.
8490 * config/sparc/sparc.c (sparc_expand_move): Likewise.
8491 (emit_soft_tfmode_cvt): Likewise.
8492 (sparc_preferred_simd_mode): Likewise.
8493 (output_cbranch): Likewise.
8494 (sparc_print_operand): Likewise.
8495 (sparc_expand_vec_perm_bmask): Likewise.
8496 (vector_init_bshuffle): Likewise.
8497 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8498 (spu_vector_mode_supported_p): Likewise.
8499 (spu_expand_insv): Likewise.
8500 (spu_emit_branch_or_set): Likewise.
8501 (spu_handle_vector_attribute): Likewise.
8502 (spu_builtin_splats): Likewise.
8503 (spu_builtin_extract): Likewise.
8504 (spu_builtin_promote): Likewise.
8505 (spu_expand_sign_extend): Likewise.
8506 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8507 (tilegx_simd_int): Likewise.
8508 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
8509 (tilepro_simd_int): Likewise.
8510 * config/v850/v850.c (const_double_split): Likewise.
8511 (v850_print_operand): Likewise.
8512 (ep_memory_offset): Likewise.
8513 * config/vax/vax.c (vax_rtx_costs): Likewise.
8514 (vax_output_int_move): Likewise.
8515 (vax_output_int_add): Likewise.
8516 (vax_output_int_subtract): Likewise.
8517 * config/visium/predicates.md (visium_branch_operator): Likewise.
8518 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
8519 (visium_print_operand_address): Likewise.
8520 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8521 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
8522 (xtensa_expand_conditional_branch): Likewise.
8523 (xtensa_copy_incoming_a7): Likewise.
8524 (xtensa_output_literal): Likewise.
8525 * dfp.c (decimal_real_maxval): Likewise.
8526 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8529 Alan Hayward <alan.hayward@arm.com>
8530 David Sherwood <david.sherwood@arm.com>
8532 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
8533 (mode_nunits_inline): Likewise.
8534 (mode_inner_inline): Likewise.
8535 (mode_unit_size_inline): Likewise.
8536 (mode_unit_precision_inline): Likewise.
8537 (emit_insn_modes_h): Likewise. Also emit a #define of the
8539 (emit_mode_wider): Add an E_ prefix to mode names.
8540 (emit_mode_complex): Likewise.
8541 (emit_mode_inner): Likewise.
8542 (emit_mode_adjustments): Likewise.
8543 (emit_mode_int_n): Likewise.
8544 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
8545 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
8546 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
8547 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
8548 (CRC32_BUILTIN, ENTRY): Likewise.
8549 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
8550 (aarch64_pop_regs): Likewise.
8551 (aarch64_process_components): Likewise.
8552 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
8553 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
8554 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
8555 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
8556 * config/arm/arm.c (arm_init_libfuncs): Likewise.
8557 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
8559 * config/i386/i386-builtin.def (pcmpestr): Likewise.
8560 (pcmpistr): Likewise.
8561 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
8562 * config/mmix/mmix.c (mmix_output_condition): Likewise.
8563 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8565 * config/rl78/rl78.c (mduc_regs): Likewise.
8566 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8567 (htm_expand_builtin): Likewise.
8568 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
8569 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
8570 * config/xtensa/xtensa.c (print_operand): Likewise.
8571 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
8572 (NUM_MODE_VECTOR_INT): Likewise.
8573 * genoutput.c (null_operand): Likewise.
8574 (output_operand_data): Likewise.
8575 * genrecog.c (print_parameter_value): Likewise.
8576 * lra.c (debug_operand_data): Likewise.
8578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8580 * dwarf2out.c (scompare_loc_descriptor_wide)
8581 (scompare_loc_descriptor_narrow): New functions, split out from...
8582 (scompare_loc_descriptor): ...here.
8583 * expmed.c (emit_store_flag_int): New function, split out from...
8584 (emit_store_flag): ...here.
8586 2017-08-30 Richard Biener <rguenther@suse.de>
8588 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
8589 (dwarf2out_early_finish): Move setting of AT_pubnames from
8590 early debug output to early finish.
8592 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8594 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
8595 and -mdata-region to the assembler.
8597 2017-08-30 Richard Biener <rguenther@suse.de>
8599 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
8601 (gen_subprogram_die): Add DW_AT_object_pointer only early.
8602 (dwarf2out_early_global_decl): Only generate a DIE for the
8603 abstract origin if it doesn't already exist or is a declaration DIE.
8604 (resolve_addr): Do not add the linkage name twice when
8605 generating a stub DIE for the DW_TAG_GNU_call_site target.
8607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8609 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8610 Use machine_mode rather than int for arg1_mode.
8612 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8615 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
8616 that the second argument of the built-in functions to unpack
8617 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
8618 switch statement instead a lot of if statements.
8619 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
8620 Allow 64-bit values to be in Altivec registers as well as
8621 traditional floating point registers.
8622 (pack<mode>, FMOVE128_VSX iterator): Likewise.
8624 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
8626 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
8627 MAX_REGS_PER_ADDRESS == 1.
8629 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
8631 * config/i386/i386.opt (flag_fentry): Do not init to -1.
8632 * config/i386/i386.c (ix86_option_override_internal): Simplify
8633 setting of opts->x_flag_entry.
8635 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8636 Jakub Jelinek <jakub@redhat.com>
8637 Richard Biener <rguenther@suse.de>
8639 PR tree-optimization/81503
8640 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
8641 folded constant fits in the target type; reorder tests for clarity.
8643 2017-08-29 Martin Liska <mliska@suse.cz>
8645 * passes.def: Include pass_lower_switch.
8646 * stmt.c (dump_case_nodes): Remove and move to
8647 tree-switch-conversion.
8648 (case_values_threshold): Likewise.
8649 (expand_switch_as_decision_tree_p): Likewise.
8650 (emit_case_decision_tree): Likewise.
8651 (expand_case): Likewise.
8652 (balance_case_nodes): Likewise.
8653 (node_has_low_bound): Likewise.
8654 (node_has_high_bound): Likewise.
8655 (node_is_bounded): Likewise.
8656 (emit_case_nodes): Likewise.
8657 (struct simple_case_node): New struct.
8658 (add_case_node): Remove.
8659 (emit_case_dispatch_table): Use vector instead of case_list.
8660 (reset_out_edges_aux): Remove.
8661 (compute_cases_per_edge): Likewise.
8662 (expand_case): Build list of simple_case_node.
8663 (expand_sjlj_dispatch_table): Use it.
8664 * tree-switch-conversion.c (struct case_node): Moved from
8665 stmt.c and adjusted.
8666 (emit_case_nodes): Likewise.
8667 (node_has_low_bound): Likewise.
8668 (node_has_high_bound): Likewise.
8669 (node_is_bounded): Likewise.
8670 (case_values_threshold): Likewise.
8671 (reset_out_edges_aux): Likewise.
8672 (compute_cases_per_edge): Likewise.
8673 (add_case_node): Likewise.
8674 (dump_case_nodes): Likewise.
8675 (balance_case_nodes): Likewise.
8676 (expand_switch_as_decision_tree_p): Likewise.
8677 (emit_jump): Likewise.
8678 (emit_case_decision_tree): Likewise.
8679 (try_switch_expansion): Likewise.
8680 (do_jump_if_equal): Likewise.
8681 (emit_cmp_and_jump_insns): Likewise.
8682 (fix_phi_operands_for_edge): New function.
8683 (record_phi_operand_mapping): Likewise.
8684 (class pass_lower_switch): New pass.
8685 (pass_lower_switch::execute): New function.
8686 (make_pass_lower_switch): Likewise.
8687 (conditional_probability):
8688 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
8689 * tree-pass.h: Add make_pass_lower_switch.
8691 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8694 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
8697 2017-08-29 Richard Biener <rguenther@suse.de>
8699 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
8700 we do not add a DW_AT_inline attribute twice.
8701 (gen_subprogram_die): Remove code setting DW_AT_inline on
8702 DECL_ABSTRACT_P nodes.
8704 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
8706 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
8707 calls to internal functions.
8708 (gimplify_modify_expr): Likewise.
8709 * tree-call-cdce.c (use_internal_fn): Likewise.
8710 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
8711 (convert_to_divmod): Set the nothrow flag.
8712 * tree-if-conv.c (predicate_mem_writes): Likewise.
8713 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8714 (vectorizable_call): Likewise.
8715 (vectorizable_store): Likewise.
8716 (vectorizable_load): Likewise.
8717 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
8718 (vect_recog_mask_conversion_pattern): Likewise.
8720 2017-08-29 Martin Liska <mliska@suse.cz>
8723 * gcc.c (driver_handle_option): Add new argument.
8724 * opts-common.c (handle_option): Pass
8725 target_option_override_hook.
8726 * opts-global.c (lang_handle_option): Add new option.
8727 (set_default_handlers): Add new argument.
8728 (decode_options): Likewise.
8729 * opts.c (target_handle_option): Likewise.
8730 (common_handle_option): Call target_option_override_hook.
8731 * opts.h (struct cl_option_handler_func): Add hook for
8732 target option override.
8733 (struct cl_option_handlers): Likewise.
8734 (set_default_handlers): Add new argument.
8735 (decode_options): Likewise.
8736 (common_handle_option): Likewise.
8737 (target_handle_option): Likewise.
8738 * toplev.c (toplev::main): Pass targetm.target_option.override
8741 2017-08-29 Richard Biener <rguenther@suse.de>
8742 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8744 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
8745 life to the active subtree.
8747 2017-08-28 Jeff Law <law@redhat.com>
8749 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
8750 derive_equivalences.
8751 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
8753 (edge_info::derive_equivalences): New private member function
8755 * tree-ssa-dom.c (class edge_info): Changed from a struct
8756 to a class. Add ctor/dtor, methods and data members.
8757 (edge_info::edge_info): Renamed from allocate_edge_info.
8758 Initialize additional members.
8759 (edge_info::~edge_info): New.
8760 (free_dom_edge_info): Delete the edge info.
8761 (record_edge_info): Use new class & associated member functions.
8762 Tighten forms for testing for edge equivalences.
8763 (record_temporary_equivalences): Iterate over the simple
8764 equivalences rather than assuming there's only one per edge.
8765 (cprop_into_successor_phis): Iterate over the simple
8766 equivalences rather than assuming there's only one per edge.
8767 (optimize_stmt): Use operand_equal_p rather than pointer
8768 equality for mini-DSE code.
8770 2017-08-28 Nathan Sidwell <nathan@acm.org>
8772 * gcc.c (execute): Fold SIGPIPE handling into switch
8773 statement. Adjust internal error message.
8775 2017-08-28 Richard Biener <rguenther@suse.de>
8778 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
8779 Do nothing for removed DIEs.
8781 2017-08-28 Richard Biener <rguenther@suse.de>
8783 PR tree-optimization/81977
8784 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
8787 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
8790 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
8791 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
8792 using targetm.gen_mem_thread_fence.
8794 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
8797 * config/i386/i386.md (*<btsc><mode>): Change operand 2
8798 predicate to register_operand. Reorder operands.
8799 (*btr<mode>): Ditto.
8800 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
8801 (*btr<mode>_mask): Ditto.
8803 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
8805 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
8806 * config/rs6000/xmmintrin.h: New file.
8807 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
8809 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8812 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
8813 parameter and_insn and return it.
8814 (recombine_lvx_pattern): Insert a copy to ensure availability of
8815 the base register of the copied masking operation at the point of
8816 the instruction replacement.
8817 (recombine_stvx_pattern): Likewise.
8819 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
8821 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
8822 undocumented switches.
8823 (-mpower9-dform-vector): Likewise.
8824 (-mpower9-dform): Likewise.
8825 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
8826 comments to delete references to -mpower9-dform* switches.
8827 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
8828 Delete reference to -mpower9-dform* switches, test for
8829 -mpower9-vector instead.
8830 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
8831 (OTHER_P9_VECTOR_MASKS): Likewise.
8832 (POWERPC_MASKS): Likewise.
8833 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
8834 tests against -mpower9-dform* to -mpower9-vector. Delete code
8835 that checked for -mpower9-dform* consistancy with other options.
8836 Add test for -mpower9-misc to enable other power9 switches.
8837 (rs6000_init_hard_regno_mode_ok): Likewise.
8838 (rs6000_option_override_internal): Likewise.
8839 (rs6000_emit_prologue): Likewise.
8840 (rs6000_emit_epilogue): Likewise.
8841 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
8842 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
8843 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
8845 (emit_fusion_p9_store): Likewise.
8846 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
8847 resetting these macros if the assembler does not support ISA 3.0
8849 (TARGET_P9_DFORM_VECTOR): Likewise.
8850 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
8851 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
8853 2017-08-25 Alan Modra <amodra@gmail.com>
8855 PR rtl-optimization/81747
8856 * cse.c (cse_extended_basic_block): Don't attempt to record
8857 equivalences for degenerate conditional branches that branch
8858 to their fall-through.
8860 2017-08-24 Martin Sebor <msebor@redhat.com>
8863 * gimple-fold.c (size_must_be_zero_p): New function.
8864 (gimple_fold_builtin_memory_op): Call it.
8866 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
8868 * config/rs6000/mm_malloc.h: New file.
8870 2017-08-24 Bin Cheng <bin.cheng@arm.com>
8872 PR tree-optimization/81913
8873 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
8874 analysis when either IVs in condition can wrap.
8876 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
8878 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
8879 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
8881 2017-08-24 Richard Biener <rguenther@suse.de>
8884 * targhooks.c (default_target_can_inline_p): Properly
8885 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
8886 is present and always compare.
8887 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
8888 infer -mfpmath=sse from TARGET_SSE_P.
8889 (ix86_can_inline_p): Properly use target_option_default_node when
8890 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
8892 2017-08-24 Richard Biener <rguenther@suse.de>
8895 * dwarf2out.c (output_die): Handle flag_generate_offload like
8897 (output_comp_unit): Likewise.
8898 (gen_array_type_die): Likewise.
8899 (dwarf2out_early_finish): Likewise.
8900 (note_variable_value_in_expr): Likewise.
8901 (dwarf2out_finish): Likewise. Adjust assert.
8902 * cgraphunit.c (symbol_table::compile): Move setting of
8903 flag_generate_offload earlier ...
8904 (symbol_table::finalize_compilation_unit): ... here, before
8905 early debug finalization.
8907 2017-08-24 Richard Biener <rguenther@suse.de>
8909 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
8910 and ipa-fnsummary.h.
8911 (ix86_can_inline_p): When ix86_fpmath flags do not match
8912 check whether the callee uses FP math at all.
8914 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
8917 * tree-ssanames.c (get_nonzero_bits): Use element_precision
8918 instead of TYPE_PRECISION.
8920 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
8921 Alan Hayward <alan.hayward@arm.com>
8922 David Sherwood <david.sherwood@arm.com>
8924 * combine.c (make_extraction): Use subreg_offset_from_lsb.
8926 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
8928 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
8930 (ix86_frame::stack_realign_allocate): New field.
8931 (struct machine_frame_state): Modify comments.
8932 (machine_frame_state::sp_realigned_fp_end): New field.
8933 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
8935 (sp_valid_at): Add assertion to assure no attempt to access invalid
8936 offset of a realigned stack.
8937 (fp_valid_at): Likewise.
8938 (choose_baseaddr): Modify comments.
8939 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
8940 ix86_expand_prologue.
8941 (ix86_expand_prologue): Modify stack realignment and allocation.
8942 (ix86_expand_epilogue): Modify comments.
8943 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
8944 avx2_runtime, avx512f, and avx512f_runtime.
8946 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
8948 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
8949 (mstackrealign): Do not init to -1.
8950 * config/i386/i386.c (ix86_option_override_internal):
8951 Check opts_set, not opts when setting default value of
8952 opts->x_ix86_force_align_arg_pointer.
8954 2017-08-23 Richard Biener <rguenther@suse.de>
8956 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
8957 lang_hooks.decl_printable_name.
8958 * print-rtl-function.c (print_rtx_function): Likewise.
8959 * tree-pretty-print.c (dump_function_header): Likewise.
8961 2017-08-23 Richard Biener <rguenther@suse.de>
8964 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
8965 -g0 at compile-time.
8967 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8970 * doc/sourcebuild.texi (Other hardware attributes):
8973 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8976 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
8977 Fix single-use check.
8979 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
8981 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
8983 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8985 * doc/install.texi: Modify to add more details on running selected
8988 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8990 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
8991 is combined with -mabi=ms.
8992 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
8995 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8997 PR tree-optimization/81488
8998 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
8999 and cached_basis fields.
9000 (MAX_SPREAD): New constant.
9001 (alloc_cand_and_find_basis): Initialize new fields.
9002 (clear_visited): New function.
9003 (create_phi_basis_1): Rename from create_phi_basis, set visited
9004 and cached_basis fields.
9005 (create_phi_basis): New wrapper function.
9006 (phi_add_costs_1): Rename from phi_add_costs, add spread
9007 parameter, set visited field, short-circuit when limits reached.
9008 (phi_add_costs): New wrapper function.
9009 (record_phi_increments_1): Rename from record_phi_increments, set
9011 (record_phi_increments): New wrapper function.
9012 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
9013 (phi_incr_cost): New wrapper function.
9014 (all_phi_incrs_profitable_1): Rename from
9015 all_phi_incrs_profitable, set visited field.
9016 (all_phi_incrs_profitable): New wrapper function.
9018 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9019 Alan Hayward <alan.hayward@arm.com>
9020 David Sherwood <david.sherwood@arm.com>
9022 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
9023 that takes the outer and inner modes.
9024 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
9025 comparison as the canonical test for a paradoxical subreg.
9026 * combine.c (simplify_set): Use paradoxical_subreg_p.
9027 (make_extraction): Likewise.
9028 (force_to_mode): Likewise.
9029 (rtx_equal_for_field_assignment_p): Likewise.
9030 (gen_lowpart_for_combine): Likewise.
9031 (simplify_comparison): Likewise.
9032 * cse.c (equiv_constant): Likewise.
9033 * expmed.c (store_bit_field_1): Likewise.
9034 * final.c (alter_subreg): Likewise.
9035 * fwprop.c (propagate_rtx): Likewise.
9036 (forward_propagate_subreg): Likewise.
9037 * ira-conflicts.c (ira_build_conflicts): Likewise.
9038 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
9039 * lra-constraints.c (curr_insn_transform): Likewise.
9040 (split_reg): Likewise.
9041 * lra-eliminations.c (move_plus_up): Likewise.
9042 (lra_eliminate_regs_1): Likewise.
9043 * recog.c (general_operand): Likewise.
9044 * ree.c (combine_reaching_defs): Likewise.
9045 * reload.c (push_reload): Likewise.
9046 (find_reloads): Likewise.
9047 * reload1.c (elimination_effects): Likewise.
9048 (compute_reload_subreg_offset): Likewise.
9049 (choose_reload_regs): Likewise.
9050 * rtlanal.c (subreg_lsb_1): Likewise.
9051 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9052 (simplify_subreg): Likewise.
9053 * var-tracking.c (track_loc_p): Likewise.
9054 * emit-rtl.c (byte_lowpart_offset): Likewise.
9055 (paradoxical_subreg_p): Delete out-of-line definition.
9057 2017-08-22 Jeff Law <law@redhat.com>
9059 PR tree-optimization/81741
9060 PR tree-optimization/71947
9061 * tree-ssa-dom.c: Include tree-inline.h.
9062 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
9063 equivalences if one is more expensive to compute than the other.
9064 * tree-ssa-scopedtables.h (class const_or_copies): Make
9065 record_const_or_copy_raw method private.
9066 (class avail_exprs_stack): New method simplify_binary_operation.
9067 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
9068 avail_exprs_stack::simplify_binary_operation as needed.
9069 (avail_exprs_stack::simplify_binary_operation): New function.
9071 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
9073 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
9074 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
9075 (DOT_SYMBOLS): Likewise.
9076 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
9077 (RELOCATABLE_NEEDS_FIXUP): Likewise.
9078 (RS6000_ABI_NAME): Likewise.
9079 (TARGET_CMODEL): Likewise.
9080 (TOC_SECTION_ASM_OP): Likewise.
9081 (SET_CMODEL): New macro.
9082 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
9084 2017-08-22 Richard Biener <rguenther@suse.de>
9086 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
9087 to allow for free-lang-data replacements similar to verify_type_variant.
9089 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9090 Alan Hayward <alan.hayward@arm.com>
9091 David Sherwood <david.sherwood@arm.com>
9093 * config/aarch64/aarch64.md (casesi): Use DImode rather than
9094 VOIDmode for the LABEL_REF.
9096 2017-08-22 Richard Biener <rguenther@suse.de>
9098 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
9100 2017-08-22 Richard Biener <rguenther@suse.de>
9102 * common.opt (feliminate-dwarf2-dups): Ignore.
9103 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
9104 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
9105 same_die_p_wrap, compute_section_prefix,
9106 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
9107 (comdat_symbol_id, comdat_symbol_number): Likewise.
9108 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
9110 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
9111 (output_die): Mark unreachable path unreachable.
9112 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
9113 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
9114 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
9115 (dwarf2out_early_finish): Likewise.
9117 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
9119 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
9121 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
9124 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
9125 not VAR_P. Filter attribute warnings with OPT_Wattributes.
9126 (avr_attribute_table) <io, io_low, address>: Initialize
9127 .decl_required with true.
9129 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
9131 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
9132 undocumented debugging options.
9133 (-mvsx-scalar-double): Likewise.
9134 (-mallow-df-permute): Likewise.
9135 (-mvectorize-builtins): Likewise.
9136 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9137 (rs6000_builtin_vectorized_function): Likewise.
9138 (rs6000_builtin_md_vectorized_function): Likewise.
9139 (rs6000_opt_vars): Likewise.
9141 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
9144 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
9145 (*btrq_imm): Rename from *btrq.
9146 (*btcq_imm): Rename from *btcq.
9147 (btsc): New code attribute.
9148 (*<btsc><mode>): New insn pattern.
9149 (*btr<mode>): Ditto.
9150 (*<btsc><mode>_mask): New insn_and_split pattern.
9151 (*btr<mode>_mask): Ditto.
9153 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9154 Alan Hayward <alan.hayward@arm.com>
9155 David Sherwood <david.sherwood@arm.com>
9157 * function.c (pad_below): Simplify padding calculation.
9159 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9160 Alan Hayward <alan.hayward@arm.com>
9161 David Sherwood <david.sherwood@arm.com>
9163 * target.def (function_prologue): Remove frame size argument.
9164 (function_epilogue): Likewise.
9165 * doc/tm.texi: Regenerate.
9166 * final.c (final_start_function): Update call to function_prologue.
9167 (final_end_function): Update call to function_epilogue.
9168 (default_function_pro_epilogue): Remove frame size argument.
9169 * output.h (default_function_pro_epilogue): Likewise.
9170 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
9171 (arm_output_function_prologue): Likewise.
9172 * config/frv/frv.c (frv_function_prologue): Likewise.
9173 (frv_function_epilogue): Likewise.
9174 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
9175 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
9176 (ia64_output_function_epilogue): Likewise.
9177 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
9178 (m32r_output_function_epilogue): Likewise.
9179 * config/microblaze/microblaze.c (microblaze_function_prologue)
9180 (microblaze_function_epilogue): Likewise.
9181 * config/mips/mips.c (mips_output_function_prologue): Likewise.
9182 (mips_output_function_epilogue): Likewise.
9183 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
9184 (mmix_target_asm_function_epilogue): Likewise.
9185 * config/msp430/msp430.c (msp430_start_function): Likewise.
9186 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
9187 (nds32_asm_function_epilogue): Likewise.
9188 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
9189 * config/pa/pa.c (pa_output_function_prologue): Likewise.
9190 (pa_output_function_epilogue): Likewise.
9191 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
9192 (rs6000_output_function_epilogue): Likewise.
9193 * config/rl78/rl78.c (rl78_start_function): Likewise.
9194 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
9195 (rs6000_output_function_epilogue): Likewise.
9196 * config/rx/rx.c (rx_output_function_prologue): Likewise.
9197 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
9198 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
9199 (sparc_asm_function_epilogue): Likewise.
9201 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9203 * tree.h (type_has_mode_precision_p): New function.
9204 * convert.c (convert_to_integer_1): Use it.
9205 * expr.c (expand_expr_real_2): Likewise.
9206 (expand_expr_real_1): Likewise.
9207 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
9208 * match.pd: Likewise.
9209 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
9210 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
9211 * tree-tailcall.c (process_assignment): Likewise.
9212 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9213 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
9214 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
9215 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9216 (vectorizable_assignment): Likewise.
9217 (vectorizable_shift): Likewise.
9218 (vectorizable_operation): Likewise.
9219 * tree-vrp.c (register_edge_assert_for_2): Likewise.
9221 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
9223 * match.pd: Add pow (C, x) simplification.
9225 2017-08-21 Richard Biener <rguenther@suse.de>
9227 PR tree-optimization/81900
9228 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
9229 for blocks with abnormal predecessors.
9230 (compute_antic): Do not set visited flag prematurely.
9232 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
9235 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
9237 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9239 * stor-layout.h (vector_type_mode): Move to...
9240 * tree.h (vector_type_mode): ...here.
9241 * stor-layout.c (vector_type_mode): Move to...
9242 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
9244 2017-08-21 Richard Biener <rguenther@suse.de>
9246 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
9247 register_external_die hooks.
9248 (debug_false_tree_charstarstar_uhwistar): Declare.
9249 (debug_nothing_tree_charstar_uhwi): Likewise.
9250 * debug.c (do_nothing_debug_hooks): Adjust.
9251 (debug_false_tree_charstarstar_uhwistar): New do nothing.
9252 (debug_nothing_tree_charstar_uhwi): Likewise.
9253 * dbxout.c (dbx_debug_hooks): Adjust.
9254 (xcoff_debug_hooks): Likewise.
9255 * sdbout.c (sdb_debug_hooks): Likewise.
9256 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9257 * dwarf2out.c (macinfo_label_base): New global.
9258 (dwarf2out_register_external_die): New function for the
9259 register_external_die hook.
9260 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
9261 (dwarf2_debug_hooks): Use them.
9262 (dwarf2_lineno_debug_hooks): Adjust.
9263 (struct die_struct): Add with_offset flag.
9264 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
9265 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
9266 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
9267 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
9268 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
9269 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
9270 defining section names for the early LTO debug variants.
9271 (reset_indirect_string): New helper.
9272 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
9273 (print_dw_val): Add support for offsetted symbol references.
9274 (get_ultimate_context): Split out from is_cxx.
9275 (is_cxx): Use get_ultimate_context.
9276 (is_fortran): Add decl overload.
9277 (compute_comp_unit_symbol): Split out worker from
9278 compute_section_prefix.
9279 (compute_section_prefix): Call compute_comp_unit_symbol and
9280 set comdat_type_p here.
9281 (output_die): Skip DIE symbol output for the LTO added one.
9282 Handle DIE symbol references with offset.
9283 (output_comp_unit): Guard section name mangling properly.
9284 For LTO debug sections emit a symbol at the section beginning
9285 which we use to refer to its DIEs.
9286 (add_abstract_origin_attribute): For DIEs registered via
9287 dwarf2out_register_external_die directly refer to the early
9288 DIE rather than indirectly through the shadow one we created.
9289 Remove obsolete call to dwarf2out_abstract_function for
9290 non-function/block origins.
9291 (gen_array_type_die): When generating early LTO debug do
9292 not emit DW_AT_string_length.
9293 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
9294 late when in LTO. As suggested place a gcc_unreachable for
9295 the DECL_ABSTRACT_P case.
9296 (gen_subprogram_die): Avoid another specification DIE
9297 for early built declarations/definitions for the late LTO case.
9298 (gen_variable_die): Add type references for late duplicated VLA dies
9300 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
9301 we have the abstract instance already.
9302 (process_scope_var): Adjust decl DIE contexts in LTO which
9303 first puts them in limbo.
9304 (gen_decl_die): Do not generate type DIEs late apart from
9305 types for VLAs or for decls we do not yet have a DIE. Do not
9306 call dwarf2out_abstract_function late.
9307 (dwarf2out_early_global_decl): Make sure to create DIEs
9308 for abstract instances of a decl first.
9309 (dwarf2out_late_global_decl): Adjust comment.
9310 (output_macinfo_op): With multiple macro sections use
9311 macinfo_label_base to distinguish labels.
9312 (output_macinfo): Likewise. Update macinfo_label_base.
9313 Pass in the line info label.
9314 (note_variable_value_in_expr): When generating LTO resolve
9315 all variable values here by generating DIEs as needed.
9316 (init_sections_and_labels): Add early LTO debug flag parameter
9317 and generate different sections and names if set. Add generation
9318 counter for the labels so we can have multiple of them.
9319 (reset_dies): Helper to allow DIEs to be output multiple times.
9320 (dwarf2out_finish): When outputting DIEs to the fat part of an
9321 LTO object first reset DIEs.
9322 (dwarf2out_early_finish): Output early DIEs when generating LTO.
9323 (modified_type_die): Check for decl_ultimate_origin being self
9325 (gen_type_die_with_usage): Likewise.
9326 (gen_typedef_die): Allow decl_ultimate_origin being self.
9327 (set_decl_abstract_flags): Remove.
9328 (set_block_abstract_flags): Likewise.
9329 (dwarf2out_abstract_function): Treat the early generated DIEs
9330 as the abstract copy and only add DW_AT_inline and
9331 DW_AT_artificial here and call set_decl_origin_self.
9332 If the DIE has an abstract origin don't do anything.
9333 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
9334 if we have none yet (Go fails to build one, PR78628).
9335 (variably_modified_type_p): Prevent endless recursion for Ada
9336 cyclic pointer types.
9337 * lto-streamer-in.c: Include debug.h.
9338 (dref_queue): New global.
9339 (lto_read_tree_1): Stream in DIE references.
9340 (lto_input_tree): Register DIE references.
9341 (input_function): Stream DECL_DEBUG_ARGS.
9342 * lto-streamer-out.c: Include debug.h.
9343 (lto_write_tree_1): Output DIE references.
9344 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
9345 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
9346 (output_function): Stream DECL_DEBUG_ARGS.
9347 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
9348 Stream DECL_ABSTRACT_ORIGIN.
9349 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
9350 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
9351 DECL_CONTEXT for file-scope decls.
9352 * lto-streamer.h (struct dref_entry): Declare.
9353 (dref_queue): Likewise.
9354 * cfgexpand.c (pass_expand::execute): Do not call the
9355 outlining_inline_function hook here.
9356 * lto-wrapper.c (debug_obj): New global.
9357 (tool_cleanup): Unlink it if required.
9358 (debug_objcopy): New function.
9359 (run_gcc): Handle early debug sections in the IL files by
9360 extracting them to separate files, partially linkin them and
9361 feeding the result back as result to the linker.
9362 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
9363 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
9364 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
9365 sections into a separate segment.
9366 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
9368 (darwin_asm_dwarf_section): Likewise.
9369 (darwin_asm_output_dwarf_offset): Likewise.
9370 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
9372 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9373 Alan Hayward <alan.hayward@arm.com>
9374 David Sherwood <david.sherwood@arm.com>
9376 * read-md.h (md_reader::record_potential_iterator_use): Replace
9377 pointer argument with an rtx and an index.
9378 * read-rtl.c (iterator_group::apply_iterator): Likewise.
9379 (apply_mode_iterator): Likewise.
9380 (apply_code_iterator): Likewise.
9381 (apply_int_iterator): Likewise.
9382 (apply_subst_iterator): Likewise.
9383 (record_iterator_use): Likewise.
9384 (record_attribute_use): Likewise.
9385 (md_reader::record_potential_iterator_use): Likewise. Update calls
9386 to record_iterator_use and apply_iterator.
9387 (iterator_use): Replace ptr with x and index.
9388 (attribute_use): Likewise.
9389 (apply_attribute_uses): Update calls to apply_iterator.
9390 (apply_iterators): Likewise. Update initialization of iterator_use.
9391 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
9392 and record_potential_iterator_use.
9393 (rtx_reader::read_rtx_operand): Likewise.
9395 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9396 Alan Hayward <alan.hayward@arm.com>
9397 David Sherwood <david.sherwood@arm.com>
9399 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
9402 2017-08-21 Richard Biener <rguenther@suse.de>
9405 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
9406 at struct end conservatively when comparing common bases.
9408 2017-08-21 Richard Biener <rguenther@suse.de>
9410 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
9411 (mem_ref_in_stmt): Remove.
9412 (determine_max_movement): Use ref index to get at the reference.
9413 (invariantness_dom_walker::before_dom_children): Deal with
9414 lim data already initialized.
9415 (gather_mem_refs_stmt): Initialize lim data and record ref index.
9417 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9419 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
9420 (TARGET_ISA_ROUND): Ditto.
9421 (TARGET_ROUND): Ditto.
9422 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
9423 * config/i386/i386.md: Ditto.
9424 * config/i386/sse.md: Ditto.
9425 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
9426 with OPTION_MASK_ISA_SSE4_1.
9428 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9431 * doc/extend.texi (x86 Built-in Functions): Correct the name of
9432 __builtin_ia32_lzcnt_u16.
9434 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9437 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
9438 (rs6000_set_current_function): Rewrite function to use it.
9440 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
9443 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
9444 and TYPE_WARN_IF_NOT_ALIGN.
9445 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
9446 (handle_warn_if_not_align): New.
9447 (place_union_field): Call handle_warn_if_not_align.
9448 (place_field): Call handle_warn_if_not_align.
9449 Copy TYPE_WARN_IF_NOT_ALIGN.
9450 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
9451 (layout_type): Likewise.
9452 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
9454 (tree_decl_common): Add warn_if_not_align.
9455 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
9456 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
9457 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
9458 (DECL_WARN_IF_NOT_ALIGN): Likewise.
9459 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
9460 * doc/extend.texi: Document warn_if_not_aligned attribute.
9461 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
9463 2017-08-17 Martin Liska <mliska@suse.cz>
9466 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
9467 (get_data_dependence): Use it as pointer type.
9468 (distribute_loop): Likewise.
9470 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9472 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
9473 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
9474 (p8_vmrgow_<mode>_direct): New define_insn.
9475 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
9476 handle endianness for vmrgew and vmrgow permute patterns.
9478 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9480 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
9481 * config/rs6000/rs6000-cpus.def: Remove comment.
9482 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
9483 (POWERPC_MASKS): Likewise.
9484 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
9485 use of TARGET_VSX_TIMODE.
9486 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
9487 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
9488 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
9489 (rs6000_option_override_internal): Remove dead code.
9490 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
9491 (rs6000_legitimize_reload_address): Likewise.
9492 (rs6000_legitimate_address_p): Likewise.
9493 (rs6000_opt_masks): Delete "vsx-timode".
9494 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
9495 from function comment.
9496 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
9497 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
9498 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
9500 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
9501 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
9502 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
9505 (define_peephole2): Likewise.
9507 2017-08-17 Martin Sebor <msebor@redhat.com>
9510 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
9511 past the end of an array.
9512 (test_pp_format): Add test cases.
9514 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
9516 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
9517 missing ECF_NOTHROW flags.
9519 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9522 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
9523 operands residing in integer registers.
9524 (*vsx_le_perm_load_<mode>): Likewise.
9525 (*vsx_le_perm_store_<mode>): Likewise.
9526 (define_peephole2): Add peepholes to optimize the above.
9528 2017-08-17 Marek Polacek <polacek@redhat.com>
9531 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
9532 to mimic what shorten_compare did. Change the return type to bool.
9533 (fold_cond_expr_with_comparison): Update call to
9534 operand_equal_for_comparison_p.
9535 (fold_ternary_loc): Likewise.
9537 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
9539 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
9541 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
9543 2017-08-17 Richard Biener <rguenther@suse.de>
9545 * tree-ssa-structalias.c (solve_graph): When propagating
9546 to successors update the graphs succ edges and avoid duplicate work.
9548 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
9551 * config/i386/i386.c (ix86_option_override_internal): Save target
9552 specific options after ix86_stack_protector_guard_reg was changed.
9554 2017-08-17 Richard Biener <rguenther@suse.de>
9556 PR tree-optimization/81827
9557 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
9558 (new_var_info): Initialize it conservatively.
9559 (get_call_vi): Mark register vars.
9560 (new_scalar_tmp_constraint_exp): Likewise.
9561 (handle_rhs_call): Likewise.
9562 (handle_const_call): Likewise.
9563 (create_function_info_for): Likewise.
9564 (solve_constraints): Sort varinfos to separate register from
9565 non-register vars to pack points-to solution bitmaps during
9568 2017-08-17 Marek Polacek <polacek@redhat.com>
9570 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
9572 2017-08-17 Richard Biener <rguenther@suse.de>
9574 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
9575 to true when overflow is undefined and we saturated the result.
9577 2017-08-17 Alan Modra <amodra@gmail.com>
9580 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
9581 Don't use store multiple if only one reg needs saving.
9582 (interesting_frame_related_regno): New function.
9583 (rs6000_frame_related): Don't emit frame info for regs that
9585 (rs6000_emit_epilogue): Likewise.
9587 2017-08-16 Nathan Sidwell <nathan@acm.org>
9589 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
9590 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
9591 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
9592 * tree.c (free_lang_data_in_type): Use else-if chain. Always
9593 clear TYPE_LANG_1. Remove obsolete member-function stripping.
9594 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
9595 (verify_type): Adjust for TYPE_BINFO move.
9596 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
9597 process TYPE_BINFO directly.
9598 (hash_tree): Likewise.
9599 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9601 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9604 2017-08-16 David Malcolm <dmalcolm@redhat.com>
9606 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
9608 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
9611 * config/i386/i386.md (*anddi_1_btr): Change predicates of
9612 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
9613 Add ix86_binary_operator_ok to insn constraint.
9614 (*iordi_1_bts): Ditto.
9615 (*xordi_1_btc): Ditto.
9616 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
9617 Update corresponding peephole2 pattern.
9621 2017-08-16 Bin Cheng <bin.cheng@arm.com>
9623 PR tree-optimization/81832
9624 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
9625 copy loop header which has IFN_LOOP_DIST_ALIAS call.
9627 2017-08-16 Marek Polacek <polacek@redhat.com>
9630 * fold-const.c (fold_indirect_ref_1): Restore original behavior
9631 regarding size_zero_node.
9633 2017-08-16 Martin Liska <mliska@suse.cz>
9636 * config.gcc: Respect previously set extra_objs in case
9639 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
9641 PR tree-optimization/81835
9642 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
9643 the phi SSA_NAME. Check that the condition in a COND_EXPR does
9644 not depend on the phi.
9646 2017-08-16 Alan Modra <amodra@gmail.com>
9648 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
9651 2017-08-16 Alan Modra <amodra@gmail.com>
9653 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
9654 (save_reg_p): ..into this. Update all callers.
9655 (first_reg_to_save): Simplify.
9657 2017-08-16 Alan Modra <amodra@gmail.com>
9659 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
9662 2017-08-15 Joseph Myers <joseph@codesourcery.com>
9666 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
9667 constant count if that count is less than 32.
9669 2017-08-15 Nathan Sidwell <nathan@acm.org>
9671 * gcc.c (execute): Emit friendlier message if inferior is killed
9672 by an external cause.
9674 2017-08-15 Richard Biener <rguenther@suse.de>
9676 PR tree-optimization/81790
9677 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
9678 CONSTRUCTORs from simplifying and VN.
9680 2017-08-14 Martin Sebor <msebor@redhat.com>
9682 * builtin-attrs.def: Add comments.
9684 2017-08-14 Martin Sebor <msebor@redhat.com>
9687 * doc/extend.texi (attribute nonstring): Document new attribute.
9689 2017-08-14 Martin Sebor <msebor@redhat.com>
9692 * tree-diagnostic.c (default_tree_printer): Handle %G.
9693 * gimple-pretty-print.h (percent_G_format): Declare new function.
9694 * gimple-pretty-print.c (percent_G_format): Define.
9695 * tree-pretty-print.c (percent_K_format): Add argument.
9697 2017-08-14 Martin Sebor <msebor@redhat.com>
9699 PR translation/79998
9700 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
9701 Remove a stray space.
9703 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
9706 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
9707 (*iordi_1_bts): Ditto.
9708 (*xordi_1_btc): Ditto.
9710 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9713 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
9714 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9716 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
9717 quoted strings, and make more translator-friendly.
9718 (darwin_rs6000_override_options): Likewise.
9719 (rs6000_option_override_internal): Likewise.
9720 (rs6000_return_in_memory): Fix overlong line.
9721 (init_cmulative_args): Use quoted strings, and make more
9722 translator-friendly.
9723 (rs6000_pass_by_reference): Fix overlong line.
9724 (def_builtin): Use quoted strings.
9725 (altivec_expand_predicate_builtin): Use quoted strings, and make
9726 more translator-friendly.
9727 (htm_expand_builtin): Use quoted strings.
9728 (cpu_expand_builtin): Use quoted strings, and make more
9729 translator-friendly.
9730 (altivec_expand_builtin): Likewise.
9731 (paired_expand_predicate_builtin): Likewise.
9732 (rs6000_invalid_builtin): Likewise.
9733 (builtin_function_type): Use quoted strings.
9734 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
9735 more translator-friendly.
9736 (rs6000_trampoline_init): Likewise.
9737 (rs6000_handle_altivec_attribute): Likewise.
9738 (rs6000_inner_target_options): Use quoted strings.
9739 (rs6000_disable_incompatible_switches): Likewise.
9740 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
9741 strings, and make more translator-friendly.
9742 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
9744 2017-08-14 Bin Cheng <bin.cheng@arm.com>
9746 PR tree-optimization/81799
9747 * tree-loop-distribution.c (version_loop_by_alias_check): Force
9748 cond_expr to simple gimple operand.
9750 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
9753 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
9755 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
9759 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
9760 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
9761 TARGET_GASISR_PROLOGUES.
9762 * config/avr/avr.c (avr_option_override): Same.
9763 (avr_pass_pre_proep::execute): Same.
9765 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
9768 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
9769 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
9771 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
9773 * config/i386/i386.md (*load_tp_<mode>): Redefine as
9774 define_insn_and_split. Split to a memory load from 0 in
9775 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
9776 using PTR mode iterator.
9777 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
9778 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
9779 (*add_tp_<mode>): Redefine as define_insn_and_split.
9780 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
9781 address space. Merge with *add_tp_x32 using PTR mode iterator.
9782 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
9783 Split to an add with a memory load from 0 in
9784 DEFAULT_TLS_SEG_REG address space.
9786 2017-08-12 Andrew Pinski <apinski@cavium.com>
9788 * config/aarch64/aarch64-option-extensions.def (rdma):
9789 Fix feature string to what Linux prints out in /proc/cpuinfo.
9791 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
9794 * dwarf2out.c (modified_type_die): For C typedef types that have
9795 an ultimate origin, process the ultimate origin instead of the
9797 (gen_typedef_die): Assert that input DECLs have no ultimate
9799 (gen_type_die_with_usage): For typedef variants that have an
9800 ultimate origin, just call gen_decl_die on the original DECL.
9801 (process_scope_var): Avoid creating DIEs for local typedefs and
9802 concrete static variables.
9804 2017-08-12 Alan Modra <amodra@gmail.com>
9808 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
9809 match gnu-user.h startfile.
9810 (ENDFILE_LINUX_SPEC): Similarly.
9812 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
9815 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
9817 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
9819 2017-08-11 Tamar Christina <tamar.christina@arm.com>
9820 * config/aarch64/aarch64.md (mov<mode>): Change.
9821 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9822 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
9823 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
9825 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
9827 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
9828 for storage order barriers.
9830 2017-08-11 Martin Liska <mliska@suse.cz>
9833 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
9834 variables of void type.
9836 2017-08-11 Martin Liska <mliska@suse.cz>
9838 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
9839 TARGET_SUPPORTS_ALIASES.
9840 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
9841 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
9842 (optimize_weakref): Likewise.
9843 * symtab.c (symtab_node::noninterposable_alias): Likewise.
9844 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
9845 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
9847 2017-08-11 Martin Liska <mliska@suse.cz>
9850 * config/i386/i386.c (make_resolver_func): Do complete
9851 refactoring of the function.
9853 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9856 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
9857 * config/i386/i386.c (ix86_stack_protect_guard): Use
9858 ix86_stack_protect_guard_symbol_str to generate varible declaration.
9859 * doc/invoke.texi (x86 Options): Document
9860 -mstack-protector-guard-symbol= option.
9862 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9864 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
9865 * config/i386/i386.c (ix86_split_stack_guard): New function.
9866 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
9867 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
9868 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
9869 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
9870 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
9871 (split_stack_space_check): Call ix86_split_stack_guard.
9873 2017-08-10 Martin Sebor <msebor@redhat.com>
9875 * print-tree.c (print_node): Print location using the established
9877 Replace spaces with colons.
9878 (debug_raw, debug): Ditto.
9880 2017-08-10 Martin Sebor <msebor@redhat.com>
9883 * pretty-print.c (pp_format): Correct the handling of %s precision.
9885 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
9888 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9890 (ix86_finalize_stack_frame_flags): This. Also clear
9891 frame_pointer_needed if -fno-omit-frame-pointer is used without
9893 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9894 with ix86_finalize_stack_frame_flags.
9895 (ix86_expand_epilogue): Likewise.
9896 (ix86_expand_split_stack_prologue): Likewise.
9897 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
9899 2017-08-10 Martin Liska <mliska@suse.cz>
9902 * c-attribs.c (handle_target_attribute):
9903 Report warning for an empty string argument of target attribute.
9905 2017-08-09 Jakub Jelinek <jakub@redhat.com>
9908 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
9910 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
9911 or DECL_NONLOCAL labels.
9912 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
9913 or DECL_NONLOCAL labels here.
9915 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
9917 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
9918 to indicate when early gimple folding has been disabled.
9919 (rs6000_gimple_fold_builtin): Add debug content.
9920 (rs6000_invalid_builtin): Fix whitespace.
9921 (rs6000_expand_builtin): Fix whitespace.
9922 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
9924 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
9927 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
9928 SAVE_MULTIPLE if not all the registers that saves, should be saved.
9930 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
9932 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
9933 (qdf24xx): Likewise.
9934 * config/aarch64/aarch64.md: Include falkor.md.
9935 * config/aarch64/falkor.md: New.
9937 2017-08-09 Marek Polacek <polacek@redhat.com>
9940 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
9941 * diagnostic.c (emit_diagnostic): Add a comment.
9942 (emit_diagnostic_valist): New function.
9944 2017-08-09 Marek Polacek <polacek@redhat.com>
9947 * input.c (make_location): New overload.
9948 * input.h (make_location): Declare.
9950 2017-08-08 Alan Modra <amodra@gmail.com>
9951 H.J. Lu <hongjiu.lu@intel.com>
9954 * gcc.c (NO_PIE_SPEC): Delete.
9955 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
9957 (LINK_PIE_SPEC): ..to here.
9958 (LINK_COMMAND_SPEC): Support -no-pie.
9959 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
9960 chain of crtbegin*.o selection, update for PIE_SPEC changes and
9962 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
9963 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
9964 (ENDFILE_CRTEND_SPEC): Similarly.
9966 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
9969 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
9970 (mstack-protector-guard-offset=): Ditto.
9971 * config/i386/i386.c (ix86_option_override): Handle
9972 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
9974 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
9975 ix86_stack_protect_guard_offset variables.
9976 (TARGET_STACK_PROTECT_GUARD): Always define.
9977 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
9978 and -mstack-protector-guard-offset= options.
9980 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9982 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
9983 boundary case for the last candidate.
9985 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9987 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
9988 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
9990 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9993 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
9994 * config/aarch64/aarch64-builtins.c
9995 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
9996 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
9997 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
9999 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10000 Andrew Pinski <pinskia@gmail.com>
10002 PR middle-end/19706
10003 * internal-fn.def (XORSIGN): New.
10004 * optabs.def (xorsign_optab): New.
10005 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
10006 (convert_expand_mult_copysign): New.
10007 (pass_optimize_widening_mul::execute): Call
10008 convert_expand_mult_copysign.
10010 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10012 PR tree-optimization/81354
10013 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
10014 Insert on edges rather than explicitly creating landing pads.
10015 (analyze_candidates_and_replace): Commit edge inserts.
10017 2017-08-08 Richard Biener <rguenther@suse.de>
10019 PR middle-end/81719
10020 * tree-ssa-loop-niter.c: Include tree-dfa.h.
10021 (expand_simple_operations): Also look through ADDR_EXPRs with
10022 MEM_REF bases treating them as POINTER_PLUS_EXPR.
10024 2017-08-08 Richard Biener <rguenther@suse.de>
10026 PR tree-optimization/81723
10027 * tree-vect-slp.c (struct bst_traits): New hash traits.
10028 (bst_fail): New global.
10029 (vect_build_slp_tree_2): New worker, split out from ...
10030 (vect_build_slp_tree): ... this now wrapping it with using
10031 bst_fail set to cache SLP tree build fails. Properly handle
10033 (vect_analyze_slp_instance): Allocate and free bst_fail.
10035 2017-08-08 Martin Liska <mliska@suse.cz>
10038 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
10039 LABEL_DECLs that can be from a different function.
10041 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10043 PR tree-optimization/81744
10044 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
10045 loop's number of iterations.
10047 2017-08-08 Martin Liska <mliska@suse.cz>
10049 * asan.c: Include header files.
10050 * attribs.c (build_decl_attribute_variant): New function moved
10052 (build_type_attribute_qual_variant): Likewise.
10053 (cmp_attrib_identifiers): Likewise.
10054 (simple_cst_list_equal): Likewise.
10055 (omp_declare_simd_clauses_equal): Likewise.
10056 (attribute_value_equal): Likewise.
10057 (comp_type_attributes): Likewise.
10058 (build_type_attribute_variant): Likewise.
10059 (lookup_ident_attribute): Likewise.
10060 (remove_attribute): Likewise.
10061 (merge_attributes): Likewise.
10062 (merge_type_attributes): Likewise.
10063 (merge_decl_attributes): Likewise.
10064 (merge_dllimport_decl_attributes): Likewise.
10065 (handle_dll_attribute): Likewise.
10066 (attribute_list_equal): Likewise.
10067 (attribute_list_contained): Likewise.
10068 * attribs.h (lookup_attribute): New function moved from tree.[ch].
10069 (lookup_attribute_by_prefix): Likewise.
10070 * bb-reorder.c: Include header files.
10071 * builtins.c: Likewise.
10072 * calls.c: Likewise.
10073 * cfgexpand.c: Likewise.
10074 * cgraph.c: Likewise.
10075 * cgraphunit.c: Likewise.
10076 * convert.c: Likewise.
10077 * dwarf2out.c: Likewise.
10078 * final.c: Likewise.
10079 * fold-const.c: Likewise.
10080 * function.c: Likewise.
10081 * gimple-expr.c: Likewise.
10082 * gimple-fold.c: Likewise.
10083 * gimple-pretty-print.c: Likewise.
10084 * gimple.c: Likewise.
10085 * gimplify.c: Likewise.
10086 * hsa-common.c: Likewise.
10087 * hsa-gen.c: Likewise.
10088 * internal-fn.c: Likewise.
10089 * ipa-chkp.c: Likewise.
10090 * ipa-cp.c: Likewise.
10091 * ipa-devirt.c: Likewise.
10092 * ipa-fnsummary.c: Likewise.
10093 * ipa-inline.c: Likewise.
10094 * ipa-visibility.c: Likewise.
10096 * lto-cgraph.c: Likewise.
10097 * omp-expand.c: Likewise.
10098 * omp-general.c: Likewise.
10099 * omp-low.c: Likewise.
10100 * omp-offload.c: Likewise.
10101 * omp-simd-clone.c: Likewise.
10102 * opts-global.c: Likewise.
10103 * passes.c: Likewise.
10104 * predict.c: Likewise.
10105 * sancov.c: Likewise.
10106 * sanopt.c: Likewise.
10107 * symtab.c: Likewise.
10108 * toplev.c: Likewise.
10109 * trans-mem.c: Likewise.
10110 * tree-chkp.c: Likewise.
10111 * tree-eh.c: Likewise.
10112 * tree-into-ssa.c: Likewise.
10113 * tree-object-size.c: Likewise.
10114 * tree-parloops.c: Likewise.
10115 * tree-profile.c: Likewise.
10116 * tree-ssa-ccp.c: Likewise.
10117 * tree-ssa-live.c: Likewise.
10118 * tree-ssa-loop.c: Likewise.
10119 * tree-ssa-sccvn.c: Likewise.
10120 * tree-ssa-structalias.c: Likewise.
10121 * tree-ssa.c: Likewise.
10122 * tree-streamer-in.c: Likewise.
10123 * tree-vectorizer.c: Likewise.
10124 * tree-vrp.c: Likewise.
10125 * tsan.c: Likewise.
10126 * ubsan.c: Likewise.
10127 * varasm.c: Likewise.
10128 * varpool.c: Likewise.
10129 * tree.c: Remove functions moved to attribs.[ch].
10130 * tree.h: Likewise.
10131 * config/aarch64/aarch64.c: Add attrs.h header file.
10132 * config/alpha/alpha.c: Likewise.
10133 * config/arc/arc.c: Likewise.
10134 * config/arm/arm.c: Likewise.
10135 * config/avr/avr.c: Likewise.
10136 * config/bfin/bfin.c: Likewise.
10137 * config/c6x/c6x.c: Likewise.
10138 * config/cr16/cr16.c: Likewise.
10139 * config/cris/cris.c: Likewise.
10140 * config/darwin.c: Likewise.
10141 * config/epiphany/epiphany.c: Likewise.
10142 * config/fr30/fr30.c: Likewise.
10143 * config/frv/frv.c: Likewise.
10144 * config/ft32/ft32.c: Likewise.
10145 * config/h8300/h8300.c: Likewise.
10146 * config/i386/winnt.c: Likewise.
10147 * config/ia64/ia64.c: Likewise.
10148 * config/iq2000/iq2000.c: Likewise.
10149 * config/lm32/lm32.c: Likewise.
10150 * config/m32c/m32c.c: Likewise.
10151 * config/m32r/m32r.c: Likewise.
10152 * config/m68k/m68k.c: Likewise.
10153 * config/mcore/mcore.c: Likewise.
10154 * config/microblaze/microblaze.c: Likewise.
10155 * config/mips/mips.c: Likewise.
10156 * config/mmix/mmix.c: Likewise.
10157 * config/mn10300/mn10300.c: Likewise.
10158 * config/moxie/moxie.c: Likewise.
10159 * config/msp430/msp430.c: Likewise.
10160 * config/nds32/nds32-isr.c: Likewise.
10161 * config/nds32/nds32.c: Likewise.
10162 * config/nios2/nios2.c: Likewise.
10163 * config/nvptx/nvptx.c: Likewise.
10164 * config/pa/pa.c: Likewise.
10165 * config/pdp11/pdp11.c: Likewise.
10166 * config/powerpcspe/powerpcspe.c: Likewise.
10167 * config/riscv/riscv.c: Likewise.
10168 * config/rl78/rl78.c: Likewise.
10169 * config/rx/rx.c: Likewise.
10170 * config/s390/s390.c: Likewise.
10171 * config/sh/sh.c: Likewise.
10172 * config/sol2.c: Likewise.
10173 * config/sparc/sparc.c: Likewise.
10174 * config/spu/spu.c: Likewise.
10175 * config/stormy16/stormy16.c: Likewise.
10176 * config/tilegx/tilegx.c: Likewise.
10177 * config/tilepro/tilepro.c: Likewise.
10178 * config/v850/v850.c: Likewise.
10179 * config/vax/vax.c: Likewise.
10180 * config/visium/visium.c: Likewise.
10181 * config/xtensa/xtensa.c: Likewise.
10183 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10186 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
10187 constraints since the -mupper-regs-* switches have been
10189 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
10190 into a vector from a double word element that was extracted from
10191 another vector, and eliminate extra XXPERMDI instructions.
10192 (vsx_concat_<mode>_2): Likewise.
10193 (vsx_concat_<mode>_3): Likewise.
10194 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
10195 concat to allow optimizing inserts from previous extracts.
10197 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
10199 * config/i386/i386.c (ix86_stack_protect_guard): Generate
10200 memory reference to a SSP offset in TLS address space.
10201 (ix86_print_operand) <case '@'>: Remove.
10202 (ix86_print_operand_punct_valid_p): Remove '@' code.
10203 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
10204 UNSPEC_SP_TLS_TEST.
10205 (stack_tls_protect_set_<mode>): Remove.
10206 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
10207 (stack_tls_protect_test_<mode>): Remove.
10208 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
10210 2017-08-07 Olivier Hainque <hainque@adacore.com>
10213 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
10215 2017-08-07 Douglas Rupp <rupp@adacore.com>
10217 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
10218 variable was referenced as multidir in command.
10220 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10223 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
10226 2017-08-07 Martin Liska <mliska@suse.cz>
10228 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
10229 * config/rl78/rl78.c: Add include of attribs.h.
10230 * config/sh/sh.c: Likewise.
10231 * config/v850/v850.c: Likewise.
10233 2017-08-07 Tom de Vries <tom@codesourcery.com>
10235 PR middle-end/78266
10236 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
10238 2017-08-07 Martin Liska <mliska@suse.cz>
10240 * config/mips/mips.c: Include attribs.h.
10242 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
10245 * doc/invoke.texi: Document change in behvaior for -Ofast for
10248 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
10250 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
10252 (aarch64_pop_regs): Likewise.
10253 (aarch64_gen_load_pair): Likewise.
10254 (aarch64_save_callee_saves): Likewise.
10255 (aarch64_restore_callee_saves): Likewise.
10257 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10259 * config/i386/i386.c: Revert the last change.
10261 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10264 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10266 (ix86_finalize_stack_frame_flags): This. Also clear
10267 frame_pointer_needed if -fno-omit-frame-pointer is used without
10269 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10270 with ix86_finalize_stack_frame_flags.
10271 (ix86_expand_epilogue): Likewise.
10272 (ix86_expand_split_stack_prologue): Likewise.
10274 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10277 * config/i386/i386.c (get_builtin_code_for_version): Set priority
10278 to P_AES for Westmere.
10280 2017-08-07 Jonathan Yong <10walls@gmail.com>
10282 * config/i386/mingw.opt (fset-stack-executable): Removed.
10283 * config/i386/cygming.opt (fset-stack-executable): Moved
10285 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
10287 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
10289 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
10291 2017-08-07 Marek Polacek <polacek@redhat.com>
10293 PR middle-end/81737
10294 * fold-const.c (fold_indirect_ref_1): Check type_domain.
10296 2017-08-07 Martin Liska <mliska@suse.cz>
10298 * attribs.h (canonicalize_attr_name): New function.
10299 (cmp_attribs): Move from c-format.c and adjusted.
10300 (is_attribute_p): Moved from tree.h.
10301 * tree-inline.c: Add new includes.
10302 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
10303 (private_is_attribute_p): Remove.
10304 (private_lookup_attribute): Likewise.
10305 (private_lookup_attribute_by_prefix): Simplify.
10306 (remove_attribute): Use is_attribute_p.
10307 * tree.h: Remove removed declarations.
10309 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10311 PR middle-end/81698
10312 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
10313 instead of computing it in the function. Formatting fix.
10314 (expand_case): Don't rely on default_edge being the first edge,
10315 clear it if removing it, pass default_edge to
10316 emit_case_dispatch_table.
10317 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
10320 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
10322 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
10323 insn in the function, emit NOP after the insn.
10325 2017-08-06 Tom de Vries <tom@codesourcery.com>
10327 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
10330 2017-08-06 Tom de Vries <tom@codesourcery.com>
10332 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
10335 2017-08-04 Yury Gribov <tetra2005@gmail.com>
10337 PR tree-optimization/57371
10338 * match.pd: New pattern.
10340 2017-08-04 Marek Polacek <polacek@redhat.com>
10342 PR middle-end/81695
10343 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
10344 perform the computation in offset_int.
10346 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10348 PR tree-optimization/81136
10349 * tree-vectorizer.h: Include tree-hash-traits.h.
10350 (vec_base_alignments): New typedef.
10351 (vec_info): Add a base_alignments field.
10352 (vect_record_base_alignments): Declare.
10353 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
10355 (DR_IS_CONDITIONAL_IN_STMT): New macro.
10356 (create_data_ref): Add an is_conditional_in_stmt argument.
10357 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
10358 the is_conditional_in_stmt field.
10359 (data_ref_loc): Add an is_conditional_in_stmt field.
10360 (get_references_in_stmt): Set the is_conditional_in_stmt field.
10361 (find_data_references_in_stmt): Update call to create_data_ref.
10362 (graphite_find_data_references_in_stmt): Likewise.
10363 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
10364 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
10365 (vect_record_base_alignment): New function.
10366 (vect_record_base_alignments): Likewise.
10367 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
10368 for nested statements even if we fail to compute a misalignment.
10369 Use pooled base alignments for unconditional references.
10370 (vect_find_same_alignment_drs): Compare base addresses instead
10372 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
10373 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
10375 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10377 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
10378 Add an explicit name for the enum. Use auto_vec for slp_instances
10379 and grouped_stores.
10380 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
10382 (_bb_vec_info): Add a constructor and destructor.
10383 (vinfo_for_stmt): Return NULL for uids of -1 as well.
10384 (destroy_loop_vec_info): Delete.
10385 (vect_destroy_datarefs): Likewise.
10386 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
10387 (vec_info::vec_info): New function.
10388 (vec_info::~vec_info): Likewise.
10389 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
10390 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
10391 destroy_loop_vec_info.
10392 * tree-vect-loop.c (new_loop_vec_info): Replace with...
10393 (_loop_vec_info::_loop_vec_info): ...this.
10394 (destroy_loop_vec_info): Replace with...
10395 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
10396 the stmt_vec_infos. Leave handling of vec_info information to its
10397 destructor. Remove explicit vector releases.
10398 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
10399 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
10400 * tree-vect-slp.c (new_bb_vec_info): Replace with...
10401 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
10402 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
10403 (destroy_bb_vec_info): Replace with...
10404 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
10405 information to its destructor.
10406 (vect_slp_analyze_bb_1): Use new and delete instead of
10407 new_bb_vec_info and destroy_bb_vec_info.
10408 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
10411 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10413 * tree-data-ref.h (subscript): Add access_fn field.
10414 (data_dependence_relation): Add could_be_independent_p.
10415 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
10416 (same_access_functions): Move to tree-data-ref.c.
10417 * tree-data-ref.c (ref_contains_union_access_p): New function.
10418 (access_fn_component_p): Likewise.
10419 (access_fn_components_comparable_p): Likewise.
10420 (dr_analyze_indices): Add a reference to access_fn_component_p.
10421 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
10423 (constant_access_functions): Likewise.
10424 (add_other_self_distances): Likewise.
10425 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
10426 (initialize_data_dependence_relation): Use XCNEW and remove
10427 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
10428 of access functions that have the same type. Allow the
10429 subsequence to end with different bases in some circumstances.
10430 Record the chosen access functions in SUB_ACCESS_FN.
10431 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
10432 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
10433 (subscript_dependence_tester_1): Likewise dra and drb.
10434 (build_classic_dist_vector): Update calls accordingly.
10435 (subscript_dependence_tester): Likewise.
10436 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
10437 DDR_COULD_BE_INDEPENDENT_P.
10438 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
10439 comp_alias_ddrs instead of may_alias_ddrs.
10440 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
10442 (vect_analyze_data_ref_dependence): Use it if
10443 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
10444 distance vectors if that fails.
10445 (dependence_distance_ge_vf): New function.
10446 (vect_prune_runtime_alias_test_list): Use it. Don't clear
10447 LOOP_VINFO_MAY_ALIAS_DDRS.
10449 2017-08-04 Richard Biener <rguenther@suse.de>
10451 PR middle-end/81705
10452 * fold-const.c (fold_binary_loc): Properly restrict
10453 minus_var0 && minus_var1 case when associating undefined overflow
10456 2017-08-04 Tom de Vries <tom@codesourcery.com>
10458 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
10460 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10462 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10463 Don't start diagnostic messages with a capital letter.
10464 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10466 (rs6000_invalid_builtin): Likewise.
10467 (rs6000_trampoline_init): Likewise.
10469 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10472 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
10473 after setting changeable df flags.
10475 2017-08-03 Richard Biener <rguenther@suse.de>
10477 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
10478 up if the use is in USE - X.
10480 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
10482 * toplev.c (dumpfile.h): New include.
10483 (internal_error_reentered): New static function. Use it...
10484 (internal_error_function): ...here to handle reentered internal_error.
10486 2017-08-03 Richard Biener <rguenther@suse.de>
10488 PR middle-end/81148
10489 * fold-const.c (split_tree): Add minus_var and minus_con
10490 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
10491 here but always use minus_*.
10492 (associate_trees): Assert we never associate with MINUS_EXPR
10493 and NULL first operand. Do not recurse for PLUS_EXPR operands
10494 when associating as MINUS_EXPR either.
10495 (fold_binary_loc): Track minus_var and minus_con.
10497 2017-08-03 Tom de Vries <tom@codesourcery.com>
10500 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
10501 ACCEL_COMPILER, apply finish_options on
10502 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
10504 2017-08-03 Tom de Vries <tom@codesourcery.com>
10507 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
10508 function_entry_patch_area_size > 0.
10510 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10513 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
10514 instead of 10??LU, perform unit multiplication in wide_int,
10515 don't change alloc_object_size_limit if the limit is larger
10518 PR tree-optimization/81655
10519 PR tree-optimization/81588
10520 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
10521 the case when ranges[i].low and high are 1 for unsigned type with
10524 PR middle-end/81052
10525 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
10526 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
10528 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10530 * tree-vrp.h: Add include guard.
10532 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
10535 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
10536 (ud2): New insn pattern.
10537 * config/i386/i386.c (ix86_expand_epilogue):
10538 For naked functions, generate ud2 instead of trap insn.
10540 2017-08-02 Marek Polacek <polacek@redhat.com>
10543 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
10545 2017-08-02 Tom de Vries <tom@codesourcery.com>
10546 Cesar Philippidis <cesar@codesourcery.com>
10548 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
10549 Add missing edge probabilities.
10551 2017-08-02 Tamar Christina <tamar.christina@arm.com>
10553 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
10554 Correct endianness.
10556 2017-08-02 Jakub Jelinek <jakub@redhat.com>
10558 PR middle-end/79499
10559 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
10560 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
10561 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
10562 prologue_seq sequences - if any.
10564 2017-08-02 Richard Biener <rguenther@suse.de>
10566 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
10567 via vectors if supported, integer extracts via punning if supported
10568 or otherwise vector extracts.
10570 2017-08-02 Richard Biener <rguenther@suse.de>
10572 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
10574 (bitmap_insert_into_set): ... this.
10576 2017-08-02 Richard Biener <rguenther@suse.de>
10578 PR tree-optimization/81633
10580 2015-08-17 Alan Hayward <alan.hayward@arm.com>
10582 PR tree-optimization/71752
10583 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
10585 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
10587 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
10588 (machine_function::call_ms2sysv_pad_out): Remove field.
10589 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
10590 (ix86_compute_frame_layout): Likewise.
10592 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
10595 * config/i386/i386.c (ix86_set_func_type): Disallow naked
10596 attribute with interrupt attribute.
10598 2017-08-01 Andrew Pinski <apinski@cavium.com>
10600 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
10601 BIT_INSERT_EXPR's operand 1
10602 to see if the types precision matches.
10604 2017-08-01 Martin Liska <mliska@suse.cz>
10606 PR middle-end/70140
10607 * builtins.c (expand_builtin_memcpy_args): Remove.
10608 (expand_builtin_memcpy): Call newly added function
10609 expand_builtin_memory_copy_args.
10610 (expand_builtin_memcpy_with_bounds): Likewise.
10611 (expand_builtin_mempcpy): Remove last argument.
10612 (expand_builtin_mempcpy_with_bounds): Likewise.
10613 (expand_builtin_memory_copy_args): New function created from
10614 expand_builtin_mempcpy_args with small modifications.
10615 (expand_builtin_mempcpy_args): Remove.
10616 (expand_builtin_stpcpy): Remove unused argument.
10617 (expand_builtin): Likewise.
10618 (expand_builtin_with_bounds): Likewise.
10620 2017-08-01 Martin Liska <mliska@suse.cz>
10623 Make mempcpy more optimal (PR middle-end/70140).
10625 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10628 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10629 __builtin_vec_cmpne verify both arguments are compatible vectors
10630 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
10631 verify arg1_type is a pointer or array type. For __builtin_vec_st,
10632 move computation of aligned to after checking the argument types.
10636 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10637 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
10640 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
10641 Doug Rupp <rupp@adacore.com>
10642 Olivier Hainque <hainque@adacore.com>
10644 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
10645 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
10647 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
10648 for TARGET_OS_CPP_BUILTIN.
10649 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
10650 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
10652 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
10653 passing required abi options to the assembler for EABI configurations.
10654 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
10655 of .text.hot and .text.unlikely sections for kernel modules when
10656 using ARM style exceptions.
10657 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
10658 options. Add EXTRA_CC1_SPEC.
10659 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
10661 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
10663 (ARM_TARGET2_DWARF_FORMAT): Define.
10664 * config/arm/t-vxworks: Adjust multilib control to removal of the
10665 Diab command line options.
10667 2017-08-01 Martin Liska <mliska@suse.cz>
10669 PR gcov-profile/81561
10670 * gcov.c (unblock): Make unblocking safe as we need to preserve
10671 index correspondence of blocks and block_lists.
10673 2017-08-01 Richard Biener <rguenther@suse.de>
10675 PR tree-optimization/81181
10676 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10677 (compute_antic): ... end of iteration here.
10679 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
10681 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
10682 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
10683 (ftree-slp-vectorize): Likewise.
10684 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
10685 can no longer be set independent of flag_tree_loop_vectorize.
10686 * omp-general.c (emp_max_vf): Likewise.
10687 * opts.c (enable_fdo_optimizations): Remove references to
10688 flag_tree_vectorize, these are now implicit.
10689 (common_handle_option): Remove handling for OPT_ftree_vectorize,
10690 and leave it for the options machinery.
10692 2017-08-01 Martin Liska <mliska@suse.cz>
10694 PR middle-end/70140
10695 * builtins.c (expand_builtin_memcpy_args): Remove.
10696 (expand_builtin_memcpy): Call newly added function
10697 expand_builtin_memory_copy_args.
10698 (expand_builtin_memcpy_with_bounds): Likewise.
10699 (expand_builtin_mempcpy): Remove last argument.
10700 (expand_builtin_mempcpy_with_bounds): Likewise.
10701 (expand_builtin_memory_copy_args): New function created from
10702 expand_builtin_mempcpy_args with small modifications.
10703 (expand_builtin_mempcpy_args): Remove.
10704 (expand_builtin_stpcpy): Remove unused argument.
10705 (expand_builtin): Likewise.
10706 (expand_builtin_with_bounds): Likewise.
10708 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10711 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
10712 print "ds:" only for immediates in generic address space.
10714 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10717 * config/i386/i386.c (ix86_funciton_naked): New prototype.
10718 (ix86_function_ok_for_sibcall): Return false for naked functions.
10720 2017-08-01 Richard Biener <rguenther@suse.de>
10722 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
10723 (compute_antic): Seed worklist with exit block predecessors.
10724 * cfganal.c (dfs_find_deadend): For a cycle return the source
10725 of the edge closing it.
10727 2017-08-01 Tamar Christina <tamar.christina@arm.com>
10729 * config/aarch64/aarch64.c
10730 (aarch64_can_const_movi_rtx_p): Move 0 check.
10732 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10734 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
10735 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
10737 * match.pd: Ditto in address comparison pattern.
10739 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10741 PR tree-optimization/81627
10742 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
10743 closed ssa form for store-store chain.
10745 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10747 PR tree-optimization/81620
10748 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
10749 for store-store chain.
10751 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10753 PR tree-optimization/81588
10754 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
10755 ranges[i].in_p, invert comparison code ccode. For >/>=,
10756 swap rhs1 and rhs2 and comparison code unconditionally,
10757 for </<= don't do that. Don't swap rhs1/rhs2 again if
10758 ranges[i].in_p, instead invert comparison code ccode if
10759 opcode or oe->rank is BIT_IOR_EXPR.
10762 * optabs.def (vec_extract_optab, vec_init_optab): Change from
10763 a direct optab to conversion optab.
10764 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
10765 with GET_MODE_INNER as last argument instead of optab_handler.
10766 * expmed.c (extract_bit_field_1): Likewise. Use vector from
10767 vector extraction if possible and optab is available.
10768 * expr.c (store_constructor): Use convert_optab_handler instead
10769 of optab_handler. Use vector initialization from smaller
10770 vectors if possible and optab is available.
10771 * tree-vect-stmts.c (vectorizable_load): Likewise.
10772 * doc/md.texi (vec_extract, vec_init): Document that the optabs
10773 now have two modes.
10774 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
10775 of vec_init from half-sized vectors with the same element mode.
10776 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
10777 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
10778 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
10779 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
10780 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
10781 after mode in gen_vec_extract* calls.
10782 (vec_extract<mode>): Renamed to ...
10783 (vec_extract<mode><ssescalarmodelower>): ... this.
10784 (vec_extract<mode><ssehalfvecmodelower>): New expander.
10785 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
10786 element mode after mode in gen_vec_init* calls.
10787 (VEC_INIT_HALF_MODE): New mode iterator.
10788 (vec_init<mode>): Renamed to ...
10789 (vec_init<mode><ssescalarmodelower>): ... this.
10790 (vec_init<mode><ssehalfvecmodelower>): New expander.
10791 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
10792 (vec_extractv2sfsf): ... this.
10793 (vec_initv2sf): Renamed to ...
10794 (vec_initv2sfsf): ... this.
10795 (vec_extractv2si): Renamed to ...
10796 (vec_extractv2sisi): ... this.
10797 (vec_initv2si): Renamed to ...
10798 (vec_initv2sisi): ... this.
10799 (vec_extractv4hi): Renamed to ...
10800 (vec_extractv4hihi): ... this.
10801 (vec_initv4hi): Renamed to ...
10802 (vec_initv4hihi): ... this.
10803 (vec_extractv8qi): Renamed to ...
10804 (vec_extractv8qiqi): ... this.
10805 (vec_initv8qi): Renamed to ...
10806 (vec_initv8qiqi): ... this.
10807 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
10808 (vec_init<mode>): Renamed to ...
10809 (vec_init<mode><VEC_base_l>): ... this.
10810 (vec_extract<mode>): Renamed to ...
10811 (vec_extract<mode><VEC_base_l>): ... this.
10812 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
10813 (vec_initv2sfsf): ... this.
10814 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
10815 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10816 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
10817 element mode after mode in gen_vec_init* calls.
10818 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
10819 (vec_init<mode><Vel>): ... this.
10820 (vec_extract<mode>): Renamed to ...
10821 (vec_extract<mode><Vel>): ... this.
10822 * config/aarch64/iterators.md (Vel): New mode attribute.
10823 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
10824 Add element mode after mode in gen_vec_extract* calls.
10825 * config/s390/vector.md (non_vec_l): New mode attribute.
10826 (vec_extract<mode>): Renamed to ...
10827 (vec_extract<mode><non_vec_l>): ... this.
10828 (vec_init<mode>): Renamed to ...
10829 (vec_init<mode><non_vec_l>): ... this.
10830 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
10831 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
10833 * config/arm/iterators.md (V_elem_l): New mode attribute.
10834 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
10835 (vec_extract<mode><V_elem_l>): ... this.
10836 (vec_extractv2di): Renamed to ...
10837 (vec_extractv2didi): ... this.
10838 (vec_init<mode>): Renamed to ...
10839 (vec_init<mode><V_elem_l>): ... this.
10840 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
10841 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
10842 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
10843 Add element mode after gen_vec_extract* calls.
10844 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
10845 (vec_init<mode><unitmode>): ... this.
10846 (vec_extract<mode>): Renamed to ...
10847 (vec_extract<mode><unitmode>): ... this.
10848 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
10849 (vec_init<mode><unitmode>): ... this.
10850 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
10851 (vec_initv2sfsf): ... this.
10852 (vec_extractv2sf): Renamed to ...
10853 (vec_extractv2sfsf): ... this.
10854 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
10855 Add element mode after gen_vec_extract* calls.
10856 * config/mips/mips.md (unitmode): New mode iterator.
10857 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
10858 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
10859 * config/spu/spu.md (inner_l): New mode attribute.
10860 (vec_init<mode>): Renamed to ...
10861 (vec_init<mode><inner_l>): ... this.
10862 (vec_extract<mode>): Renamed to ...
10863 (vec_extract<mode><inner_l>): ... this.
10864 * config/sparc/sparc.md (veltmode): New mode iterator.
10865 (vec_init<VMALL:mode>): Renamed to ...
10866 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
10867 * config/ia64/vect.md (vec_initv2si): Renamed to ...
10868 (vec_initv2sisi): ... this.
10869 (vec_initv2sf): Renamed to ...
10870 (vec_initv2sfsf): ... this.
10871 (vec_extractv2sf): Renamed to ...
10872 (vec_extractv2sfsf): ... this.
10873 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
10874 (vec_init<mode>): Renamed to ...
10875 (vec_init<mode><VEC_base_l>): ... this.
10876 (vec_extract<mode>): Renamed to ...
10877 (vec_extract<mode><VEC_base_l>): ... this.
10878 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
10879 (vec_initv2sfsf): ... this.
10880 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
10881 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10882 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
10883 gen_vec_init* calls.
10885 2017-08-01 Richard Biener <rguenther@suse.de>
10887 PR tree-optimization/81297
10888 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
10889 TREE_OVERFLOW from INTEGER_CSTs.
10891 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
10893 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
10895 2017-07-31 Carl Love <cel@us.ibm.com>
10897 * config/rs6000/rs6000-c: Add support for built-in functions
10898 vector signed char vec_xl_be (signed long long, signed char *);
10899 vector unsigned char vec_xl_be (signed long long, unsigned char *);
10900 vector signed int vec_xl_be (signed long long, signed int *);
10901 vector unsigned int vec_xl_be (signed long long, unsigned int *);
10902 vector signed long long vec_xl_be (signed long long, signed long long *);
10903 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
10904 vector signed short vec_xl_be (signed long long, signed short *);
10905 vector unsigned short vec_xl_be (signed long long, unsigned short *);
10906 vector double vec_xl_be (signed long long, double *);
10907 vector float vec_xl_be (signed long long, float *);
10908 * config/rs6000/altivec.h (vec_xl_be): Add #define.
10909 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
10910 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
10912 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
10913 (altivec_expand_builtin): Add switch statement to call
10914 altivec_expand_xl_be for each builtin.
10915 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
10916 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
10917 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
10918 __builtin_vsx_le_be_v16qi.
10919 * doc/extend.texi: Update the built-in documentation file for the
10920 new built-in functions.
10922 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10925 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
10927 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10929 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10931 * config.gcc: Add z14.
10932 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
10933 CPU model numbers for z13s and z14.
10934 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
10936 * config/s390/s390-opts.h (enum processor_type): Rename
10937 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10938 * config/s390/s390.c (processor_table): Add field for CPU name to
10939 be passed to Binutils.
10940 (s390_asm_output_machine_for_arch): Use the new field in
10941 processor_table for Binutils.
10942 (s390_expand_builtin): Replace arch12 with z14.
10943 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10944 (s390_get_sched_attrmask): Likewise.
10945 (s390_get_unit_mask): Likewise.
10946 * config/s390/s390.opt: Add z14 to processor_type enum.
10948 2017-07-31 Martin Jambor <mjambor@suse.cz>
10951 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
10952 regardless of optimization level.
10954 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
10955 Martin Liska <mliska@suse.cz>
10957 * predict.def: Remove old comment and adjust probability.
10958 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
10959 PREDICT statements.
10961 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10964 * config/i386/i386.c (ix86_function_naked): New function.
10965 (ix86_can_use_return_insn_p): Return false for naked functions.
10966 (ix86_expand_prologue): Skip prologue for naked functions.
10967 (ix86_expand_epilogue): Skip epilogue for naked functions
10968 and emit trap instruction.
10969 (ix86_warn_func_return): New function.
10970 (ix86_attribute_table): Add "naked" attribute specification.
10971 (TARGET_WARN_FUNC_RETURN): Define.
10972 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
10974 2017-07-31 Martin Liska <mliska@suse.cz>
10976 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
10977 (dump_gimple_bb_header): Always dump BB info.
10978 (pp_cfg_jump): Do not append info about BB when dumping a jump.
10980 2017-07-31 Martin Liska <mliska@suse.cz>
10983 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
10984 also with current_function_decl non-null equality.
10986 2017-07-31 Jakub Jelinek <jakub@redhat.com>
10989 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
10990 change type to the element type, instead add eltype variable and
10991 use it where we are interested in the element type.
10993 PR tree-optimization/81603
10994 * ipa-polymorphic-call.c
10995 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
10996 offset arithmetic in offset_int, bail out if the resulting bit offset
10997 doesn't fit into shwi.
10999 2017-07-31 Martin Liska <mliska@suse.cz>
11001 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
11002 (gimplify_save_expr): Fix comment.
11004 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
11007 * config/i386/i386.c (ix86_function_arg): Update arguments for
11009 (ix86_compute_frame_layout): Set the initial stack offset to
11010 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
11011 INCOMING_FRAME_SP_OFFSET.
11012 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
11013 stack before exception handler returns.
11014 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
11015 the 'ERROR_CODE' for exception handler.
11017 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
11019 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
11020 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
11021 (ASM_OUTPUT_REG_POP): Ditto.
11022 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
11023 instead of asm_fprintf to output pure string.
11025 2017-07-29 Jakub Jelinek <jakub@redhat.com>
11027 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
11028 to imported_module_or_decl hook.
11029 (debug_nothing_tree_tree_tree_bool): Remove.
11030 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
11031 * debug.c (do_nothing_debug_hooks): Use
11032 debug_nothing_tree_tree_tree_bool_bool instead of
11033 debug_nothing_tree_tree_tree_bool.
11034 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11035 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
11036 * sdbout.c (sdb_debug_hooks): Likewise.
11037 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
11038 (gen_namespace_die): Add DW_AT_export_symbols attribute if
11040 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
11041 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
11042 attribute, don't add anything.
11044 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11046 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
11047 (fold_build2_stat_loc): Likewise.
11048 (fold_build3_stat_loc): Likewise.
11049 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
11050 (fold_build1_loc): Remove macro.
11051 (fold_build2_loc): Likewise.
11052 (fold_build3_loc): Likewise.
11054 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11056 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
11057 (gimple_build_debug_bind_source_stat): Likewise.
11058 * gimple.h (gimple_build_debug_bind): Remove macro.
11059 (gimple_build_debug_bind_source): Likewise.
11061 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11063 * bitmap.c (bitmap_alloc): Adjust.
11064 (bitmap_gc_alloc): Likewise.
11065 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
11067 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11069 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
11070 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
11071 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
11072 (bitmap_gc_alloc_stat): Likewise.
11073 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
11075 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11077 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
11078 * rtl.h (shallow_copy_rtx): Remove macro.
11080 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11082 * emit-rtl.c (gen_raw_REG): Adjust.
11083 * gengenrtl.c (gendef): Likewise.
11084 * rtl.c (rtx_alloc_stat): Remove _stat from name.
11085 * rtl.h (rtx_alloc): Remove macro.
11087 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11089 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
11090 (build_tree_list_stat): Likewise.
11091 * tree.h (build_tree_list): Remove macro.
11092 (build_tree_list_vec): Likewise.
11094 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11096 * tree.c (make_vector_stat): Remove _stat from name.
11097 (build_vector_stat): Likewise.
11098 * tree.h (make_vector_stat): Remove macro.
11099 (build_vector_stat): Likewise.
11101 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11103 * tree.h (build_var_debug_value): Remove prototype.
11105 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11107 * tree.c (tree_cons_stat): Remove _stat from name.
11108 * tree.h (tree_cons): Remove macro.
11110 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11112 * tree.c (build_vl_exp_stat): Remove _stat from name.
11113 * tree.h (build_vl_exp): Remove macro.
11115 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11117 * tree.c (build_decl_stat): Remove _stat from name.
11118 * tree.h (build_decl): Remove macro.
11120 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11122 * gimple.c (gimple_build_with_ops_stat): Adjust.
11123 (gimple_alloc_stat): Remove _stat from name.
11124 * gimple.h (gimple_alloc): Remove macro.
11126 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11128 * tree.c (make_tree_vec_stat): Remove _stat from name.
11129 (grow_tree_vec_stat): Likewise.
11130 * tree.h (make_tree_vec_stat): Adjust prototype.
11131 (grow_tree_vec_stat): Likewise.
11132 (make_tree_vec): Remove macro.
11133 (grow_tree_vec): Likewise.
11135 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11137 * fold-const.c (fold_build1_stat_loc): Adjust.
11138 (fold_build2_stat_loc): Likewise.
11139 (fold_build3_stat_loc): Likewise.
11140 * tree.c (build0_stat): Remove _stat from name.
11141 (build1_stat): Likewise.
11142 (build2_stat): Likewise.
11143 (build3_stat): Likewise.
11144 (build4_stat): Likewise.
11145 (build5_stat): Likewise.
11146 * tree.h (build1_loc): Remove macro, and rename _stat function
11148 (build2_loc): Likewise.
11149 (build3_loc): Likewise.
11150 (build4_loc): Likewise.
11151 (build5_loc): Likewise.
11153 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11155 * tree.c (make_int_cst_stat): Remove _stat from name.
11156 * tree.h (make_int_cst_stat): Adjust prototype.
11157 (make_int_cst): Remove macro.
11159 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11161 * tree.c (make_tre_binfo_stat): Remove _stat from name.
11162 * tree.h (make_tree_binfo_stat): Adjust prototype.
11163 (make_tree_binfo): Remove.
11165 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11167 * tree.c (copy_node_stat): Rename to copy_node.
11168 (build_distinct_type_copy): Adjust.
11169 * tree.h (copy_node_stat): Adjust prototype.
11170 (copy_node): Remove macro.
11172 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11174 * tree.c (make_node_stat): rename to make_node.
11175 (build_tree_list_stat): Adjust.
11176 (build0_stat): Likewise.
11177 (build2_stat): Likewise.
11178 (build3_stat): Likewise.
11179 (build4_stat): Likewise.
11180 (build5_stat): Likewise.
11181 (build_decl_stat): Likewise.
11182 * tree.h (make_node_stat): Adjust prototype.
11183 (make_node): remove macro.
11185 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
11187 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
11188 (PPC_FEATURE2_SCV): Likewise.
11189 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
11191 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11193 * config/aarch64/aarch64.c
11194 (aarch64_internal_mov_immediate): Add new special pattern.
11195 * config/aarch64/aarch64.md (*movdi_aarch64):
11196 Add reg/32bit const mov case.
11198 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11199 Richard Sandiford <richard.sandiford@linaro.org>
11201 * config/aarch64/aarch64.md (mov<mode>): Generalize.
11202 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11203 Add integer and movi cases.
11204 (movi-split-hf-df-sf split, fp16): New.
11205 (enabled): Added TARGET_FP_F16INST.
11206 * config/aarch64/iterators.md (GPF_HF): New.
11207 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
11209 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11211 * config/aarch64/aarch64.c
11212 (aarch64_simd_container_mode): Add prototype.
11213 (aarch64_expand_mov_immediate): Add HI support.
11214 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
11215 (aarch64_can_const_movi_rtx_p): New.
11216 (aarch64_preferred_reload_class):
11217 Remove restrictions of using FP registers for certain SIMD operations.
11218 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
11219 (aarch64_valid_floating_const): Add integer move validation.
11220 (aarch64_simd_imm_scalar_p): Remove.
11221 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
11222 (aarch64_legitimate_constant_p): Expand list of supported cases.
11223 * config/aarch64/aarch64-protos.h
11224 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
11225 (aarch64_reinterpret_float_as_int): New.
11226 (aarch64_simd_imm_scalar_p): Remove.
11227 * config/aarch64/constraints.md (Uvi): New.
11228 (Dd): Split into Ds and new Dd.
11229 * config/aarch64/aarch64.md (*movsi_aarch64):
11231 (*movdi_aarch64): Add SIMD mov case.
11233 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11235 * tree-predcom.c: (struct chain): Handle store-store chain in which
11236 stores for elimination only store loop invariant values.
11237 (execute_pred_commoning_chain): Ditto.
11238 (prepare_initializers_chain_store_elim): Ditto.
11239 (prepare_finalizers): Ditto.
11240 (is_inv_store_elimination_chain): New function.
11241 (initialize_root_vars_store_elim_1): New function.
11243 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11245 * tree-predcom.c: Revise general description of the pass.
11246 (enum chain_type): New enum type for store elimination.
11247 (struct chain): New field supporting store elimination.
11248 (struct component): Ditto.
11249 (dump_chain): Dump store-stores chain.
11250 (release_chain): Release resources.
11251 (split_data_refs_to_components): Compute and create component
11252 contains only stores for elimination.
11253 (get_chain_last_ref_at): New function.
11254 (make_invariant_chain): Initialization.
11255 (make_rooted_chain): Specify chain type in parameter and record it.
11256 (add_looparound_copies): Skip for store-stores chain.
11257 (determine_roots_comp): Compute type of chain and pass it to
11259 (initialize_root_vars_store_elim_2): New function.
11260 (finalize_eliminated_stores): New function.
11261 (remove_stmt): Handle store for elimination.
11262 (execute_pred_commoning_chain): Execute predictive commoning on
11263 store-store chains.
11264 (determine_unroll_factor): Skip unroll for store-stores chain.
11265 (prepare_initializers_chain_store_elim): New function.
11266 (prepare_initializers_chain): Hanlde store-store chain.
11267 (prepare_finalizers_chain, prepare_finalizers): New function.
11268 (tree_predictive_commoning_loop): Return integer value indicating
11269 if loop is unrolled or lcssa form is corrupted.
11270 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
11272 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11274 * tree-predcom.c (initialize_root): Delete.
11275 (execute_pred_commoning_chain): Initialize root vars and replace
11276 reference of non-combined chain directly, rather than call above
11279 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11281 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
11282 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
11284 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11286 * tree-predcom.c (struct chain): New field init_seq.
11287 (release_chain): Release init_seq.
11288 (prepare_initializers_chain): Record intialization stmts in above
11290 (insert_init_seqs): New function.
11291 (tree_predictive_commoning_loop): Call insert_init_seqs.
11293 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11295 * tree-predcom.c (determine_roots_comp): Skip trivial components.
11297 2017-07-28 Richard Biener <rguenther@suse.de>
11299 * match.pd: Remove superfluous :c.
11300 * genmatch.c (simplify::id): Add member.
11301 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
11303 (current_id): New global.
11304 (dt_node::parent): Move from ...
11305 (dt_operand::parent): ... here. Add for_id member.
11306 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
11307 (decision_tree::find_node): Relax order requirement when
11308 merging DT_TRUE nodes to ones inbetween the current simplify
11309 and the one we try to merge with. Add diagnostic whenever
11310 we need to enforce pattern order by not merging.
11311 (decision_tree::insert): Set current_id.
11312 (decision_tree::print_node): Dump parent node and for_id.
11313 (parser::last_id): Add member.
11314 (parser::push_simplify): Assign unique id.
11315 (parser::parser): Initialize last_id.
11317 2017-07-28 Martin Liska <mliska@suse.cz>
11320 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
11321 gimple_build_debug_bind.
11323 2017-07-28 Richard Biener <rguenther@suse.de>
11325 PR tree-optimization/81502
11326 * match.pd: Add pattern combining BIT_INSERT_EXPR with
11328 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
11330 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
11331 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
11332 for BIT_FIELD_REF args.
11333 * fold-const.c (make_bit_field_ref): Likewise.
11334 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
11336 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11339 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
11340 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
11341 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
11342 Or it into SANITIZER_UNDEFINED.
11343 * ubsan.c: Include gimple-fold.h and varasm.h.
11344 (ubsan_expand_ptr_ifn): New function.
11345 (instrument_pointer_overflow): New function.
11346 (maybe_instrument_pointer_overflow): New function.
11347 (instrument_object_size): Formatting fix.
11348 (pass_ubsan::execute): Call instrument_pointer_overflow
11349 and maybe_instrument_pointer_overflow.
11350 * internal-fn.c (expand_UBSAN_PTR): New function.
11351 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
11352 * sanitizer.def (__ubsan_handle_pointer_overflow,
11353 __ubsan_handle_pointer_overflow_abort): New builtins.
11354 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
11355 * internal-fn.def (UBSAN_PTR): New internal function.
11356 * opts.c (sanitizer_opts): Add pointer-overflow.
11357 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
11358 * fold-const.c (build_range_check): Compute pointer range check in
11359 integral type if pointer arithmetics would be needed. Formatting
11362 2017-07-28 Martin Liska <mliska@suse.cz>
11365 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
11366 parameters that are of a variable-length.
11368 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
11370 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
11372 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
11373 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
11374 (CRT_CALL_STATIC_FUNCTION): Likewise.
11375 (ASM_DEFAULT_SPEC): New define.
11376 (ASM_SPEC32): Likewise.
11377 (ASM_SPEC64): Likewise.
11378 (ASM_SPEC_COMMON): Likewise.
11379 (ASM_SPEC): Likewise.
11380 (INVALID_64BIT): Likewise.
11381 (LINK_OS_DEFAULT_SPEC): Likewise.
11382 (LINK_OS_SPEC32): Likewise.
11383 (LINK_OS_SPEC64): Likewise.
11384 (POWERPC_LINUX): Likewise.
11385 (PTRDIFF_TYPE): Likewise.
11386 (RESTORE_FP_PREFIX): Likewise.
11387 (RESTORE_FP_SUFFIX): Likewise.
11388 (SAVE_FP_PREFIX): Likewise.
11389 (SAVE_FP_SUFFIX): Likewise.
11390 (SIZE_TYPE): Likewise.
11391 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
11392 (TARGET_64BIT): Likewise.
11393 (TARGET_64BIT): Likewise.
11394 (TARGET_AIX): Likewise.
11395 (WCHAR_TYPE_SIZE): Likewise.
11396 (WCHAR_TYPE): Undefine.
11397 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
11398 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
11399 (CPP_OS_RTEMS_SPEC): Delete.
11400 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
11401 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
11403 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
11405 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11407 PR tree-optimization/81578
11408 * tree-parloops.c (build_new_reduction): Bail out if
11409 reduction_code isn't one of the standard OpenMP reductions.
11410 Move the details printing after that decision.
11412 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11414 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
11415 related to reload_in_progress.
11416 (splat_input_operand): Likewise.
11417 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
11419 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
11421 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
11422 (TARGET_INSTANTIATE_DECLS): Likewise.
11423 (legitimate_indexed_address_p): Delete reload_in_progress code.
11424 (rs6000_debug_legitimate_address_p): Likewise.
11425 (rs6000_eliminate_indexed_memrefs): Likewise.
11426 (rs6000_emit_le_vsx_store): Likewise.
11427 (rs6000_emit_move_si_sf_subreg): Likewise.
11428 (rs6000_emit_move): Likewise.
11429 (register_to_reg_type): Likewise.
11430 (rs6000_pre_atomic_barrier): Likewise.
11431 (rs6000_machopic_legitimize_pic_address): Likewise.
11432 (rs6000_allocate_stack_temp): Likewise.
11433 (rs6000_address_for_fpconvert): Likewise.
11434 (rs6000_address_for_altivec): Likewise.
11435 (rs6000_secondary_memory_needed_rtx): Delete function.
11436 (rs6000_check_sdmode): Likewise.
11437 (rs6000_alloc_sdmode_stack_slot): Likewise.
11438 (rs6000_instantiate_decls): Likewise.
11439 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
11440 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
11441 Delete reload_in_progress.
11442 (*vec_reload_and_plus_<mptrsize>): Likewise.
11443 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
11444 (vsx_div_v2di): Likewise.
11445 (vsx_udiv_v2di): Likewise.
11447 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11449 * config/rs6000/rs6000.opt (mlra): Replace with stub.
11450 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
11451 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
11452 (rs6000_debug_reg_global): Delete print of LRA status.
11453 (rs6000_option_override_internal): Delete dead LRA related code.
11454 (rs6000_lra_p): Delete function.
11455 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
11457 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
11459 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
11460 * config/riscv/rtems.h: New file.
11462 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11463 Sudakshina Das <sudi.das@arm.com>
11465 * config/aarch64/aarch64.md
11466 (define_split for and<mode>3nr_compare): Move
11467 non aarch64_logical_operand to a register.
11468 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
11469 register immediate operand to a register.
11470 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
11472 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11474 PR middle-end/81564
11475 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
11477 2017-07-27 Richard Biener <rguenther@suse.de>
11479 PR tree-optimization/81573
11480 PR tree-optimization/81494
11481 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
11482 multi defuse cycle case.
11484 2017-07-27 Richard Biener <rguenther@suse.de>
11486 PR tree-optimization/81571
11487 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
11490 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
11492 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
11493 earlier and only if MASK_FPU is set. Adjust formatting.
11495 2017-07-27 Martin Liska <mliska@suse.cz>
11497 * opt-functions.awk: Add validation of value of Init.
11498 * optc-gen.awk: Pass new argument.
11500 2017-07-27 Martin Liska <mliska@suse.cz>
11502 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
11503 Fix wrong condition.
11505 2017-07-27 Martin Liska <mliska@suse.cz>
11507 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
11508 BBs and edges seen by autoFDO.
11510 2017-07-27 Richard Biener <rguenther@suse.de>
11512 PR tree-optimization/81502
11513 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
11514 with incompatible but same sized type.
11515 (execute_update_addresses_taken): Likewise.
11517 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
11519 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
11520 flag_tree_loop_vectorize rather than flag_tree_vectorize.
11522 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11525 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
11526 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
11527 Change s_operand to memory_operand.
11529 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
11531 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
11532 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
11533 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
11534 Emit instructions rather than returning an expression. Handle TFmode
11535 and KFmode by casting to TImode.
11536 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
11537 (rs6000_emit_le_vsx_store): Likewise.
11538 * config/rs6000/vsx.md (VSX_TI): New iterator.
11539 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
11540 (*vsx_le_undo_permute_<mode>): Likewise.
11541 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
11542 emit the split sequence.
11543 (*vsx_le_perm_store_<mode>): Likewise.
11545 2017-07-27 Jakub Jelinek <jakub@redhat.com>
11547 PR tree-optimization/81555
11548 PR tree-optimization/81556
11549 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
11550 if true, force CHANGED for the recursive invocation.
11551 (reassociate_bb): Remember original length of ops array, pass
11552 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
11554 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
11555 attributes for noipa attribute. For naked attribute use
11556 lookup_attribute first before lookup_attribute_spec.
11557 * final.c (rest_of_handle_final): Disable IPA RA for functions with
11559 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
11560 for functions with noipa attribute.
11561 (cgraph_externally_visible_p): Return true for functions with noipa
11563 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
11564 for functions with noipa attribute.
11565 * doc/extend.texi: Document noipa function attribute.
11566 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
11567 also for functions with noipa attribute.
11568 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
11570 2017-07-26 Andrew Pinski <apinski@cavium.com>
11572 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
11573 vec_unalign_load_cost and vec_unalign_store_cost.
11575 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11577 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
11578 -mvsx-small-integer option.
11579 (ISA_3_0_MASKS_IEEE): Likewise.
11580 (OTHER_VSX_VECTOR_MASKS): Likewise.
11581 (POWERPC_MASKS): Likewise.
11582 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
11583 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
11584 code, only testing for DImode being allowed in non-VSX floating
11586 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
11587 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
11589 (rs6000_option_override_internal): Delete -mvsx-small-integer.
11590 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
11591 TARGET_P8_VECTOR test.
11592 (rs6000_secondary_reload_simple_move): Likewise.
11593 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
11594 since TARGET_P9_VECTOR was already tested.
11595 (rs6000_opt_masks): Remove -mvsx-small-integer.
11596 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
11597 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11599 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
11600 test for TARGET_VEXTRACTUB was used, and that uses
11602 (p9 extract splitter): Likewise.
11603 (vsx_extract_<mode>_di_p9): Likewise.
11604 (vsx_extract_<mode>_store_p9): Likewise.
11605 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
11606 for TARGET_P9_VECTOR was used. Delete code that is now dead with
11607 the elimination of TARGET_VSX_SMALL_INTEGER.
11608 (vsx_extract_<mode>_p8): Likewise.
11609 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
11610 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
11611 (vsx_set_<mode>_p9): Likewise.
11612 (vsx_set_v4sf_p9): Likewise.
11613 (vsx_set_v4sf_p9_zero): Likewise.
11614 (vsx_insert_extract_v4sf_p9): Likewise.
11615 (vsx_insert_extract_v4sf_p9_2): Likewise.
11616 * config/rs6000/rs6000.md (sign extend splitter): Change
11617 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
11618 (floatsi<mode>2_lfiwax_mem): Likewise.
11619 (floatunssi<mode>2_lfiwzx_mem): Likewise.
11620 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
11621 since a test for TARGET_P9_VECTOR was used.
11622 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11623 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
11624 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11625 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11626 TARGET_P8_VECTOR test.
11627 (fix_trunc<mode>si2_stfiwx): Likewise.
11628 (fix_trunc<mode>si2_internal): Likewise.
11629 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
11630 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11632 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11633 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11634 TARGET_P8_VECTOR test.
11635 (fixuns_trunc<mode>si2_stfiwx): Likewise.
11636 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
11637 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11639 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11640 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
11641 since a test for TARGET_P9_VECTOR was used.
11642 (splitter for loading small constants): Likewise.
11644 2017-07-26 Andrew Pinski <apinski@cavium.com>
11646 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
11649 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
11652 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
11653 (fp_valid_at): Likewise.
11655 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11657 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
11658 (qdf24xx_addrcost_table): Likewise.
11659 (cortexa57_tunings): Update to use generic_branch_cost.
11660 (cortexa72_tunings): Likewise.
11661 (cortexa73_tunings): Likewise.
11662 (qdf24xx_tunings): Likewise.
11664 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11666 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
11667 (thunderx2t99_branch_cost): Likewise.
11668 (cortexa35_tunings): Update to use generic_branch_cost.
11669 (cortexa53_tunings): Likewise.
11670 (cortexa57_tunings): Likewise.
11671 (cortexa72_tunings): Likewise.
11672 (cortexa73_tunings): Likewise.
11673 (thunderx2t99_tunings): Likewise.
11675 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11677 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
11678 (sparc_option_override): Honour MASK_FSMULD.
11679 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
11680 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
11681 * config/sparc/sparc.opt (mfsmuld): New option.
11682 * doc/invoke.texi (mfsmuld): Document option.
11684 2017-07-26 Marek Polacek <polacek@redhat.com>
11686 PR middle-end/70992
11687 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
11689 2017-07-26 Richard Biener <rguenther@suse.de>
11691 * gimple-match-head.c (do_valueize): Return OP if valueize
11693 (get_def): New helper to get at the def stmt of a SSA name
11694 if valueize allows.
11695 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
11696 do_valueize to get at the def stmt.
11697 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
11699 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
11701 PR middle-end/46932
11702 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
11704 2017-07-26 Martin Liska <mliska@suse.cz>
11707 * function.c (expand_function_start): Make expansion of
11708 nonlocal_goto_save_area after parm_birth_insn.
11710 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11712 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
11713 from all CPU target flags enable members.
11715 2017-07-26 Richard Biener <rguenther@suse.de>
11717 * genmatch.c (dt_simplify::gen): Make iterator vars const.
11718 (decision_tree::gen): Make 'type' const.
11719 (write_predicate): Likewise.
11721 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11723 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
11724 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
11725 (rs6000_option_override_internal): Likewise.
11726 (rs6000_expand_vector_set): Likewise.
11727 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
11728 (TARGET_UPPER_REGS_SF): Likewise.
11729 (TARGET_UPPER_REGS_DI): Likewise.
11730 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
11731 (TARGET_DIRECT_MOVE_64BIT): Likewise.
11732 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
11733 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11734 (Splitters for DI constants in Altivec registers): Likewise.
11735 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
11736 (vsx_set_v4sf_p9): Likewise.
11737 (vsx_set_v4sf_p9_zero): Likewise.
11738 (vsx_insert_extract_v4sf_p9): Likewise.
11739 (vsx_insert_extract_v4sf_p9_2): Likewise.
11741 2017-07-25 Carl Love <cel@us.ibm.com>
11743 * doc/extend.texi: Update the built-in documentation file for the
11744 existing built-in functions
11745 vector signed char vec_cnttz (vector signed char);
11746 vector unsigned char vec_cnttz (vector unsigned char);
11747 vector signed short vec_cnttz (vector signed short);
11748 vector unsigned short vec_cnttz (vector unsigned short);
11749 vector signed int vec_cnttz (vector signed int);
11750 vector unsigned int vec_cnttz (vector unsigned int);
11751 vector signed long long vec_cnttz (vector signed long long);
11752 vector unsigned long long vec_cnttz (vector unsigned long long);
11754 2017-07-25 Andrew Pinski <apinski@cavium.com>
11756 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
11757 accesses where the use is for the first operand of a BIT_INSERT.
11759 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11762 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11763 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11765 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11767 * config/i386/gstabs.h: Delete.
11768 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
11770 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
11772 * config/i386/i386.c (ix86_decompose_address): Do not check for
11773 register RTX when looking at index_reg or base_reg.
11774 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
11776 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
11778 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
11779 to update EH info here.
11781 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11783 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
11785 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11787 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
11789 2017-07-25 Torsten Duwe <duwe@suse.de>
11791 * common.opt: Introduce -fpatchable-function-entry
11792 command line option, and its variables function_entry_patch_area_size
11793 and function_entry_patch_area_start.
11794 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
11795 including a two-value parser.
11796 * target.def (print_patchable_function_entry): New target hook.
11797 * targhooks.h (default_print_patchable_function_entry): New function.
11798 * targhooks.c (default_print_patchable_function_entry): Likewise.
11799 * toplev.c (process_options): Switch off IPA-RA if
11800 patchable function entries are being generated.
11801 * varasm.c (assemble_start_function): Look at the
11802 patchable-function-entry command line switch and current
11803 function attributes and maybe generate NOP instructions by
11804 calling the print_patchable_function_entry hook.
11805 * doc/extend.texi: Document patchable_function_entry attribute.
11806 * doc/invoke.texi: Document -fpatchable_function_entry
11807 command line option.
11808 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
11810 * doc/tm.texi: Re-generate.
11812 2017-07-25 Jakub Jelinek <jakub@redhat.com>
11815 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
11816 TARGET_AVX512DQ rather than TARGET_AVX512BW.
11818 2017-07-25 Tamar Christina <tamar.christina@arm.com>
11820 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
11822 2017-07-25 Richard Biener <rguenther@suse.de>
11824 PR tree-optimization/81455
11825 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
11826 not walk in cycles when looking for guards.
11828 2017-07-25 Richard Biener <rguenther@suse.de>
11830 PR tree-optimization/81529
11831 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
11832 when optimizing backedge uses.
11834 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
11836 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
11838 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
11839 to dl_section_ref. On AIX, append an expression to subtract
11840 the size of the section length to dl_section_ref.
11842 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
11844 * configure.ac: If any of the config.* scripts fail, exit 1.
11845 * configure: Regenerate.
11847 2017-07-25 Richard Biener <rguenther@suse.de>
11849 PR middle-end/81546
11850 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
11851 of immediate uses, be more verbose on errors.
11853 2017-07-25 Richard Biener <rguenther@suse.de>
11855 PR tree-optimization/81510
11856 * tree-vect-loop.c (vect_is_simple_reduction): When the
11857 reduction stmt is not inside the loop bail out.
11859 2017-07-25 Richard Biener <rguenther@suse.de>
11861 PR tree-optimization/81303
11862 * tree-vect-loop-manip.c (vect_loop_versioning): Build
11863 profitability check against LOOP_VINFO_NITERSM1.
11865 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11867 * domwalk.c (cmp_bb_postorder): Simplify.
11868 (sort_bbs_postorder): New function. Use it...
11869 (dom_walker::walk): ...here to optimize common cases.
11871 2017-07-25 Martin Liska <mliska@suse.cz>
11874 * ipa-visibility.c (function_and_variable_visibility): Make the
11875 redirection just on target that supports aliasing.
11876 Fix GNU coding style.
11878 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
11881 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
11883 * config/arm/rtems.h: Likewise.
11884 * config/bfin/rtems.h: Likewise.
11885 * config/i386/rtemself.h: Likewise.
11886 * config/lm32/rtems.h: Likewise.
11887 * config/m32c/rtems.h: Likewise.
11888 * config/m68k/rtemself.h: Likewise.
11889 * config/microblaze/rtems.h: Likewise.
11890 * config/mips/rtems.h: Likewise.
11891 * config/moxie/rtems.h: Likewise.
11892 * config/nios2/rtems.h: Likewise.
11893 * config/powerpcspe/rtems.h: Likewise.
11894 * config/rs6000/rtems.h: Likewise.
11895 * config/rtems.h: Likewise.
11896 * config/sh/rtems.h: Likewise.
11897 * config/sh/rtemself.h: Likewise.
11898 * config/sparc/rtemself.h: Likewise.
11900 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11903 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
11904 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
11905 * tree-ssa-structalias.c (alias_get_name): Same.
11907 2017-07-25 Bin Cheng <bin.cheng@arm.com>
11910 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
11911 instructions if no du chain is found.
11913 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11915 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
11917 2017-07-25 Richard Biener <rguenther@suse.de>
11919 PR middle-end/81505
11920 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
11923 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11925 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
11926 upper-regs options.
11927 (ISA_2_7_MASKS_SERVER): Likewise.
11928 (ISA_3_0_MASKS_IEEE): Likewise.
11929 (OTHER_P8_VECTOR_MASKS): Likewise.
11930 (OTHER_VSX_VECTOR_MASKS): Likewise.
11931 (POWERPC_MASKS): Likewise.
11932 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
11933 duplicate list of options.
11934 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
11935 explicit -mupper-regs options.
11936 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
11937 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
11938 alias for -mupper-regs-df.
11939 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11940 (rs6000_init_hard_regno_mode_ok): Likewise.
11941 (rs6000_option_override_internal): Likewise.
11942 (rs6000_opt_masks): Likewise.
11943 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
11944 options in terms of whether -mvsx or -mpower8-vector was used.
11945 (TARGET_UPPER_REGS_DI): Likewise.
11946 (TARGET_UPPER_REGS_SF): Likewise.
11947 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
11948 -mupper-regs-* options.
11950 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11952 * passes.c (emergency_dump_function): Print some empty lines and a
11953 header before the RTL dump.
11955 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11957 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
11959 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
11962 * config/aarch64/aarch64.c (aarch64_classify_symbol):
11963 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
11965 2017-07-24 Carl Love <cel@us.ibm.com>
11967 * config/rs6000/rs6000-c.c: Add support for built-in functions
11968 vector float vec_extract_fp32_from_shorth (vector unsigned short);
11969 vector float vec_extract_fp32_from_shortl (vector unsigned short);
11970 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
11971 vec_extract_fp_from_shortl): Add defines for the two builtins.
11972 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
11973 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
11975 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
11976 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
11977 * doc/extend.texi: Update the built-in documentation file for the
11978 new built-in function.
11980 2017-07-24 Jakub Jelinek <jakub@redhat.com>
11983 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
11985 * doc/generic.texi: Likewise.
11986 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11987 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11989 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
11991 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
11992 (aarch64_mls_elt_merge<mode>): Likewise.
11994 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
11996 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11997 having __cxa_atexit.
11999 2017-07-23 Michael Collison <michael.collison@arm.com>
12001 * config/arm/arm.c (arm_option_override): Deprecate
12002 use of -mstructure-size-boundary.
12003 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
12004 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
12006 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12009 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12010 Reduce cost estimate for direct moves.
12012 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
12015 * config/i386/i386.c (ix86_option_override_internal): Disable
12016 BMI, BMI2 and TBM instructions for -m16.
12018 2017-07-21 Carl Love <cel@us.ibm.com>
12020 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12021 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12022 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12023 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12024 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12025 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12026 VMULOSW): New enum "unspec" values.
12027 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12028 altivec_vmulosw): New patterns.
12029 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12030 VMULOSW): Add definitions.
12032 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
12034 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
12035 (qdf24xx): Likewise.
12036 * config/aarch64/aarch64-options-extensions.def (rdma); New.
12037 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
12038 (AARCH64_FL_V8_1): Renumber.
12039 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
12040 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
12041 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
12042 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
12043 rdma to feature modifiers list.
12045 2017-07-21 Yury Gribov <tetra2005@gmail.com>
12047 PR middle-end/56727
12048 * ipa-visibility (function_and_variable_visibility): Convert
12049 recursive PLT call to direct call if appropriate.
12051 2017-07-21 Andrew Pinski <apinski@cavium.com>
12053 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
12054 operand 1 to see if the types precision matches.
12055 * fold-const.c (operand_equal_p): Likewise.
12057 2017-07-21 Richard Biener <rguenther@suse.de>
12059 PR tree-optimization/81303
12060 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
12061 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
12062 (vect_peeling_hash_get_lowest_cost): Adjust.
12063 (vect_enhance_data_refs_alignment): Likewise. Use
12064 vect_get_peeling_costs_all_drs to compute the penalty for no
12065 peeling to match up costs.
12067 2017-07-21 Richard Biener <rguenther@suse.de>
12069 PR tree-optimization/81500
12070 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
12071 we didn't identify a reduction path.
12073 2017-07-21 Tom de Vries <tom@codesourcery.com>
12074 Cesar Philippidis <cesar@codesourcery.com>
12076 PR gcov-profile/81442
12077 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
12080 2017-07-21 Tom de Vries <tom@codesourcery.com>
12083 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
12085 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
12086 nvptx_override_options_after_change.
12088 2017-07-21 Ulrich Drepper <drepper@redhat.com>
12090 * dwarf2out.c (output_file_names): Avoid double testing for
12091 dwarf_version >= 5.
12093 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
12095 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
12097 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12099 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
12101 (try_crossjump_to_edge): Do not punt on partitioned functions.
12103 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12105 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
12106 Put all BBs reachable only via paths crossing cold region to cold
12108 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
12110 2016-07-21 Richard Biener <rguenther@suse.de>
12112 PR tree-optimization/81303
12113 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
12114 into account prologue and epilogue iterations when raising
12115 min_profitable_iters to sth at least covering one vector iteration.
12117 2017-07-21 Tamar Christina <tamar.christina@arm.com>
12119 * config/arm/arm.c (arm_test_cpu_arch_dat):
12122 2017-07-20 Nathan Sidwell <nathan@acm.org>
12124 Remove TYPE_METHODS.
12125 * tree.h (TYPE_METHODS): Delete.
12126 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
12127 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
12128 (dbxout_type_methods): Scan TYPE_FIELDS.
12129 (dbxout_type): Don't check TYPE_METHODS here.
12130 * function.c (use_register_for_decl): Always ignore register for
12131 class types when not optimizing.
12132 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
12133 * tree.c (free_lang_data_in_type): Stitch out member functions and
12134 templates from TYPE_FIELDS.
12135 (build_distinct_type_copy, verify_type_variant,
12136 verify_type): Member fns are on TYPE_FIELDS.
12137 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
12138 * tree-pretty-print.c (dump_generic_node): Likewise.
12140 2017-07-20 Jakub Jelinek <jakub@redhat.com>
12143 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
12144 V2TImode and V4TImode.
12145 (ix86_expand_vector_extract): Likewise.
12146 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
12147 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
12148 (ssescalarmode): Handle V4TImode and V2TImode.
12149 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
12150 (*vec_extractv2ti, *vec_extractv4ti): New insns.
12151 (VEXTRACTI128_MODE): New mode iterator.
12152 (splitter for *vec_extractv?ti first element): New.
12153 (VEC_INIT_MODE): New mode iterator.
12154 (vec_init<mode>): Consolidate 3 expanders into one using
12155 VEC_INIT_MODE mode iterator.
12157 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12159 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
12160 non_spilled_static_chain_regno_p.
12162 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12164 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
12166 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
12168 * bb-reorder.c (connect_traces): Allow copying of blocks within
12171 2017-07-20 Richard Biener <rguenther@suse.de>
12173 * gimple.h (gimple_phi_result): Add gphi * overload.
12174 (gimple_phi_result_ptr): Likewise.
12175 (gimple_phi_arg): Likewise. Adjust index assert to only
12176 allow actual argument accesses rather than all slots available
12178 (gimple_phi_arg_def): Add gphi * overload.
12179 * tree-phinodes.c (make_phi_node): Initialize only actual
12181 (resize_phi_node): Clear memory not covered by old node,
12182 do not initialize excess argument slots.
12183 (reserve_phi_args_for_new_edge): Initialize new argument slot
12186 2017-07-20 Bin Cheng <bin.cheng@arm.com>
12188 PR tree-optimization/81388
12190 2016-07-21 Bin Cheng <bin.cheng@arm.com>
12192 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
12193 by removing computation of may_be_zero.
12195 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12196 Tom de Vries <tom@codesourcery.com>
12198 PR middle-end/81030
12199 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
12200 when gimple level profile disagrees with what RTL expander did.
12202 2017-07-20 Richard Biener <rguenther@suse.de>
12204 PR tree-optimization/61171
12205 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
12206 (vect_analyze_stmt): Add slp instance parameter.
12207 (vectorizable_reduction): Likewise.
12208 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
12209 (vect_is_simple_reduction): Deal with chains not detected
12210 as SLP reduction chain, specifically not properly associated
12211 chains containing a mix of plus/minus.
12212 (get_reduction_op): Remove.
12213 (get_initial_defs_for_reduction): Simplify, pass in whether
12214 this is a reduction chain, pass in the SLP node for the PHIs.
12215 (vect_create_epilog_for_reduction): Get the SLP instance as
12217 (vectorizable_reduction): Get the SLP instance as arg.
12218 During analysis remember the SLP node with the PHIs in the
12219 instance. Simplify getting at the vectorized reduction PHIs.
12220 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
12221 through SLP instance.
12222 (vect_slp_analyze_operations): Likewise.
12223 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
12224 (vect_transform_stmt): Likewise.
12226 2017-07-20 Tom de Vries <tom@codesourcery.com>
12228 PR tree-optimization/81489
12229 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
12230 read of phi arg location to before loop that modifies phi.
12232 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12234 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
12237 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12239 PR middle-end/81331
12240 * except.c (execute): Fix ordering issue.
12242 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12244 PR rtl-optimization/81423
12245 * combine.c (make_compound_operation_int): Don't try to optimize
12246 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
12248 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12250 PR rtl-optimization/81423
12251 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
12252 with a constant that is -1 in the truncated to mode.
12254 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12256 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
12257 (determine_unlikely_bbs): ... here.
12258 * predict.h (propagate_unlikely_bbs_forward): Declare.
12259 * cfgexpand.c (pass_expand::execute): Use it.
12260 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
12262 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
12263 propagate_unlikely_bbs_forward.
12265 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12267 PR middle-end/81331
12268 * except.c (maybe_add_nop_after_section_switch): New function.
12271 2017-07-19 Tom de Vries <tom@codesourcery.com>
12273 * gimple.h (gimple_phi_set_arg): Make assert more strict.
12275 2017-07-19 Tom de Vries <tom@codesourcery.com>
12277 * gimple.h (gimple_phi_arg): Make assert more strict.
12279 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
12281 * config.gcc (powerpc*-*-*): Add mmintrin.h.
12282 * config/rs6000/mmintrin.h: New file.
12283 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
12285 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12287 PR tree-optimization/81346
12288 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
12290 2017-07-19 Tom de Vries <tom@codesourcery.com>
12292 * config/nvptx/nvptx.md (VECIM): Add V2DI.
12294 2017-07-19 Tom de Vries <tom@codesourcery.com>
12296 * config/nvptx/nvptx-modes.def: Add V2DImode.
12297 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
12298 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
12299 (nvptx_output_mov_insn): Handle lack of mov.b128.
12300 (nvptx_print_operand): Handle 'H' and 'L' codes.
12301 (nvptx_vector_mode_supported): Allow V2DImode.
12302 (nvptx_preferred_simd_mode): New function.
12303 (nvptx_data_alignment): New function.
12304 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
12305 nvptx_preferred_simd_mode.
12306 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
12308 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
12310 2017-07-19 Tom de Vries <tom@codesourcery.com>
12312 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
12313 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
12314 (nvptx_vector_mode_supported): New function. Allow V2SImode.
12315 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
12316 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
12317 (mov<VECIM>_insn): New define_insn.
12318 (define_expand "mov<VECIM>): New define_expand.
12320 2017-07-19 Tom de Vries <tom@codesourcery.com>
12322 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
12324 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12326 PR tree-optimization/81346
12327 * fold-const.h (fold_div_compare, range_check_type): Declare.
12328 * fold-const.c (range_check_type): New function.
12329 (build_range_check): Use range_check_type.
12330 (fold_div_compare): No longer static, rewritten into
12331 a match.pd helper function.
12332 (fold_comparison): Don't call fold_div_compare here.
12333 * match.pd (X / C1 op C2): New optimization using fold_div_compare
12334 as helper function.
12336 2017-07-19 Nathan Sidwell <nathan@acm.org>
12338 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
12339 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
12340 * tree.c (find_decls_types_r, verify_type): Use
12341 TYPE_{MIN,MAX}_VALUE_RAW.
12342 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
12343 (hash_tree): Likewise.
12344 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12346 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12349 2017-07-18 Tom de Vries <tom@codesourcery.com>
12351 PR middle-end/81464
12352 * omp-expand.c (expand_omp_for_static_chunk): Handle
12353 equal-argument loop exit phi.
12355 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
12358 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
12359 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
12360 operand 2 predicate.
12361 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
12362 operand 2 predicate.
12363 (ror,rol -> rorx splitters): Use const_int_operand as
12364 operand 2 predicate.
12366 2017-06-18 Richard Biener <rguenther@suse.de>
12368 PR tree-optimization/81410
12369 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
12370 the gap in the ! slp_perm SLP case after each group.
12372 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12374 PR middle-end/81463
12375 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
12378 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12380 PR middle-end/81462
12381 * predict.c (set_even_probabilities): Cleanup; do not affect
12382 probabilities that are already known.
12383 (combine_predictions_for_bb): Call even when count is set.
12385 2017-07-18 Nathan Sidwell <nathan@acm.org>
12387 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
12390 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12393 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
12394 optimization for loop niter analysis.
12396 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
12399 * config/avr/avr.c (avr_optimize_casesi): Don't use
12400 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
12402 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12404 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
12405 body_cost_vec from _vect_peel_extended_info.
12406 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
12407 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
12410 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12412 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
12414 2017-07-18 Richard Biener <rguenther@suse.de>
12416 PR tree-optimization/80620
12417 PR tree-optimization/81403
12418 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
12419 info when re-using a VN table entry.
12421 2017-07-18 Richard Biener <rguenther@suse.de>
12423 PR tree-optimization/81418
12424 * tree-vect-loop.c (vectorizable_reduction): Properly compute
12425 vectype_in. Verify that with lane-reducing reduction operations
12426 we have a single def-use cycle.
12428 2017-07-17 Carl Love <cel@us.ibm.com>
12430 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
12432 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12433 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12434 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12435 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12436 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12437 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12438 VMULOSW): New enum "unspec" values.
12439 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12440 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12441 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12442 altivec_vmulosw): New patterns.
12443 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12444 VMULOSW): Add definitions.
12445 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
12447 * config/alpha/alpha.c: Include predict.h.
12449 2017-07-17 Yury Gribov <tetra2005@gmail.com>
12451 * tree-vrp.c (compare_assert_loc): Fix comparison function
12452 to return predictable results.
12454 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12456 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
12458 (subdi3): Likewise.
12459 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
12460 * doc/invoke.texi (mexpand-adddi): Update text.
12462 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12464 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
12465 that also clobbers the CC register. The old expand code is moved
12467 (*arc_clzsi2): ... here.
12468 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
12469 the CC register. The old expand code is moved to ...
12470 (arc_ctzsi2): ... here.
12472 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12474 * config/arc/arc.opt (mindexed-loads): Use initial value
12475 TARGET_INDEXED_LOADS_DEFAULT.
12476 (mauto-modify-reg): Use initial value
12477 TARGET_AUTO_MODIFY_REG_DEFAULT.
12478 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12479 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12480 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12481 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12483 2017-07-17 Martin Liska <mliska@suse.cz>
12486 * opts.c (finish_options): Do not allow -fgnu-tm
12487 w/ -fsanitize={kernel-,}address. Say sorry.
12489 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12492 * tree-loop-distribution.c (classify_partition): Only assert on
12493 numer of iterations.
12494 (merge_dep_scc_partitions): Delete prameter. Update function call.
12495 (distribute_loop): Remove code handling loop with unknown niters.
12496 (pass_loop_distribution::execute): Skip loop with unknown niters.
12498 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12501 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
12502 function sort_partitions_by_post_order.
12504 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12506 PR tree-optimization/81374
12507 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
12508 the max index of basic blocks, rather than number of basic blocks.
12510 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12512 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
12514 (arc_legitimate_pic_operand_p): Likewise.
12515 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
12517 (arc_needs_pcl_p): Likewise.
12518 (arc_legitimate_pc_offset_p): Likewise.
12519 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
12520 function is also used in constrains.md.
12521 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
12522 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
12523 PLUS. Only return true/false in known cases, otherwise assert.
12524 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
12525 is already called in arc_legitimate_constant_p.
12526 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
12528 (LEGITIMATE_PIC_OPERAND_P): Use
12529 arc_raw_symbolic_reference_mentioned_p function.
12530 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
12535 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12536 Andrew Burgess <andrew.burgess@embecosm.com>
12538 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
12539 (arc_return_address_register): New function.
12540 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
12541 (arc_handle_fndecl_attribute): Add naked attribute.
12542 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12543 (TARGET_WARN_FUNC_RETURN): Likewise.
12544 (arc_allocate_stack_slots_for_args): New function.
12545 (arc_warn_func_return): Likewise.
12546 (machine_function): Change type fn_type.
12547 (arc_compute_function_type): Consider new naked function type,
12548 change function return type.
12549 (arc_must_save_register): Adapt to handle new
12550 arc_compute_function_type's return type.
12551 (arc_expand_prologue): Likewise.
12552 (arc_expand_epilogue): Likewise.
12553 (arc_return_address_regs): Delete.
12554 (arc_return_address_register): New function.
12555 (arc_epilogue_uses): Use above function.
12556 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
12557 (arc_function_type): Change encoding, add naked type.
12558 (ARC_INTERRUPT_P): Change to handle the new encoding.
12559 (ARC_FAST_INTERRUPT_P): Likewise.
12560 (ARC_NORMAL_P): Define.
12561 (ARC_NAKED_P): Likewise.
12562 (arc_compute_function_type): Delete prototype.
12563 * config/arc/arc.md (in_ret_delay_slot): Use
12564 arc_return_address_register function.
12565 (simple_return): Likewise.
12566 (p_return_i): Likewise.
12568 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12570 PR tree-optimization/81428
12571 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
12572 can't be built for those types.
12574 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12576 Remove stuff dead since r239246.
12578 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
12579 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
12580 (avr_inform_devices): Remove dead stuff.
12582 2017-07-17 Tamar Christina <tamar.christina@arm.com>
12584 * config/arm/arm_neon.h: Fix softp typo.
12586 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12588 PR tree-optimization/81365
12589 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
12590 aggregate moves onto bb predecessor edges, make sure there are no
12591 loads that could alias the lhs in between the start of bb and the
12594 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12597 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
12598 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
12599 [LSHIFTRT, outer_code = TRUNCATE]: Same.
12601 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12603 PR tree-optimization/81396
12604 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
12605 (init_symbolic_number): Initialize it to 1.
12606 (perform_symbolic_merge): Add n_ops from both operands into the new
12608 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
12609 without base_addr as useless if they need more than one operation.
12610 (bswap_replace): Handle !bswap case for NULL base_addr.
12612 2017-07-17 Tom de Vries <tom@codesourcery.com>
12615 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
12618 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
12620 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
12621 conditional builtin define __FIX_LEON3FT_B2BST.
12623 2017-07-17 Daniel Cederman <cederman@gaisler.com>
12625 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
12626 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
12629 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12631 PR rtl-optimization/81424
12632 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
12633 to remove potential trapping from operands if -fnon-call-exceptions.
12635 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12637 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
12638 profile_proability for scalling.
12639 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
12641 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12643 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
12645 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12647 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
12648 fixpoint arithmetics.
12650 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12652 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
12653 fixpoint arithmetics.
12655 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12657 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
12658 fixpoint arithmetics.
12660 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12662 * profile-count.h (profile_probability::from_reg_br_prob_note,
12663 profile_probability::to_reg_br_prob_note): New functions.
12664 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
12665 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
12666 * predict.c (probability_reliable_p): Update.
12667 (edge_probability_reliable_p): Update.
12668 (br_prob_note_reliable_p): Update.
12669 (invert_br_probabilities): Update.
12670 (add_reg_br_prob_note): New function.
12671 (combine_predictions_for_insn): Update.
12672 * asan.c (asan_clear_shadow): Update.
12673 * cfgbuild.c (compute_outgoing_frequencies): Update.
12674 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12675 (update_br_prob_note): Update.
12676 (rtl_verify_edges): Update.
12677 (purge_dead_edges): Update.
12678 (fixup_reorder_chain): Update.
12679 * emit-rtl.c (try_split): Update.
12680 * ifcvt.c (cond_exec_process_insns): Update.
12681 (cond_exec_process_if_block): Update.
12682 (dead_or_predicable): Update.
12683 * internal-fn.c (expand_addsub_overflow): Update.
12684 (expand_neg_overflow): Update.
12685 (expand_mul_overflow): Update.
12686 * loop-doloop.c (doloop_modify): Update.
12687 * loop-unroll.c (compare_and_jump_seq): Update.
12688 * optabs.c (emit_cmp_and_jump_insn_1): Update.
12689 * predict.h: Update.
12690 * reorg.c (mostly_true_jump): Update.
12692 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
12693 * config/alpha/alpha.c (emit_unlikely_jump): Update.
12694 * config/arc/arc.c: (emit_unlikely_jump): Update.
12695 * config/arm/arm.c: (emit_unlikely_jump): Update.
12696 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
12697 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
12698 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
12699 (ix86_print_operand): Update.
12700 (ix86_split_fp_branch): Update.
12701 (predict_jump): Update.
12702 * config/ia64/ia64.c (ia64_print_operand): Update.
12703 * config/mmix/mmix.c (mmix_print_operand): Update.
12704 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
12705 (rs6000_expand_split_stack_prologue): Update.
12706 * config/rs6000/rs6000.c: Update.
12707 * config/s390/s390.c (s390_expand_vec_strlen): Update.
12708 (s390_expand_vec_movstr): Update.
12709 (s390_expand_cs_tdsi): Update.
12710 (s390_expand_split_stack_prologue): Update.
12711 * config/sh/sh.c (sh_print_operand): Update.
12712 (expand_cbranchsi4): Update.
12713 (expand_cbranchdi4): Update.
12714 * config/sparc/sparc.c (output_v9branch): Update.
12715 * config/spu/spu.c (get_branch_target): Update.
12716 (ea_load_store_inline): Update.
12717 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
12718 * config/tilepro/tilepro.c: Update.
12720 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12722 * gimplify.c (mostly_copy_tree_r): Revert latest change.
12723 (gimplify_save_expr): Likewise.
12725 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12727 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
12729 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12731 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
12733 * timevar.def (TV_IPA_FNSUMMARY): Define.
12735 2017-07-16 Daniel Cederman <cederman@gaisler.com>
12737 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
12738 to back store errata sensitive sequence from being generated.
12739 (sqrtdf2_fix): Likewise.
12741 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12743 * tree-ssa-threadupdate.c (compute_path_counts,
12744 update_joiner_offpath_counts): Use profile_probability.
12746 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
12749 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12751 * config/arm/arm-c.c (arm_cpu_builtins): Define
12752 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12754 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
12756 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12757 array entries to represent __ieee128 versions of the
12758 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
12759 scalar_extract_sig, and scalar_insert_exp built-in functions.
12760 (altivec_resolve_overloaded_builtin): Add special case handling
12761 for the __builtin_scalar_insert_exp function, as represented by
12762 the P9V_BUILTIN_VEC_VSIEDP constant.
12763 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
12764 exponent support for __ieee128 argument.
12765 (VSESQP): Add scalar extract signature support for __ieee128
12767 (VSTDCNQP): Add scalar test negative support for __ieee128
12769 (VSIEQP): Add scalar insert exponent support for __int128 argument
12770 with __ieee128 result.
12771 (VSIEQPF): Add scalar insert exponent support for __ieee128
12772 argument with __ieee128 result.
12773 (VSTDCQP): Add scalar test data class support for __ieee128
12775 (VSTDCNQP): Add overload support for scalar test negative with
12776 __ieee128 argument.
12777 (VSTDCQP): Add overload support for scalar test data class
12778 __ieee128 argument.
12779 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
12780 UNSPEC_VSX_SXSIGDP.
12781 (UNSPEC_VSX_SIEXPQP): New constant.
12782 (xsxexpqp): New insn for VSX scalar extract exponent quad
12784 (xsxsigqp): New insn for VSX scalar extract significand quad
12786 (xsiexpqpf): New insn for VSX scalar insert exponent quad
12787 precision with floating point argument.
12788 (xststdcqp): New expand for VSX scalar test data class quad
12790 (xststdcnegqp): New expand for VSX scalar test negative quad
12792 (xststdcqp): New insn to match expansions for VSX scalar test data
12793 class quad precision and VSX scalar test negative quad precision.
12794 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
12795 special case operand checking to enforce that second operand of
12796 VSX scalar test data class with quad precision argument is a 7-bit
12798 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12799 prototypes and descriptions of __ieee128 versions of
12800 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
12801 scalar_test_data_class, and scalar_test_neg built-in functions.
12803 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12805 PR tree-optimization/81162
12806 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
12807 replace a negate with an add.
12809 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
12811 * doc/invoke.texi (arm/-mcpu): Document +crypto.
12813 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12815 * config/arm/arm-c.c (arm_cpu_builtins): Define
12816 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12818 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12820 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
12821 (armv8-r): Set ARM Cortex-R52 as default CPU.
12822 * config/arm/arm-tables.opt: Regenerate.
12823 * config/arm/arm-tune.md: Regenerate.
12824 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12826 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
12827 extension for -mcpu=cortex-r52.
12829 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12831 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
12832 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
12833 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
12834 (fp-armv8): Define it as FP_ARMv8 only.
12835 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
12836 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
12838 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
12839 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
12840 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
12841 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
12842 than TARGET_FPU_ARMV8.
12843 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
12844 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
12845 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
12847 * config/arm/neon.md (neon_vrint): Likewise.
12848 (neon_vcvt): Likewise.
12849 (neon_<fmaxmin_op><mode>): Likewise.
12850 (<fmaxmin><mode>3): Likewise.
12851 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
12852 * config/arm/predicates.md (arm_cond_move_operator): Check against
12853 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
12855 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
12857 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
12858 to top of function.
12860 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12862 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
12863 loop in comment with memset.
12865 2017-07-14 Martin Liska <mliska@suse.cz>
12867 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
12868 * dwarf2out.c (is_java): Remove the function.
12869 (output_pubname): Remove usage of the function.
12870 (lower_bound_default): Remove usage of DW_LANG_Java.
12871 (gen_compile_unit_die): Likewise.
12872 * gcc.c: Remove compiler defaults for .java and .zip files.
12873 * gimple-expr.c (remove_suffix): Change as there's no longer
12874 extension than 4-letter one.
12875 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
12876 (gimplify_save_expr): Likewise.
12877 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
12878 as it's possible even for other languages than Java.
12879 * langhooks.h (struct lang_hooks): Remove Java from a comment.
12880 * lto-opts.c (lto_write_options): Remove reference to Java.
12881 * opts.c (strip_off_ending): Update file extension handling.
12882 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
12883 * tree-eh.c (lower_resx): Likewise.
12884 * tree.c (free_lang_data_in_type): Remove dead code.
12885 (find_decls_types_r): Likewise.
12886 (build_common_builtin_nodes): Remove Java from a comment.
12887 (verify_type): Remove dead code.
12888 * varasm.c (assemble_external): Remove Java from a comment.
12890 2017-07-14 Martin Liska <mliska@suse.cz>
12892 * opts.c (finish_options): Add quotes.
12893 (common_handle_option): Likewise.
12895 2017-07-14 Martin Liska <mliska@suse.cz>
12897 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
12898 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
12899 Remove N_SO_PASCAL.
12900 * dwarf2out.c (lower_bound_default): Do not handle
12902 (gen_compile_unit_die): Likewise.
12903 * gcc.c: Remove default extension binding for GNU Pascal.
12904 * stmt.c: Remove Pascal language from a comment.
12905 * xcoffout.c: Likewise.
12907 2017-07-13 David Malcolm <dmalcolm@redhat.com>
12910 * diagnostic-show-locus.c (fixit_cmp): New function.
12911 (layout::layout): Sort m_fixit_hints.
12912 (column_range::column_range): Assert that the values are valid.
12913 (struct char_span): New struct.
12914 (correction::overwrite): New method.
12915 (struct source_line): New struct.
12916 (line_corrections::add_hint): Add assertions. Reimplement memcpy
12917 calls in terms of classes source_line and char_span, and
12918 correction::overwrite.
12919 (selftest::test_overlapped_fixit_printing_2): New function.
12920 (selftest::diagnostic_show_locus_c_tests): Call it.
12922 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
12924 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
12925 early if there is no lhs.
12927 2017-07-13 Martin Liska <mliska@suse.cz>
12929 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
12930 (gen_reference_type_die): Likewise.
12931 * stor-layout.c: Remove Pascal-related comment.
12933 2017-07-13 Martin Liska <mliska@suse.cz>
12935 * opts.c (finish_options): Add quotes to error messages.
12936 (parse_sanitizer_options): Likewise.
12938 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
12940 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
12942 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
12944 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
12946 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
12948 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
12950 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
12952 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12955 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
12956 provides the hardware capability bits, define the macro
12957 __BUILTIN_CPU_SUPPORTS__.
12958 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
12959 if GLIBC does not provide the hardware capability bits. Add a
12960 gcc_unreachable call if the built-in cpu function is neither
12961 __builtin_cpu_is nor __builtin_cpu_supports.
12962 (rs6000_get_function_versions_dispatcher): Change the warning
12963 that an old GLIBC is used which does not export the capability
12964 bits to be an error.
12965 * doc/extend.texi (target_clones attribute): Document the
12966 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
12967 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
12968 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
12969 the macros defined by GCC if the newer GLIBC is available.
12971 2017-07-12 Jeff Law <law@redhat.com>
12973 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
12974 remaining includes slightly.
12975 * config/riscv/riscv-builtins.c: Include profile-count.h.
12977 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12980 * config/avr/avr.c (avr_set_current_function): In diagnostic
12981 messages: Quote keywords and (parts of) identifiers.
12982 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
12985 2017-07-12 Carl Love <cel@us.ibm.com>
12987 * config/rs6000/rs6000-c.c: Add support for built-in functions
12988 vector bool char vec_revb (vector bool char);
12989 vector bool short vec_revb (vector short char);
12990 vector bool int vec_revb (vector bool int);
12991 vector bool long long vec_revb (vector bool long long);
12992 * doc/extend.texi: Update the built-in documentation file for the
12993 new built-in functions.
12995 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12997 * config/s390/s390.md: Remove movcc splitter.
12999 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13001 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
13002 load/store on condition.
13004 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13007 * config/avr/avr.c (avr_encode_section_info)
13008 [progmem && !TREE_READONLY]: Error if progmem object needs
13011 2017-07-11 Michael Collison <michael.collison@arm.com>
13013 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
13016 2017-07-11 Carl Love <cel@us.ibm.com>
13018 * config/rs6000/rs6000-c.c: Add support for builtins
13019 vector unsigned int vec_parity_lsbb (vector signed int);
13020 vector unsigned int vec_parity_lsbb (vector unsigned int);
13021 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
13022 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
13023 vector unsigned long long vec_parity_lsbb (vector signed long long);
13024 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
13025 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
13026 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
13027 * doc/extend.texi: Update the built-in documentation file for the
13028 new built-in functions.
13030 2017-07-11 David Malcolm <dmalcolm@redhat.com>
13032 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
13033 (layout::m_primary_loc): New field.
13034 (layout::layout): Initialize new field. Move location filtering
13035 logic from here to...
13036 (layout::maybe_add_location_range): ...this new method. Add
13037 support for filtering to just the lines already specified by other
13039 (layout::will_show_line_p): New method.
13040 (gcc_rich_location::add_location_if_nearby): New method.
13041 (selftest::test_add_location_if_nearby): New test function.
13042 (selftest::diagnostic_show_locus_c_tests): Call it.
13043 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
13046 2017-07-11 Tom de Vries <tom@codesourcery.com>
13048 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
13049 (bb_first_real_insn): New function.
13050 (nvptx_single): Add extra initialization of broadcasted condition
13053 2017-07-11 Nathan Sidwell <nathan@acm.org>
13055 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
13057 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
13059 * doc/extend.texi (AVR Function Attributes): Remove weblink to
13060 Binutils doc as TEXI will mess them up.
13061 * doc/invoke.texi (AVR Options): Same here.
13063 2017-07-11 Daniel Cederman <cederman@gaisler.com>
13065 * config/sparc/sparc.opt (mfix-ut700): New option.
13066 (mfix-gr712rc): Likewise.
13067 (sparc_fix_b2bst): New variable.
13068 * doc/invoke.texi (SPARC options): Document them.
13069 (ARM options): Fix warnings.
13070 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
13071 instructions to prevent sequences that can trigger the store-store
13072 errata for certain LEON3FT processors.
13073 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
13074 (sparc_option_override): Set sparc_fix_b2bst appropriately.
13075 * config/sparc/sparc.md (fix_b2bst): New attribute.
13076 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
13078 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
13081 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
13083 (*rsqrtsf2_sse): Ditto.
13085 (div<mode>3): Macroize insn from divdf3 and divsf3
13086 using MODEF mode iterator.
13088 2017-07-10 Martin Sebor <msebor@redhat.com>
13090 PR tree-optimization/80397
13091 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
13092 instead of testing for equality to INTEGER_TYPE.
13094 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
13096 * config.gcc: Remove uclibc from arc target spec.
13098 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
13100 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
13102 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13105 * lto-wrapper.c (remove_option): New function.
13106 (merge_and_complain): Merge PIC/PIE options more realistically.
13108 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13110 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
13114 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
13115 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
13116 * config.in: Regenerate.
13117 * configure: Regenerate.
13118 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
13119 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
13120 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
13121 (TARGET_GASISR_PROLOGUES): ...target mask.
13122 * common/config/avr/avr-common.c
13123 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
13124 Set -mgas-isr-prologues.
13125 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
13126 INSERT_PASS_BEFORE for it.
13127 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
13128 * config/avr/avr.c (avr_option_override)
13129 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
13130 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
13131 (avr_attribute_table) <no_gccisr>: Add new function attribute.
13132 (avr_set_current_function) <is_no_gccisr>: Init machine field.
13133 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
13135 (make_avr_pass_pre_proep): New function.
13136 (emit_push_sfr) <treg>: Add argument to function and use it
13137 instead of TMP_REG.
13138 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
13139 and set machine->gasisr.yes.
13140 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
13141 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
13142 __gcc_isr.n_pushed to .L__stack_usage.
13143 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
13144 (avr_asm_final_postscan_insn): ...this new static function.
13145 * config/avr/avr.h (machine_function)
13146 <is_no_gccisr, use_L__stack_usage>: New fields.
13147 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
13148 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
13149 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
13150 (gasisr, *gasisr): New expander and insn.
13151 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13152 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
13153 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
13155 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
13157 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
13160 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13162 Move jump-tables out of .text again.
13165 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
13166 (ASM_OUTPUT_ADDR_VEC): New function.
13167 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
13168 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
13169 INSN_ADDRESSes as asm comment.
13170 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
13171 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
13172 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
13173 * config/avr/avr.md (*tablejump): Adjust comment.
13174 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
13175 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
13177 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
13178 (avr_output_addr_vec): New proto.
13179 (avr_log_t) <insn_addresses>: New field.
13181 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
13184 * config/i386/i386.c (ix86_function_arg_advance): Set
13185 outgoing_args_on_stack to true if there are outgoing arguments
13187 (ix86_function_arg): Likewise.
13188 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
13189 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
13190 * config/i386/i386.h (machine_function): Add
13191 outgoing_args_on_stack.
13193 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
13195 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13196 supporting pthreds.
13197 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
13199 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
13201 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
13202 (REAL_H): Remove $(MACHMODE_H).
13203 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
13205 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
13206 $(MACHMODE_H) and double-int.h.
13207 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
13209 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
13212 2017-07-07 Andrew Pinski <apinski@cavium.com>
13214 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
13215 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
13217 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13219 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
13220 Add warning if GCC was not configured to link against a GLIBC that
13221 exports the hardware capability bits.
13222 (make_resolver_func): Make resolver function private and not a
13223 COMDAT function. Create the name with clone_function_name instead
13224 of make_unique_name.
13227 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
13228 correct operand in doing the split.
13230 2017-07-07 Carl Love <cel@us.ibm.com>
13232 * config/rs6000/rs6000-c: Add support for built-in function
13233 vector unsigned short vec_pack_to_short_fp32 (vector float,
13235 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
13236 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
13237 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
13238 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
13239 (convert_4f32_8i16): Add define_expand.
13240 * doc/extend.texi: Update the built-in documentation file for the
13241 new built-in function.
13243 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13245 * config/sparc/m8.md: New file.
13246 * config/sparc/sparc.md: Include m8.md.
13248 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13250 * config/sparc/sparc.opt: New option -mvis4b.
13251 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
13252 (sparc_option_override): Handle VIS4B.
13253 (enum sparc_builtins): Define
13254 SPARC_BUILTIN_DICTUNPACK{8,16,32},
13255 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
13256 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
13257 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
13258 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
13259 (check_constant_argument): New function.
13260 (sparc_vis_init_builtins): Define builtins
13261 __builtin_vis_dictunpack{8,16,32},
13262 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
13263 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
13264 __builtin_vis_fpcmpde{8,16,32}shl and
13265 __builtin_vis_fpcmpur{8,16,32}shl.
13266 (sparc_expand_builtin): Check that the constant operands to
13267 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
13268 constant and in range.
13269 * config/sparc/sparc-c.c (sparc_target_macros): Handle
13271 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
13272 (SPARC_IMM5_P): Likewise.
13273 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
13274 (enabled): Handle vis4b.
13275 (UNSPEC_DICTUNPACK): New unspec.
13276 (UNSPEC_FPCMPSHL): Likewise.
13277 (UNSPEC_FPUCMPSHL): Likewise.
13278 (UNSPEC_FPCMPDESHL): Likewise.
13279 (UNSPEC_FPCMPURSHL): Likewise.
13280 (cpu_feature): New CPU feature `vis4b'.
13281 (dictunpack{8,16,32}): New insns.
13282 (FPCSMODE): New mode iterator.
13283 (fpcscond): New code iterator.
13284 (fpcsucond): Likewise.
13285 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
13286 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
13287 (fpcmpde{8,16,32}{si,di}shl): Likewise.
13288 (fpcmpur{8,16,32}{si,di}shl): Likewise.
13289 * config/sparc/constraints.md: Define constraints `q' for unsigned
13290 2-bit integer constants and `t' for unsigned 5-bit integer
13292 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
13294 (imm5_operand_dictunpack16): Likewise.
13295 (imm5_operand_dictunpack32): Likewise.
13296 (imm2_operand): Likewise.
13297 * doc/invoke.texi (SPARC Options): Document -mvis4b.
13298 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
13299 ditunpack* and fpcmp*shl builtins.
13301 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13303 * config.gcc: Handle m8 in --with-{cpu,tune} options.
13304 * config.in: Add HAVE_AS_SPARC6 define.
13305 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
13307 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
13309 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
13310 (CPP_CPU_SPEC): Handle m8.
13311 (ASM_CPU_SPEC): Likewise.
13312 * config/sparc/sparc-opts.h (enum processor_type): Add
13314 * config/sparc/sparc.c (m8_costs): New struct.
13315 (sparc_option_override): Handle TARGET_CPU_m8.
13316 (sparc32_initialize_trampoline): Likewise.
13317 (sparc64_initialize_trampoline): Likewise.
13318 (sparc_issue_rate): Likewise.
13319 (sparc_register_move_cost): Likewise.
13320 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
13321 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
13322 (ASM_CPU64_DEFAULT_SPEC): Likewise.
13323 (CPP_CPU_SPEC): Handle M8.
13324 (ASM_CPU_SPEC): Likewise.
13325 (AS_M8_FLAG): Define.
13326 * config/sparc/sparc.md: Add m8 to the cpu attribute.
13327 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
13328 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
13330 * configure: Regenerate.
13331 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
13334 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13336 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
13338 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
13339 ("*movdi_insn_sp32"): Do not set v3pipe.
13340 ("*movsi_insn"): Likewise.
13341 ("*movdi_insn_sp64"): Likewise.
13342 ("*movsf_insn"): Likewise.
13343 ("*movdf_insn_sp32"): Likewise.
13344 ("*movdf_insn_sp64"): Likewise.
13345 ("*zero_extendsidi2_insn_sp64"): Likewise.
13346 ("*sign_extendsidi2_insn"): Likewise.
13347 ("*mov<VM32:mode>_insn"): Likewise.
13348 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13349 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13350 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13351 ("<vlop:code><VL:mode>3"): Likewise.
13352 ("*not_<vlop:code><VL:mode>3"): Likewise.
13353 ("*nand<VL:mode>_vis"): Likewise.
13354 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
13355 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
13356 ("one_cmpl<VL:mode>2"): Likewise.
13357 ("faligndata<VM64:mode>_vis"): Likewise.
13358 ("alignaddrsi_vis"): Likewise.
13359 ("alignaddrdi_vis"): Likweise.
13360 ("alignaddrlsi_vis"): Likewise.
13361 ("alignaddrldi_vis"): Likewise.
13362 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13363 ("bmaskdi_vis"): Likewise.
13364 ("bmasksi_vis"): Likewise.
13365 ("bshuffle<VM64:mode>_vis"): Likewise.
13366 ("cmask8<P:mode>_vis"): Likewise.
13367 ("cmask16<P:mode>_vis"): Likewise.
13368 ("cmask32<P:mode>_vis"): Likewise.
13369 ("pdistn<P:mode>_vis"): Likewise.
13370 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13372 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13374 * config/sparc/sparc.md ("subtype"): New insn attribute.
13375 ("*wrgsr_sp64"): Set insn subtype.
13376 ("*rdgsr_sp64"): Likewise.
13377 ("alignaddrsi_vis"): Likewise.
13378 ("alignaddrdi_vis"): Likewise.
13379 ("alignaddrlsi_vis"): Likewise.
13380 ("alignaddrldi_vis"): Likewise.
13381 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13382 ("fexpand_vis"): Likewise.
13383 ("fpmerge_vis"): Likewise.
13384 ("faligndata<VM64:mode>_vis"): Likewise.
13385 ("bshuffle<VM64:mode>_vis"): Likewise.
13386 ("cmask8<P:mode>_vis"): Likewise.
13387 ("cmask16<P:mode>_vis"): Likewise.
13388 ("cmask32<P:mode>_vis"): Likewise.
13389 ("fchksm16_vis"): Likewise.
13390 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
13391 ("fmean16_vis"): Likewise.
13392 ("fp<plusminus_insn>64_vis"): Likewise.
13393 ("<plusminus_insn>v8qi3"): Likewise.
13394 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13395 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
13396 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
13397 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
13398 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
13399 ("*movqi_insn"): Likewise.
13400 ("*movhi_insn"): Likewise.
13401 ("*movsi_insn"): Likewise.
13402 ("movsi_pic_gotdata_op"): Likewise.
13403 ("*movdi_insn_sp32"): Likewise.
13404 ("*movdi_insn_sp64"): Likewise.
13405 ("movdi_pic_gotdata_op"): Likewise.
13406 ("*movsf_insn"): Likewise.
13407 ("*movdf_insn_sp32"): Likewise.
13408 ("*movdf_insn_sp64"): Likewise.
13409 ("*zero_extendhisi2_insn"): Likewise.
13410 ("*zero_extendqihi2_insn"): Likewise.
13411 ("*zero_extendqisi2_insn"): Likewise.
13412 ("*zero_extendqidi2_insn"): Likewise.
13413 ("*zero_extendhidi2_insn"): Likewise.
13414 ("*zero_extendsidi2_insn_sp64"): Likewise.
13415 ("ldfsr"): Likewise.
13416 ("prefetch_64"): Likewise.
13417 ("prefetch_32"): Likewise.
13418 ("tie_ld32"): Likewise.
13419 ("tie_ld64"): Likewise.
13420 ("*tldo_ldub_sp32"): Likewise.
13421 ("*tldo_ldub1_sp32"): Likewise.
13422 ("*tldo_ldub2_sp32"): Likewise.
13423 ("*tldo_ldub_sp64"): Likewise.
13424 ("*tldo_ldub1_sp64"): Likewise.
13425 ("*tldo_ldub2_sp64"): Likewise.
13426 ("*tldo_ldub3_sp64"): Likewise.
13427 ("*tldo_lduh_sp32"): Likewise.
13428 ("*tldo_lduh1_sp32"): Likewise.
13429 ("*tldo_lduh_sp64"): Likewise.
13430 ("*tldo_lduh1_sp64"): Likewise.
13431 ("*tldo_lduh2_sp64"): Likewise.
13432 ("*tldo_lduw_sp32"): Likewise.
13433 ("*tldo_lduw_sp64"): Likewise.
13434 ("*tldo_lduw1_sp64"): Likewise.
13435 ("*tldo_ldx_sp64"): Likewise.
13436 ("*mov<VM32:mode>_insn"): Likewise.
13437 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13438 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13440 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13442 * config/sparc/sparc.md ("type"): New insn type viscmp.
13443 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
13445 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
13446 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
13447 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13448 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
13450 ("n7_vis_logical_11cycle"): Likewise.
13451 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
13452 * config/sparc/niagara2.md ("niag3_vis": Likewise.
13453 * config/sparc/niagara.md ("niag_vis"): Likewise.
13454 * config/sparc/ultra3.md ("us3_fga"): Likewise.
13455 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
13457 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13459 * config/sparc/sparc.md: New instruction type `bmask'.
13460 (bmaskdi_vis): Use the `bmask' type.
13461 (bmasksi_vis): Likewise.
13462 * config/sparc/ultra3.md (us3_array): Likewise.
13463 * config/sparc/niagara7.md (n7_array): Likewise.
13464 * config/sparc/niagara4.md (n4_array): Likewise.
13465 * config/sparc/niagara2.md (niag2_vis): Likewise.
13466 (niag3_vis): Likewise.
13467 * config/sparc/niagara.md (niag_vis): Likewise.
13469 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13471 * ipa-comdats.c: Remove optimize check from gate.
13472 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
13473 for functions not optimized.
13474 (ipa_fn_summary_read): Skip optimize check.
13475 (ipa_fn_summary_write): Likewise.
13476 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
13478 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
13480 (can_inline_edge_p): Check flag_pcc_struct_return for match.
13481 (check_callers): Give up on caller which is not optimized.
13482 (inline_small_functions): Likewise.
13483 (ipa_inline): Do not give up when not optimizing.
13484 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
13485 away unoptimizes cdtors.
13486 (whole_program_function_and_variable_visibility): Do
13487 ipa_discover_readonly_nonaddressable_vars in LTO mode.
13488 * ipa.c (process_references): Do not check optimize.
13489 (symbol_table::remove_unreachable_nodes): Update optimize check.
13490 (set_writeonly_bit): Update optimize check.
13491 (pass_ipa_cdtor_merge::gate): Do not check optimize.
13492 (pass_ipa_single_use::gate): Remove.
13494 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13496 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
13497 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
13498 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
13499 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
13500 permute_load, permute_store, adjust_extract, adjust_splat,
13501 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
13502 replace_swap_with_copy, dump_swap_insn_table,
13503 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
13504 recombine_lvx_pattern, recombine_stvx_pattern,
13505 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
13506 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
13507 to file rs6000-p8swap.c.
13508 * config/rs6000/rs6000-p8swap.c: New file.
13509 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
13510 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
13511 and rs6000*-*-* targets.
13513 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13515 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
13517 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13519 * lto-wrapper.c (merge_and_complain): Do not merge
13520 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
13521 fsigned_zeros, ftrapping_math, fwrapv.
13522 (append_compiler_options): Do not track these options.
13523 (append_linker_options): Likewie
13525 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13527 * cgraphunit.c (cgraph_node::finalize_function): When
13528 !flag_toplevel_reorde set no_reorder flag.
13529 (varpool_node::finalize_decl): Likewise.
13530 (symbol_table::compile): Drop no toplevel reorder path.
13532 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13534 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
13535 edges; zero probability is not better than uninitialized.
13537 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13539 * asan.h (asan_sanitize_allocas_p): Declare.
13540 * asan.c (asan_sanitize_allocas_p): New function.
13541 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
13542 (handle_builtin_alloca): Likewise.
13543 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
13544 if !asan_sanitize_allocas_p.
13545 * params.def (asan-instrument-allocas): Add new option.
13546 * params.h (ASAN_PROTECT_ALLOCAS): Define.
13547 * opts.c (common_handle_option): Disable allocas sanitization for
13550 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13552 * asan.c: Include gimple-fold.h.
13553 (get_last_alloca_addr): New function.
13554 (handle_builtin_stackrestore): Likewise.
13555 (handle_builtin_alloca): Likewise.
13556 (asan_emit_allocas_unpoison): Likewise.
13557 (get_mem_refs_of_builtin_call): Add new parameter, remove const
13558 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
13559 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
13560 (instrument_builtin_call): Pass gimple iterator to
13561 get_mem_refs_of_builtin_call.
13562 (last_alloca_addr): New global.
13563 * asan.h (asan_emit_allocas_unpoison): Declare.
13564 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
13565 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
13566 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
13567 if function calls alloca.
13568 * gimple-fold.c (replace_call_with_value): Remove static keyword.
13569 * gimple-fold.h (replace_call_with_value): Declare.
13570 * internal-fn.c: Include asan.h.
13571 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
13572 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
13574 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13576 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
13577 (C_SELFTEST_FLAGS): New.
13578 (CPP_SELFTEST_FLAGS): New.
13579 (SELFTEST_DEPS): New, from deps of s-selftest.
13580 (C_SELFTEST_DEPS): New, from deps of s-selftest.
13581 (CPP_SELFTEST_DEPS): New.
13582 (selftest): Add dependency on s-selftest-c++.
13583 (s-selftest): Rename to...
13584 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
13585 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
13586 than SELFTEST_FLAGS.
13587 (selftest-gdb): Rename to...
13588 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
13590 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
13591 (selftest-valgrind): Rename to...
13592 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
13594 (selftest-valgrind): Reintroduce as an alias for
13595 selftest-c-valgrind.
13596 (s-selftest-c++): New.
13597 (selftest-c++-gdb): New.
13598 (selftest-c++-valgrind): New.
13600 2017-07-06 Olivier Hainque <hainque@adacore.com>
13602 * gcc.c (process_command): When deciding if undefined variables
13603 should be ignored when processing specs, accept "gcc -v" as well.
13605 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13607 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
13608 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
13610 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13612 * config/arm/arm-cpus.in (armv8-r): Add new entry.
13613 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
13614 * config/arm/arm-tables.opt: Regenerate.
13615 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
13617 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
13619 2017-07-06 Carl Love <cel@us.ibm.com>
13621 * ChangeLog: Clean up from mid air collision
13623 2017-07-06 Carl Love <cel@us.ibm.com>
13625 * config/rs6000/rs6000-c.c: Add support for built-in functions
13626 vector signed int vec_subc (vector signed int, vector signed int);
13627 vector signed __int128 vec_subc (vector signed __int128,
13628 vector signed __int128);
13629 vector unsigned __int128 vec_subc (vector unsigned __int128,
13630 vector unsigned __int128);
13631 vector signed int vec_sube (vector signed int, vector signed int,
13632 vector signed int);
13633 vector unsigned int vec_sube (vector unsigned int,
13634 vector unsigned int,
13635 vector unsigned int);
13636 vector signed __int128 vec_sube (vector signed __int128,
13637 vector signed __int128,
13638 vector signed__int128);
13639 vector unsigned __int128 vec_sube (vector unsigned __int128,
13640 vector unsigned __int128,
13641 vector unsigned __int128);
13642 vector signed int vec_subec (vector signed int, vector signed int,
13643 vector signed int);
13644 vector unsigned int vec_subec (vector unsigned int,
13645 vector unsigned int,
13646 vector unsigned int);
13647 vector signed __int128 vec_subec (vector signed __int128,
13648 vector signed __int128,
13649 vector signed__int128);
13650 vector unsigned __int128 vec_subec (vector unsigned __int128,
13651 vector unsigned __int128,
13652 vector unsigned __int128);
13653 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
13654 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
13655 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
13656 BU_ALTIVEC_OVERLOAD_X definitions.
13657 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
13658 * doc/extend.texi: Update the built-in documentation file for the new
13659 built-in functions.
13661 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13664 * diagnostic-show-locus.c (layout::layout): Use start and finish
13665 spelling location for the start and finish of each range.
13666 * genmatch.c (linemap_client_expand_location_to_spelling_point):
13667 Add unused aspect param.
13668 * input.c (expand_location_1): Add "aspect" param, and use it
13669 to access the correct part of the location.
13670 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
13672 (expand_location_to_spelling_point): Likewise.
13673 (linemap_client_expand_location_to_spelling_point): Add "aspect"
13674 param, and pass it to expand_location_1.
13676 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
13678 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
13679 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
13680 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
13681 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
13682 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
13683 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
13684 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
13685 _mm_maskz_getmant_ss): New intrinsics.
13686 (__builtin_ia32_getexpss128_mask): Changed to ...
13687 __builtin_ia32_getexpss128_round ... this.
13688 (__builtin_ia32_getexpsd128_mask): Changed to ...
13689 __builtin_ia32_getexpsd128_round ... this.
13690 * config/i386/i386-builtin-types.def
13691 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
13692 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
13693 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
13694 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
13695 __builtin_ia32_getmantss_mask_round): New builtins.
13696 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
13697 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
13698 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
13699 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
13700 * config/i386/sse.md
13701 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
13702 avx512f_sgetexp<mode><mask_scalar_name>
13703 <round_saeonly_scalar_name> ... this.
13704 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
13705 %0, %1, %2<round_saeonly_op3>}): Changed to ...
13706 vgetexp<ssescalarmodesuffix>
13707 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13708 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
13709 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
13710 avx512f_vgetmant<mode><mask_scalar_name>
13711 <round_saeonly_scalar_name> ... this.
13712 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
13713 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
13714 vgetmant<ssescalarmodesuffix>
13715 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
13716 %0<mask_scalar_operand4>, %1, %2
13717 <round_saeonly_scalar_mask_op4>, %3} ... this.
13718 * config/i386/subst.md (mask_scalar_operand4,
13719 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
13720 round_saeonly_scalar_nimm_predicate): New subst attributes.
13722 2017-07-06 Julia Koval <julia.koval@intel.com>
13724 * config/i386/i386.c (ix86_erase_embedded_rounding):
13725 Remove code for old rounding pattern.
13727 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
13729 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
13731 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
13733 * doc/sourcebuild.texi (Test Directives, Variants of
13734 dg-require-support): Add documentation for dg-require-stack-check.
13736 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
13738 * config/i386/subst.md (mask_scalar, round_scalar,
13739 round_saeonly_scalar): New meta-templates.
13740 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
13741 round_scalar_mask_operand3, round_scalar_mask_op3,
13742 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
13743 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
13744 round_saeonly_scalar_constraint,
13745 round_saeonly_scalar_prefix): New subst attribute.
13746 * config/i386/sse.md
13747 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
13748 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
13749 <round_scalar_name> ... this.
13750 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
13751 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
13752 <round_scalar_name> ... this.
13753 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
13754 <sse>_vm<code><mode>3<mask_scalar_name>
13755 <round_saeonly_scalar_name> ... this.
13756 (v<plusminus_mnemonic><ssescalarmodesuffix>
13757 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13758 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13759 v<plusminus_mnemonic><ssescalarmodesuffix>
13760 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13761 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13762 (v<multdiv_mnemonic><ssescalarmodesuffix>
13763 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13764 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13765 v<multdiv_mnemonic><ssescalarmodesuffix>
13766 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13767 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13768 (v<maxmin_float><ssescalarmodesuffix>
13769 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
13770 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
13771 v<maxmin_float><ssescalarmodesuffix>
13772 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13773 %0<mask_scalar_operand3>, %1, %<iptr>2
13774 <round_saeonly_scalar_mask_op3>} ... this.
13776 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
13778 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
13779 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
13781 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
13782 Alan Hayward <alan.hayward@arm.com>
13783 David Sherwood <david.sherwood@arm.com>
13785 * combine.c (simplify_if_then_else): Remove "enum" before
13787 * compare-elim.c (can_eliminate_compare): Likewise.
13788 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13790 (aarch64_lookup_simd_builtin_type): Likewise.
13791 (aarch64_simd_builtin_type): Likewise.
13792 (aarch64_init_simd_builtin_types): Likewise.
13793 (aarch64_simd_expand_args): Likewise.
13794 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
13796 (aarch64_reverse_mask): Likewise.
13797 (aarch64_simd_emit_reg_reg_move): Likewise.
13798 (aarch64_gen_adjusted_ldpstp): Likewise.
13799 (aarch64_ccmp_mode_to_code): Likewise.
13800 (aarch64_operands_ok_for_ldpstp): Likewise.
13801 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13802 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
13804 (aarch64_min_divisions_for_recip_mul): Likewise.
13805 (aarch64_reassociation_width): Likewise.
13806 (aarch64_get_condition_code_1): Likewise.
13807 (aarch64_simd_emit_reg_reg_move): Likewise.
13808 (aarch64_simd_attr_length_rglist): Likewise.
13809 (aarch64_reverse_mask): Likewise.
13810 (aarch64_operands_ok_for_ldpstp): Likewise.
13811 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13812 (aarch64_gen_adjusted_ldpstp): Likewise.
13813 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
13815 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
13816 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13817 (arm_lookup_simd_builtin_type): Likewise.
13818 (arm_simd_builtin_type): Likewise.
13819 (arm_init_simd_builtin_types): Likewise.
13820 (arm_expand_builtin_args): Likewise.
13821 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
13822 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
13823 (ft32_setup_incoming_varargs): Likewise.
13824 (ft32_function_arg): Likewise.
13825 (ft32_function_arg_advance): Likewise.
13826 (ft32_pass_by_reference): Likewise.
13827 (ft32_arg_partial_bytes): Likewise.
13828 (ft32_valid_pointer_mode): Likewise.
13829 (ft32_addr_space_pointer_mode): Likewise.
13830 (ft32_addr_space_legitimate_address_p): Likewise.
13831 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
13833 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
13834 (ix86_emit_outlined_ms2sysv_restore): Likewise.
13835 (iamcu_alignment): Likewise.
13836 (canonicalize_vector_int_perm): Likewise.
13837 (ix86_noce_conversion_profitable_p): Likewise.
13838 (ix86_mpx_bound_mode): Likewise.
13839 (ix86_operands_ok_for_move_multiple): Likewise.
13840 * config/microblaze/microblaze-protos.h
13841 (microblaze_expand_conditional_branch_reg): Likewise.
13842 * config/microblaze/microblaze.c
13843 (microblaze_expand_conditional_branch_reg): Likewise.
13844 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13846 (rs6000_reassociation_width): Likewise.
13847 (rs6000_invalid_binary_op): Likewise.
13848 (fusion_p9_p): Likewise.
13849 (emit_fusion_p9_load): Likewise.
13850 (emit_fusion_p9_store): Likewise.
13851 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
13853 (riscv_hard_regno_mode_ok_p): Likewise.
13854 (riscv_address_insns): Likewise.
13855 (riscv_split_symbol): Likewise.
13856 (riscv_legitimize_move): Likewise.
13857 (riscv_function_value): Likewise.
13858 (riscv_hard_regno_nregs): Likewise.
13859 (riscv_expand_builtin): Likewise.
13860 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
13861 (riscv_build_integer): Likewise.
13862 (riscv_split_integer): Likewise.
13863 (riscv_legitimate_constant_p): Likewise.
13864 (riscv_cannot_force_const_mem): Likewise.
13865 (riscv_regno_mode_ok_for_base_p): Likewise.
13866 (riscv_valid_base_register_p): Likewise.
13867 (riscv_valid_offset_p): Likewise.
13868 (riscv_valid_lo_sum_p): Likewise.
13869 (riscv_classify_address): Likewise.
13870 (riscv_legitimate_address_p): Likewise.
13871 (riscv_address_insns): Likewise.
13872 (riscv_load_store_insns): Likewise.
13873 (riscv_force_binary): Likewise.
13874 (riscv_split_symbol): Likewise.
13875 (riscv_force_address): Likewise.
13876 (riscv_legitimize_address): Likewise.
13877 (riscv_move_integer): Likewise.
13878 (riscv_legitimize_const_move): Likewise.
13879 (riscv_legitimize_move): Likewise.
13880 (riscv_address_cost): Likewise.
13881 (riscv_subword): Likewise.
13882 (riscv_output_move): Likewise.
13883 (riscv_canonicalize_int_order_test): Likewise.
13884 (riscv_emit_int_order_test): Likewise.
13885 (riscv_function_arg_boundary): Likewise.
13886 (riscv_pass_mode_in_fpr_p): Likewise.
13887 (riscv_pass_fpr_single): Likewise.
13888 (riscv_pass_fpr_pair): Likewise.
13889 (riscv_get_arg_info): Likewise.
13890 (riscv_function_arg): Likewise.
13891 (riscv_function_arg_advance): Likewise.
13892 (riscv_arg_partial_bytes): Likewise.
13893 (riscv_function_value): Likewise.
13894 (riscv_pass_by_reference): Likewise.
13895 (riscv_setup_incoming_varargs): Likewise.
13896 (riscv_print_operand): Likewise.
13897 (riscv_elf_select_rtx_section): Likewise.
13898 (riscv_save_restore_reg): Likewise.
13899 (riscv_for_each_saved_reg): Likewise.
13900 (riscv_register_move_cost): Likewise.
13901 (riscv_hard_regno_mode_ok_p): Likewise.
13902 (riscv_hard_regno_nregs): Likewise.
13903 (riscv_class_max_nregs): Likewise.
13904 (riscv_memory_move_cost): Likewise.
13905 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
13906 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
13907 (rl78_addr_space_address_mode): Likewise.
13908 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13910 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13911 (rs6000_reassociation_width): Likewise.
13912 (rs6000_invalid_binary_op): Likewise.
13913 (fusion_p9_p): Likewise.
13914 (emit_fusion_p9_load): Likewise.
13915 (emit_fusion_p9_store): Likewise.
13916 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
13917 (ok_for_simple_move_operands): Likewise.
13918 (ok_for_simple_move_strict_operands): Likewise.
13919 (ok_for_simple_arith_logic_operands): Likewise.
13920 (visium_legitimize_reload_address): Likewise.
13921 (visium_select_cc_mode): Likewise.
13922 (output_cbranch): Likewise.
13923 (visium_split_double_move): Likewise.
13924 (visium_expand_copysign): Likewise.
13925 (visium_expand_int_cstore): Likewise.
13926 (visium_expand_fp_cstore): Likewise.
13927 * config/visium/visium.c (visium_pass_by_reference): Likewise.
13928 (visium_function_arg): Likewise.
13929 (visium_function_arg_advance): Likewise.
13930 (visium_libcall_value): Likewise.
13931 (visium_setup_incoming_varargs): Likewise.
13932 (visium_legitimate_constant_p): Likewise.
13933 (visium_legitimate_address_p): Likewise.
13934 (visium_legitimize_address): Likewise.
13935 (visium_secondary_reload): Likewise.
13936 (visium_register_move_cost): Likewise.
13937 (visium_memory_move_cost): Likewise.
13938 (prepare_move_operands): Likewise.
13939 (ok_for_simple_move_operands): Likewise.
13940 (ok_for_simple_move_strict_operands): Likewise.
13941 (ok_for_simple_arith_logic_operands): Likewise.
13942 (visium_function_value_1): Likewise.
13943 (rtx_ok_for_offset_p): Likewise.
13944 (visium_legitimize_reload_address): Likewise.
13945 (visium_split_double_move): Likewise.
13946 (visium_expand_copysign): Likewise.
13947 (visium_expand_int_cstore): Likewise.
13948 (visium_expand_fp_cstore): Likewise.
13949 (visium_split_cstore): Likewise.
13950 (visium_select_cc_mode): Likewise.
13951 (visium_split_cbranch): Likewise.
13952 (output_cbranch): Likewise.
13953 (visium_print_operand_address): Likewise.
13954 * expmed.c (flip_storage_order): Likewise.
13955 * expmed.h (emit_cstore): Likewise.
13956 (flip_storage_order): Likewise.
13957 * genrecog.c (validate_pattern): Likewise.
13958 * hsa-gen.c (gen_hsa_addr): Likewise.
13959 * internal-fn.c (expand_arith_overflow): Likewise.
13960 * ira-color.c (allocno_copy_cost_saving): Likewise.
13961 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13962 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
13963 (process_invariant_for_inheritance): Likewise.
13964 * lra-eliminations.c (move_plus_up): Likewise.
13965 * omp-low.c (lower_oacc_reductions): Likewise.
13966 * simplify-rtx.c (simplify_subreg): Likewise.
13967 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
13968 (TARGET_CHKP_BOUND_MODE): Likewise..
13969 * targhooks.c (default_chkp_bound_mode): Likewise.
13970 (default_setup_incoming_vararg_bounds): Likewise.
13971 * targhooks.h (default_chkp_bound_mode): Likewise.
13972 (default_setup_incoming_vararg_bounds): Likewise.
13973 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
13974 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
13975 (have_whole_vector_shift): Likewise.
13976 * tree-vect-stmts.c (vectorizable_load): Likewise.
13977 * doc/tm.texi: Regenerate.
13979 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13981 Graceful degrade if Binutils PR21472 is not available.
13984 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
13985 .rodata in flash test fails.
13986 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
13987 * confgure: Regenerate.
13988 * config.in: Regenerate.
13989 * config/avr/avr.c (avr_asm_named_section)
13990 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
13991 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
13992 (avr_asm_init_sections): Same.
13994 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13996 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
13997 (fma<VH:mode>4_intrinsic): Likewise.
13998 (*fmsub<VCVTF:mode>4): Likewise.
13999 (*fmsub<VH:mode>4_intrinsic): Likewise.
14001 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14004 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
14005 Don't depend on "optimize > 0".
14006 (out_movhi_r_mr, out_movqi_mr_r): Same.
14007 (out_movhi_mr_r, out_movqi_r_mr): Same.
14008 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
14009 io_address_operand on "optimize > 0".
14011 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14013 * tree-loop-distribution.c: Add general explanantion on the pass.
14014 (generate_loops_for_partition): Mark distributed loop.
14015 (pg_add_dependence_edges): New parameter. Handle alias data
14016 dependence specially and record it in the parameter if asked.
14017 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
14018 (init_partition_graph_vertices, add_partition_graph_edge): New.
14019 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
14020 (free_partition_graph_vdata, build_partition_graph): New.
14021 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
14022 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
14023 (data_ref_segment_size, latch_dominated_by_data_ref): New.
14024 (compute_alias_check_pairs, version_loop_by_alias_check): New.
14025 (version_for_distribution_p, finalize_partitions): New.
14026 (distribute_loop): Handle alias data dependence specially. Factor
14027 out loop fusion code as functions and call these functions.
14029 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14031 * tree-loop-distribution.c (classify_partition): New parameter and
14032 better handle reduction statement.
14033 (rdg_build_partitions): Revise comment.
14034 (distribute_loop): Compute statements in all partitions and pass it
14035 to classify_partition.
14037 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14039 * tree-loop-distribution.c (enum partition_type): New.
14040 (struct partition): New field type.
14041 (partition_merge_into): Add parameter. Update partition type.
14042 (data_dep_in_cycle_p, update_type_for_merge): New functions.
14043 (build_rdg_partition_for_vertex): Compute partition type.
14044 (rdg_build_partitions): Dump partition type.
14045 (distribute_loop): Update calls to partition_merge_into.
14047 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14049 * tree-loop-distribution.c (struct ddr_hasher): New.
14050 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
14052 (classify_partition): Call get_data_dependence.
14053 (pg_add_dependence_edges): Ditto.
14054 (distribute_loop): Release data dependence hash table.
14056 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14058 * tree-loop-distribution.c (ref_base_address): Delete.
14059 (similar_memory_accesses): Rename ...
14060 (share_memory_accesses): ... to this. Check if partitions access
14061 the same memory reference.
14062 (distribute_loop): Call share_memory_accesses.
14064 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14066 * tree-loop-distribution.c (struct partition): New field recording
14067 its data reference.
14068 (partition_alloc, partition_free): Init and release data refs.
14069 (partition_merge_into): Merge data refs.
14070 (build_rdg_partition_for_vertex): Collect data refs for partition.
14071 (pg_add_dependence_edges): Change parameters from vector to bitmap.
14073 (distribute_loop): Remve data refs from vertice data of partition
14076 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14078 * tree-loop-distribution.c (params.h): Include header file.
14079 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
14080 (datarefs_vec): New global var.
14081 (create_rdg_vertices): Use datarefs_vec directly.
14082 (free_rdg): Don't free data references.
14083 (build_rdg): Update use. Don't free data references.
14084 (distribute_loop): Compute global variable for data references.
14085 Bail out if there are too many data references.
14087 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14089 * tree-loop-distribution.c (loop_nest): New global var.
14090 (build_rdg): Use loop directly, rather than loop nest.
14091 (pg_add_dependence_edges): Remove loop nest parameter. Use global
14093 (distribute_loop): Compute global variable loop nest. Update use.
14095 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14097 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
14098 (partition_merge_into): New parameter. Dump reason for fusion.
14099 (distribute_loop): Update use of partition_merge_into.
14101 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14103 * tree-loop-distribution.c (bb_top_order_index): New.
14104 (bb_top_order_index_size, bb_top_order_cmp): New.
14105 (stmts_from_loop): Use topological order.
14106 (pass_loop_distribution::execute): Compute and release topological
14107 order for basic blocks.
14109 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14111 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
14114 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14116 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
14117 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
14118 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
14119 * internal-fn.def (LOOP_DIST_ALIAS): New.
14120 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
14121 (fold_loop_internal_call): ... this.
14122 (vect_loop_dist_alias_call): New function.
14123 (set_uid_loop_bbs): Call fold_loop_internal_call.
14124 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
14127 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14130 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
14131 Require dead FLAGS_REG at the beginning of a peephole.
14133 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14136 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
14137 arguments in the call to __builtin_ia32_sbb_u32.
14138 (_subborrow_u64): Swap _X and _Y arguments in the call to
14139 __builtin_ia32_sbb_u64.
14141 2017-07-04 Jakub Jelinek <jakub@redhat.com>
14144 * tree-vrp.c (compare_assert_loc): Turn into a function template
14145 with stable template parameter. Only test if a->e is NULL,
14146 !a->e == !b->e has been verified already. Use e == NULL or
14147 e != NULL instead of e or ! e tests. If stable is true, don't use
14148 iterative_hash_expr, on the other side allow a or b or both NULL
14149 and sort the NULLs last.
14150 (process_assert_insertions): Sort using compare_assert_loc<false>
14151 instead of compare_assert_loc, later sort using
14152 compare_assert_loc<true> before calling process_assert_insertions_for
14153 in a loop. Use break instead of continue once seen NULL pointer.
14155 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14157 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14158 Cortex-R7 and Cortex-R8 processors.
14160 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14162 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
14163 uninitialized while src is not.
14165 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
14167 * common/config/arm/arm-common.c: Adjust include path for
14169 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
14170 (arm-cpu.h): Create in build directory. Adjust dependency rules.
14171 (arm-cpu-data.h): Likewise.
14172 (arm-cpu-cdata.h): Likewise.
14173 * config/arm/arm-cpu.h: Delete.
14174 * config/arm/arm-cpu-cdata.h: Delete.
14175 * config/arm/arm-cpu-data.h: Delete.
14177 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
14179 * config/arm/arm-cpus.in (cortex-a55): New.
14180 (cortex-a75): Likewise.
14181 (cortex-a75.cortex-a55): Likewise.
14182 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
14184 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
14185 * config/arm/arm-cpu-cdata.h: Regenerate.
14186 * config/arm/arm-cpu-data.h: Regenerate.
14187 * config/arm/arm-cpu.h: Regenerate.
14188 * config/arm/arm-tables.opt: Regenerate.
14189 * config/arm/arm-tune.md: Regenerate.
14191 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14193 * haifa-sched.c (sched_create_recovery_edges): Update profile.
14195 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14197 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
14200 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
14202 PR tree-optimization/81292
14203 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
14204 full_string_p, also call adjust_related_strinfos if the adjustment
14205 is simple, otherwise invalidate related strinfos.
14207 2017-07-04 Martin Liska <mliska@suse.cz>
14210 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
14211 newly created variable as DECL_IGNORED_P.
14213 2017-07-04 Martin Liska <mliska@suse.cz>
14216 * ipa-inline.c (inline_small_functions):
14217 Use xstrdup_for_dump.
14219 2017-07-04 Tom de Vries <tom@codesourcery.com>
14221 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
14223 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14226 * config/darwin.c (darwin_function_switched_text_sections):
14229 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
14231 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
14233 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14235 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
14237 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14239 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
14240 min_profitable_iters, and th as inclusive lower bounds.
14241 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
14242 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
14243 for min_profitable_iters and min_profitable_estimate.
14244 (vect_transform_loop): Treat th as an inclusive lower bound.
14245 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
14247 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14250 * config/darwin.c (darwin_function_switched_text_sections):
14251 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
14252 in two pieces, and suppress the use of buf.
14254 2017-07-03 Nathan Sidwell <nathan@acm.org>
14256 * hash-table.h (hash_table_mod1): Fix indentation.
14258 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14260 PR middle-end/81290
14261 * predict.c (force_edge_cold): Be more careful about propagation
14263 * profile-count.h (profile_probability::guessed,
14264 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
14266 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
14268 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
14270 * doc/invoke.texi (rcpc architecture extension): Document it.
14272 2017-07-03 Richard Biener <rguenther@suse.de>
14274 PR tree-optimization/60510
14275 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
14276 the scalar reduction PHI and use it.
14277 (vectorizable_reduction): Properly guard the single_defuse_cycle
14278 path for non-SLP reduction chains where we cannot use it.
14279 Rework reduc_def/index and vector type deduction. Rework
14280 vector operand gathering during reduction op code-gen.
14281 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
14282 chains dissolve the chain and leave it to non-SLP reduction
14285 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14287 * tree-data-ref.h (dr_alignment): Declare.
14288 * tree-data-ref.c (dr_alignment): New function.
14289 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
14290 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
14292 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
14294 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14296 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
14297 and base_misalignment fields.
14298 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
14299 * tree-data-ref.c: Include builtins.h.
14300 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
14301 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
14302 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
14303 * tree-vect-data-refs.c: Include tree-cfg.h.
14304 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
14305 fields instead of calculating an alignment here.
14306 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
14307 innermost_loop_behavior fields.
14309 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14311 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
14313 (DR_STEP_ALIGNMENT): New macro.
14314 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
14315 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
14316 (create_data_ref): Print it.
14317 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
14318 to tell whether the step preserves vector (mis)alignment.
14319 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14320 Move the check for an integer step and generalise to all INTEGER_CST.
14321 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
14322 Print the outer step alignment.
14324 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14326 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
14327 with offset_alignment.
14328 (DR_ALIGNED_TO): Delete.
14329 (DR_OFFSET_ALIGNMENT): New macro.
14330 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
14331 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
14332 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
14333 (create_data_ref): Likewise.
14334 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14335 (vect_analyze_data_refs): Likewise.
14336 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
14337 creating dummy innermost behavior.
14339 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14341 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
14342 with a "innermost_loop_behavior *" and refeence tree.
14343 * tree-data-ref.c (dr_analyze_innermost): Likewise.
14344 (create_data_ref): Update call accordingly.
14345 * tree-predcom.c (find_looparound_phi): Likewise.
14347 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14349 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
14350 fields with dr_wrt_vec_loop.
14351 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
14352 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
14353 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
14354 (vect_dr_behavior): New function.
14355 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14356 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
14357 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
14358 track whether the step preserves the misalignment.
14359 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14360 Use vect_dr_behavior.
14361 (vect_setup_realignment): Update call accordingly.
14362 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
14363 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
14364 call to vect_create_addr_base_for_vector_ref.
14365 (vect_create_cond_for_align_checks): Likewise.
14366 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
14367 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
14368 (vect_recog_mask_conversion_pattern): Likewise.
14369 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
14370 (new_stmt_vec_info): Remove redundant zeroing.
14372 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14374 * common/config/arm/arm-common.c (arm_be8_option): New function.
14375 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
14376 (ISA_ARMv6): Add isa_bit_be8.
14377 * config/arm/arm.h (arm_be8_option): Add prototype.
14378 (BE8_SPEC_FUNCTION): New define.
14379 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
14380 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
14381 (mlittle-endian): Similarly.
14382 (mbe8, mbe32): New options.
14383 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
14384 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
14386 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14388 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
14390 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14392 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
14393 (cleanup_tree_cfg_bb): Use it.
14394 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
14396 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
14398 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14401 * loop-doloop.c (add_test): Update profile.
14403 2017-07-03 Martin Liska <mliska@suse.cz>
14406 * sanopt.c (rewrite_usage_of_param): New function.
14407 (sanitize_rewrite_addressable_params): Likewise.
14408 (pass_sanopt::execute): Call rewrite_usage_of_param.
14410 2017-07-03 Richard Biener <rguenther@suse.de>
14412 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
14413 back to using VIEW_CONVERT_EXPR.
14415 2017-07-03 Martin Liska <mliska@suse.cz>
14418 * doc/extend.texi: Document when a resolver function is
14419 generated for target_clones.
14421 2017-07-03 Martin Liska <mliska@suse.cz>
14423 * asan.c (asan_emit_stack_protection): Unpoison just red zones
14424 and shadow memory of auto variables which are subject of
14425 use-after-scope sanitization.
14426 (asan_expand_mark_ifn): Add do set only when is_poison.
14428 2016-07-03 Richard Biener <rguenther@suse.de>
14430 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
14432 (vect_force_simple_reduction): Record reduction def -> phi mapping.
14433 (vectorizable_reduction): Perform reduction PHI creation when
14434 visiting a reduction PHI and adjust and simplify code generation
14435 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
14436 (vect_transform_loop): Visit reduction PHIs.
14437 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
14438 defs into the SLP tree.
14439 (vect_build_slp_tree): Reduction defs terminate the recursion.
14440 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
14442 (vect_get_vec_defs_for_stmt_copy): Export.
14443 (vect_get_vec_defs): Likewise.
14444 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
14446 (vect_get_vec_defs_for_stmt_copy): Declare.
14447 (vect_get_vec_defs): Likewise.
14449 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14451 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
14452 parameter with a "loop" parameter and use it instead of the
14453 loop containing DR_STMT. Don't check simple_iv when doing
14454 BB analysis. Describe the two analysis modes in the comment.
14456 2017-07-03 Tom de Vries <tom@codesourcery.com>
14458 PR tree-optimization/69468
14459 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
14460 (find_same_succ_bb): Handle ignore_edge_flags.
14462 2017-07-03 Tom de Vries <tom@codesourcery.com>
14464 PR tree-optimization/81192
14465 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
14467 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
14469 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
14471 2017-07-03 Tom de Vries <tom@codesourcery.com>
14473 PR tree-optimization/81192
14474 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
14475 BB_SAME_SUCC (bb) == NULL.
14477 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14479 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
14482 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14484 * dumpfile.c: Include profile-count.h
14485 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
14487 (insert_cond_bb): Update profile.
14488 * tree-cfg.h (insert_cond_bb): Update prototype.
14489 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
14490 * tree-dump.c: Do not include tree-cfg.
14492 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14494 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
14496 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14498 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
14501 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14503 * tree-complex.c (expand_complex_div_wide): update profile.
14505 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14506 Alan Hayward <alan.hayward@arm.com>
14507 David Sherwood <david.sherwood@arm.com>
14509 * Makefile.in (MACHMODE_H): Remove insn-modes.h
14510 (CORETYPES_H): New define.
14511 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
14512 (insn-modes-inline.h, s-modes-inline-h): New rules.
14513 (generated_files): Add insn-modes-inline.h.
14514 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
14515 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
14516 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
14517 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
14518 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
14519 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
14520 (build/gencodes.o, build/genconditions.o): Likewise.
14521 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
14522 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
14523 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
14524 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
14525 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
14526 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
14527 * coretypes.h: Include everything up to real.h for generators.
14528 Include insn-modes.h first. Include wide-int-print.h after
14529 wide-int.h. Include insn-modes-inline.h and then machmode.h.
14530 * machmode.h: Don't include insn-modes.h here.
14531 * function-tests.c: Remove includes of signop.h, machmode.h,
14532 double-int.h and wide-int.h.
14534 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
14536 * optc-save-gen.awk: Likewise.
14537 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
14538 * godump.c: Remove include of wide-int-print.h.
14539 * pretty-print.h: Likewise.
14540 * wide-int-print.cc: Likewise.
14541 * wide-int.cc: Likewise.
14542 * hash-map-tests.c: Remove include of signop.h.
14543 * hash-set-tests.c: Likewise.
14544 * rtl-tests.c: Likewise.
14545 * mkconfig.sh: Remove include of machmode.h.
14546 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
14548 (emit_insn_modes_inline_h): ...this new function. Emit the code
14549 into an insn-modes-inline.h header file, adding appropriate
14550 include guards and end comments.
14551 (emit_insn_modes_c_header): Remove include of machmode.h.
14552 (emit_min_insn_modes_c_header): Include coretypes.h rather than
14554 (main): Handle -i flag and call emit_insn_modes_inline_h when
14557 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14559 * tree-ssa-strlen.c (strinfo): Rename the length field to
14560 nonzero_chars. Add a full_string_p field.
14561 (compare_nonzero_chars, zero_length_string_p): New functions.
14562 (get_addr_stridx): Add an offset_out parameter.
14563 Use compare_nonzero_chars.
14564 (get_stridx): Update accordingly. Use compare_nonzero_chars.
14565 (new_strinfo): Update after above changes to strinfo.
14566 (set_endptr_and_length): Set full_string_p.
14567 (get_string_length): Update after above changes to strinfo.
14568 (unshare_strinfo): Update call to new_strinfo.
14569 (maybe_invalidate): Likewise.
14570 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
14571 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
14572 as a uhwi instead of an shwi. Update after above changes to
14573 strinfo and new_strinfo.
14574 (zero_length_string): Assert that chainsi contains full strings.
14575 Use zero_length_string_p. Update call to new_strinfo.
14576 (adjust_related_strinfos): Update after above changes to strinfo.
14577 Copy full_string_p from origsi.
14578 (adjust_last_stmt): Use zero_length_string_p.
14579 (handle_builtin_strlen): Update after above changes to strinfo and
14580 new_strinfo. Install the lhs as the string length if the previous
14581 entry didn't describe a full string.
14582 (handle_builtin_strchr): Update after above changes to strinfo
14584 (handle_builtin_strcpy): Likewise.
14585 (handle_builtin_strcat): Likewise.
14586 (handle_builtin_malloc): Likewise.
14587 (handle_pointer_plus): Likewise.
14588 (handle_builtin_memcpy): Likewise. Track nonzero characters
14589 that aren't necessarily followed by a nul terminator.
14590 (handle_char_store): Likewise.
14592 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14594 PR tree-optimization/80769
14595 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
14596 for malloc and calloc. Document the new invariant that all related
14597 strinfos have delayed lengths or none do.
14598 (verify_related_strinfos): Move earlier in file.
14599 (set_endptr_and_length): New function, split out from...
14600 (get_string_length): ...here. Also set the lengths of related
14602 (zero_length_string): Assert that chainsi has known (rather than
14604 (adjust_related_strinfos): Likewise.
14606 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14608 PR tree-optimization/81136
14609 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
14610 assert that two references with the same misalignment have the same
14611 compile-time misalignment if those compile-time misalignments
14614 2017-07-01 Andi Kleen <ak@linux.intel.com>
14616 * print-tree.c (print_node): Print all attributes.
14618 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14620 * cfg.c (scale_bbs_frequencies): New function.
14621 * cfg.h (scale_bbs_frequencies): Declare it.
14622 * cfgloopanal.c (single_likely_exit): Cleanup.
14623 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
14625 (scale_loop_profile): Likewise.
14626 (loop_version): Likewise.
14627 (create_empty_loop_on_edge): Update.
14628 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
14629 scale_loop_frequencies, scale_loop_profile, loopify,
14630 loop_version): Update prototypes.
14631 * modulo-sched.c (sms_schedule): Update.
14632 * predict.c (unlikely_executed_edge_p): Also check probability.
14633 (probably_never_executed_edge_p): Fix typo.
14634 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14635 * tree-parloops.c (gen_parallel_loop): Update.
14636 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
14637 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14638 * tree-ssa-loop-split.c (split_loop): Update.
14639 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14640 * tree-vect-loop-manip.c (vect_do_peeling): Update.
14641 (vect_loop_versioning): Update.
14642 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14644 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14646 * trans-mem.c (split_bb_make_tm_edge): Update profile.
14648 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14650 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
14651 to keep profile consistent.
14653 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14655 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
14656 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
14657 * profile-count.h (max_safe_multiplier): Make unsigned.
14658 (profile_count::guessed_zero): New.
14660 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14662 * bb-reorder.c (fix_up_crossing_landing_pad,
14663 fix_crossing_conditional_branches): Use make_single_succ_edge
14664 to keep profile consistent.
14666 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14668 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
14671 2017-07-01 Jakub Jelinek <jakub@redhat.com>
14674 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
14675 the right scopes, make sure cond_jump isn't preserved between multiple
14676 iterations. Search for fallthru edge whenever there are 3+ edges and
14677 use find_fallthru_edge for it.
14679 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14681 Patch by Alexander Monakov <amonakov@ispras.ru>
14682 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
14683 probabilities consistently.
14685 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14687 * pa.c (pa_expand_compare_and_swap_loop): Update call of
14688 emit_cmp_and_jump_insns.
14690 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14693 * tree-inline.c (expand_call_inline): Combine profile statuses.
14695 2017-06-30 Andrew Pinski <apinski@cavium.com>
14697 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
14698 fold_stmt returned true.
14700 2017-06-30 Nathan Sidwell <nathan@acm.org>
14702 * ggc.h (empty_string): Delete.
14703 * cfgexpand.c (expand_asm_stmt): Use plain "".
14704 * optabs.c (expand_asm_memory_barrier): Likewise.
14705 * stringpool.c (empty_string): Delete.
14706 (digit_vector, digit_string): Delete.
14707 (ggc_alloc_string): Use plain "", don't optimize single digit
14708 strings. Use ggc_alloc_atomic.
14710 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
14712 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
14713 comparison set and one other set, use the cost of the non-comparison
14716 2017-06-30 Nathan Sidwell <nathan@acm.org>
14718 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
14721 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
14723 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
14724 loops. Remove now unneeded calls to gimple_switch_set_label() that
14725 just set removed labels to NULL_TREE.
14727 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
14729 * tree-ssanames.c (set_range_info_raw): Abstract from ...
14730 (set_range_info): ...here. Only call set_range_info_raw if domain
14732 (set_nonzero_bits): Call set_range_info_raw.
14733 * tree-ssanames.h (set_range_info_raw): New.
14735 2017-06-30 Jakub Jelinek <jakub@redhat.com>
14738 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
14739 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
14740 of nonimmediate_operand and <store_mask_constraint> instead of m
14741 for the input operand. For V8FI iterator, always split if input
14742 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
14743 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
14744 <store_mask_predicate> instead of register_operand and
14745 <store_mask_constraint> instead of v for the input operand. Make
14746 sure both operands aren't MEMs for if not <mask_applied>.
14748 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
14750 * lto-wrapper.c (copy_file) Close both file descriptors before
14753 2017-06-30 Martin Liska <mliska@suse.cz>
14756 * multiple_target.c (create_dispatcher_calls): Make ifunc
14757 also for function that don't have calls or are not referenced.
14759 2017-06-30 Richard Biener <rguenther@suse.de>
14761 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
14762 analyze the first scalar stmt. Move vector type computation
14763 for the BB case here from ...
14764 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
14765 live operation processing in the SLP case properly.
14767 2017-06-30 Richard Biener <rguenther@suse.de>
14769 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
14771 2017-06-30 Martin Liska <mliska@suse.cz>
14774 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
14775 before BUILT_IN_UNWIND_RESUME when ASAN is used.
14777 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
14779 * doc/invoke.texi (AArch64): Add missing options and remove redundant
14782 2017-06-30 Richard Biener <rguenther@suse.de>
14784 PR tree-optimization/81249
14785 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
14786 condition reduction result to original scalar type.
14788 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14790 * profile-count.h (enum profile_quality): Fix typos and whitespace
14793 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14795 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
14796 type for branch probabilities.
14798 2017-06-29 Julian Brown <julian@codesourcery.com>
14799 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14801 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
14802 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
14803 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
14804 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
14806 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14808 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
14809 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
14810 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
14811 CC usage from generic code to here.
14812 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
14813 CC usage into the target macros.
14815 2017-06-29 Maya Rashish <coypu@sdf.org>
14817 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
14820 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14822 * arm/arm-builtins.c: Include profile-count.h
14823 * except.c (sjlj_emit_function_enter): Use
14824 profile_probability::unlikely.
14826 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14828 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
14829 and tocrel_offset be pointer args rather than implicitly using
14831 (legitimate_constant_pool_address_p, rs6000_emit_move,
14832 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
14833 tocrel_offset and use in toc_relative_expr_p call.
14834 (print_operand, print_operand_address): Use static tocrel_base_oac
14835 and tocrel_offset_oac.
14836 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
14839 2017-06-29 Maya Rashish <coypu@sdf.org>
14841 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
14843 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
14845 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
14846 objects, take into account only the alignment of 'op0' and 'mode1' if
14849 2017-06-29 Steve Ellcey <sellcey@cavium.com>
14851 * ccmp.c (ccmp_tree_comparison_p): New function.
14852 (ccmp_candidate_p): Update to use above function.
14853 (get_compare_parts): New function.
14854 (expand_ccmp_next): Update to use new functions.
14855 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
14857 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
14858 take mode as argument.
14859 * ccmp.h (expand_ccmp_expr): Add mode as argument.
14860 * expr.c (expand_expr_real_1): Pass mode as argument.
14862 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
14864 * combine.c (combine_instructions): Print insns to dump_file, together
14867 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14869 * asan.c (asan_emit_stack_protection): Update.
14870 (create_cond_insert_point): Update.
14871 * auto-profile.c (afdo_propagate_circuit): Update.
14872 * basic-block.h (struct edge_def): Turn probability to
14873 profile_probability.
14874 (EDGE_FREQUENCY): Update.
14875 * bb-reorder.c (find_traces_1_round): Update.
14876 (better_edge_p): Update.
14877 (sanitize_hot_paths): Update.
14878 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
14879 (make_single_succ_edge): Update.
14880 (check_bb_profile): Update.
14881 (dump_edge_info): Update.
14882 (update_bb_profile_for_threading): Update.
14883 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
14884 probabilitycount to 0.
14885 * cfgbuild.c (compute_outgoing_frequencies): Update.
14886 * cfgcleanup.c (try_forward_edges): Update.
14887 (outgoing_edges_match): Update.
14888 (try_crossjump_to_edge): Update.
14889 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
14890 (expand_gimple_tailcall): Update.
14891 (construct_init_block): Use make_single_succ_edge.
14892 (construct_exit_block): Use make_single_succ_edge.
14893 * cfghooks.c (verify_flow_info): Update.
14894 (redirect_edge_succ_nodup): Update.
14895 (split_edge): Update.
14896 (account_profile_record): Update.
14897 * cfgloopanal.c (single_likely_exit): Update.
14898 * cfgloopmanip.c (scale_loop_profile): Update.
14899 (set_zero_probability): Remove.
14900 (duplicate_loop_to_header_edge): Update.
14901 * cfgloopmanip.h (loop_version): Update prototype.
14902 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
14903 (force_nonfallthru_and_redirect): Update.
14904 (update_br_prob_note): Update.
14905 (rtl_verify_edges): Update.
14906 (purge_dead_edges): Update.
14907 (rtl_lv_add_condition_to_bb): Update.
14908 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
14909 * cgraphunit.c (init_lowered_empty_function): Update.
14910 (cgraph_node::expand_thunk): Update.
14911 * cilk-common.c: Include profile-count.h
14912 * dojump.c (inv): Remove.
14913 (jumpifnot): Update.
14914 (jumpifnot_1): Update.
14915 (do_jump_1): Update.
14917 (do_jump_by_parts_greater_rtx): Update.
14918 (do_compare_rtx_and_jump): Update.
14919 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
14920 do_jump_1. do_compare_rtx_and_jump): Update prototype.
14921 * dwarf2cfi.c: Include profile-count.h
14922 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
14923 (sjlj_emit_dispatch_table): Likewise.
14924 * explow.c: Include profile-count.h
14925 * expmed.c (emit_store_flag_force): Update.
14926 (do_cmp_and_jump): Update.
14927 * expr.c (compare_by_pieces_d::generate): Update.
14928 (compare_by_pieces_d::finish_mode): Update.
14929 (emit_block_move_via_loop): Update.
14930 (store_expr_with_bounds): Update.
14931 (store_constructor): Update.
14932 (expand_expr_real_2): Update.
14933 (expand_expr_real_1): Update.
14934 * expr.h (try_casesi, try_tablejump): Update prototypes.
14935 * gimple-pretty-print.c (dump_probability): Update.
14936 (dump_profile): New.
14937 (dump_gimple_label): Update.
14938 (dump_gimple_bb_header): Update.
14939 * graph.c (draw_cfg_node_succ_edges): Update.
14940 * hsa-gen.c (convert_switch_statements): Update.
14941 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
14942 (find_if_case_1): Update.
14943 (find_if_case_2): Update.
14944 * internal-fn.c (expand_arith_overflow_result_store): Update.
14945 (expand_addsub_overflow): Update.
14946 (expand_neg_overflow): Update.
14947 (expand_mul_overflow): Update.
14948 (expand_vector_ubsan_overflow): Update.
14949 * ipa-cp.c (good_cloning_opportunity_p): Update.
14950 * ipa-split.c (split_function): Use make_single_succ_edge.
14951 * ipa-utils.c (ipa_merge_profiles): Update.
14952 * loop-doloop.c (add_test): Update.
14953 (doloop_modify): Update.
14954 * loop-unroll.c (compare_and_jump_seq): Update.
14955 (unroll_loop_runtime_iterations): Update.
14956 * lra-constraints.c (lra_inheritance): Update.
14957 * lto-streamer-in.c (input_cfg): Update.
14958 * lto-streamer-out.c (output_cfg): Update.
14959 * mcf.c (adjust_cfg_counts): Update.
14960 * modulo-sched.c (sms_schedule): Update.
14961 * omp-expand.c (expand_omp_for_init_counts): Update.
14962 (extract_omp_for_update_vars): Update.
14963 (expand_omp_ordered_sink): Update.
14964 (expand_omp_for_ordered_loops): Update.
14965 (expand_omp_for_generic): Update.
14966 (expand_omp_for_static_nochunk): Update.
14967 (expand_omp_for_static_chunk): Update.
14968 (expand_cilk_for): Update.
14969 (expand_omp_simd): Update.
14970 (expand_omp_taskloop_for_outer): Update.
14971 (expand_omp_taskloop_for_inner): Update.
14972 * omp-simd-clone.c (simd_clone_adjust): Update.
14973 * optabs.c (expand_doubleword_shift): Update.
14974 (expand_abs): Update.
14975 (emit_cmp_and_jump_insn_1): Update.
14976 (expand_compare_and_swap_loop): Update.
14977 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
14978 * predict.c (predictable_edge_p): Update.
14979 (edge_probability_reliable_p): Update.
14980 (set_even_probabilities): Update.
14981 (combine_predictions_for_insn): Update.
14982 (combine_predictions_for_bb): Update.
14983 (propagate_freq): Update.
14984 (estimate_bb_frequencies): Update.
14985 (force_edge_cold): Update.
14986 * profile-count.c (profile_count::dump): Add missing space into dump.
14987 (profile_count::debug): Add newline.
14988 (profile_count::differs_from_p): Explicitly convert to unsigned.
14989 (profile_count::stream_in): Update.
14990 (profile_probability::dump): New member function.
14991 (profile_probability::debug): New member function.
14992 (profile_probability::differs_from_p): New member function.
14993 (profile_probability::differs_lot_from_p): New member function.
14994 (profile_probability::stream_in): New member function.
14995 (profile_probability::stream_out): New member function.
14996 * profile-count.h (profile_count_quality): Rename to ...
14997 (profile_quality): ... this one.
14998 (profile_probability): New.
14999 (profile_count): Update.
15000 * profile.c (compute_branch_probabilities): Update.
15001 * recog.c (peep2_attempt): Update.
15002 * sched-ebb.c (schedule_ebbs): Update.
15003 * sched-rgn.c (find_single_block_region): Update.
15004 (compute_dom_prob_ps): Update.
15005 (schedule_region): Update.
15006 * sel-sched-ir.c (compute_succs_info): Update.
15007 * stmt.c (struct case_node): Update.
15008 (do_jump_if_equal): Update.
15009 (get_outgoing_edge_probs): Update.
15010 (conditional_probability): Update.
15011 (emit_case_dispatch_table): Update.
15012 (expand_case): Update.
15013 (expand_sjlj_dispatch_table): Update.
15014 (emit_case_nodes): Update.
15015 * targhooks.c: Update.
15016 * tracer.c (better_p): Update.
15017 (find_best_successor): Update.
15018 * trans-mem.c (expand_transaction): Update.
15019 * tree-call-cdce.c: Update.
15020 * tree-cfg.c (gimple_split_edge): Upate.
15021 (move_sese_region_to_fn): Upate.
15022 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
15023 * tree-eh.c (lower_resx): Upate.
15024 (cleanup_empty_eh_move_lp): Upate.
15025 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15026 * tree-inline.c (copy_edges_for_bb): Update.
15027 (copy_cfg_body): Update.
15028 * tree-parloops.c (gen_parallel_loop): Update.
15029 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
15030 (gimple_gen_time_profiler): Update.
15031 * tree-ssa-dce.c (remove_dead_stmt): Update.
15032 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15033 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
15034 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
15035 (unloop_loops): Update.
15036 (try_peel_loop): Update.
15037 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15038 * tree-ssa-loop-split.c (connect_loops): Update.
15039 (split_loop): Update.
15040 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15041 (hoist_guard): Update.
15042 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
15043 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
15044 (value_replacement): Update.
15045 * tree-ssa-reassoc.c (branch_fixup): Update.
15046 * tree-ssa-tail-merge.c (replace_block_by): Update.
15047 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
15048 (create_edge_and_update_destination_phis): Update.
15049 (compute_path_counts): Update.
15050 (recompute_probabilities): Update.
15051 (update_joiner_offpath_counts): Update.
15052 (freqs_to_counts_path): Update.
15053 (duplicate_thread_path): Update.
15054 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
15055 (struct switch_conv_info): Update.
15056 (gen_inbound_check): Update.
15057 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15058 (vect_do_peeling): Update.
15059 (vect_loop_versioning): Update.
15060 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15061 (optimize_mask_stores): Update.
15062 * ubsan.c (ubsan_expand_null_ifn): Update.
15063 * value-prof.c (gimple_divmod_fixed_value): Update.
15064 (gimple_divmod_fixed_value_transform): Update.
15065 (gimple_mod_pow2): Update.
15066 (gimple_mod_pow2_value_transform): Update.
15067 (gimple_mod_subtract): Update.
15068 (gimple_mod_subtract_transform): Update.
15069 (gimple_ic): Update.
15070 (gimple_stringop_fixed_value): Update.
15071 (gimple_stringops_transform): Update.
15072 * value-prof.h: Update.
15074 2017-06-29 Carl Love <cel@us.ibm.com>
15076 * config/rs6000/rs6000-c.c: Add support for built-in functions
15077 vector signed int vec_signed (vector float);
15078 vector signed long long vec_signed (vector double);
15079 vector signed int vec_signed2 (vector double, vector double);
15080 vector signed int vec_signede (vector double);
15081 vector signed int vec_signedo (vector double);
15082 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
15083 instruction generator.
15084 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15085 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
15086 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
15088 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
15089 vunsignede_v2df): Add define_expands.
15090 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
15091 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
15092 VEC_UNSIGNEDO): Add definitions.
15093 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15094 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
15095 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
15096 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
15097 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
15098 * config/rs6000/altivec.h (vec_signed, vec_signed2,
15099 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
15100 vec_unsignede, vec_unsignedo): Add builtin defines.
15101 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
15103 * doc/extend.texi: Update the built-in documentation file for the
15104 new built-in functions.
15106 2017-06-29 Richard Biener <rguenther@suse.de>
15108 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
15109 reduction chains to LOOP_VINFO_REDUCTIONS.
15110 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
15111 SLP reductions after processing reduction chains.
15113 2017-06-29 Nathan Sidwell <nathan@acm.org>
15115 * builtins.c (fold_builtin_FUNCTION): Use
15116 lang_hooks.decl_printable_name.
15118 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
15120 PR middle-end/81194
15121 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
15122 with only one label.
15123 * stmt.c (expand_case): Assert NCASES is greater than one.
15125 2017-06-29 Richard Biener <rguenther@suse.de>
15127 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
15129 (group_case_labels): Likewise.
15130 (find_taken_edge): Push sanity checking on val to workers...
15131 (find_taken_edge_cond_expr): ... here
15132 (find_taken_edge_switch_expr): ... and here, handle cases
15133 with just a default label.
15134 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
15135 (group_case_labels): Likewise.
15136 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
15137 group_case_labels does anything cleanup the CFG again.
15139 2017-06-29 Bin Cheng <bin.cheng@arm.com>
15141 PR tree-optimization/81196
15142 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
15143 exit condition comparing two IVs.
15145 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
15147 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
15148 profile to the dummy entry at the end of the list of architectures.
15149 * config/arm/arm-cpu-cdata.h: Regenerated.
15151 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15152 Michael Collison <michael.collison@arm.com>
15155 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
15157 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
15158 (*aarch64_reg_<mode>3_minus_mask): New pattern.
15159 (*aarch64_<optab>_reg_di3_mask2): New pattern.
15160 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
15161 of shift when the shift amount is masked with constant equal to
15162 the size of the mode.
15163 * config/aarch64/predicates.md (subreg_lowpart_operator): New
15166 2017-06-29 Martin Liska <mliska@suse.cz>
15168 * config/i386/i386.opt: Change range from [1,5] to [0,5].
15170 2017-06-29 Yury Gribov <tetra2005@gmail.com>
15173 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
15176 (edge_growth_cache_entry::edge_growth_cache_entry): New
15178 (reset_edge_growth_cache): Update to use constructor.
15180 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15182 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
15183 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15184 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15186 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
15188 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
15189 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
15191 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
15193 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
15194 (*-linux-uclibc*): Add t-uclibc tmake_file.
15195 * config/t-musl: New.
15196 * config/t-uclibc: New.
15198 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
15200 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
15202 (gen_comm_data): Emit architectural setting of arch_prof.
15203 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
15205 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
15206 (armv8-m.base, armv8-m.main): Likewise.
15207 * arm-protos.h (arm_build_target): Add profile field.
15208 (arch_option): Likewise.
15209 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
15211 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
15212 arm_active_target.profile.
15214 2017-06-28 Richard Biener <rguenther@suse.de>
15216 PR middle-end/81227
15217 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
15218 TYPE_OVERFLOW_WRAPS.
15219 * match.pd (negate_expr_p): Likewise.
15220 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
15221 fold_build2, not fold_binary.
15223 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15225 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
15226 Convert memory address to Pmode.
15227 (aarch64_print_operand): Assert MEM operands are always Pmode.
15229 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15232 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
15233 Remove redundant if.
15234 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
15235 * config/arm/aarch-common-protos.h
15236 (aarch_forward_to_shift_is_not_shifted_re): Remove.
15237 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
15239 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
15242 * multiple_target.c (create_dispatcher_calls): Set the default
15243 clone to be static, not public.
15245 2017-06-28 Richard Biener <rguenther@suse.de>
15247 * tree-vect-loop.c (vectorizable_reduction): Move special
15248 cond reduction IV var creation ...
15249 (vect_create_epilog_for_reduction): ... here. Remove induction_index
15250 parameter. Use STMT_VINFO_VECTYPE.
15251 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
15254 2017-06-28 Martin Liska <mliska@suse.cz>
15257 * ipa-visibility.c (non_local_p): Handle visibility.
15259 2017-06-28 Martin Liska <mliska@suse.cz>
15262 * common.opt: Add IntegerRange to various options.
15263 * opt-functions.awk (integer_range_info): New function.
15264 * optc-gen.awk: Add integer_range_info to cl_options struct.
15265 * opts-common.c (decode_cmdline_option): Handle
15266 CL_ERR_INT_RANGE_ARG.
15267 (cmdline_handle_error): Likewise.
15268 * opts.c (print_filtered_help): Show valid interval in
15269 when --help is provided.
15270 * opts.h (struct cl_option): Add range_min and range_max fields.
15271 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
15273 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
15275 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
15276 (x * C EQ/NE y * C): New transformation.
15278 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
15280 * genmultilib (combination_space): Accept '+' in option names.
15282 2017-06-28 Martin Liska <mliska@suse.cz>
15285 * asan.c (instrument_derefs): Bail out inner references
15286 that are hard register variables.
15288 2017-06-28 Jakub Jelinek <jakub@redhat.com>
15291 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
15292 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
15294 2017-06-28 Richard Biener <rguenther@suse.de>
15296 * tree-vectorizer.h (vect_get_vec_defs): Remove.
15297 (vect_get_slp_defs): Adjust.
15298 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
15300 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
15302 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
15303 get_initial_defs_for_reduction instead of vect_get_vec_defs.
15304 (vectorizable_reduction): Adjust.
15305 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
15307 (vect_get_slp_defs): Likewise.
15308 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
15309 (vectorizable_bswap): Adjust.
15310 (vectorizable_call): Likewise.
15311 (vectorizable_conversion): Likewise.
15312 (vectorizable_assignment): Likewise.
15313 (vectorizable_shift): Likewise.
15314 (vectorizable_operation): Likewise.
15315 (vectorizable_store): Likewise.
15316 (vectorizable_condition): Likewise.
15317 (vectorizable_comparison): Likewise.
15319 2017-06-28 Michael Collison <michael.collison@arm.com>
15322 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
15324 (arm_gen_movmemqi): Removed unused variable 'i'.
15325 Convert 'for' loop into 'while' loop.
15326 (arm_expand_prologue): Remove last unnecessary set of insn.
15327 (thumb_pop): Remove unused variable 'pushed_words'.
15328 (thumb_exit): Remove last unnecessary set of regs_to_pop.
15330 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15332 * config/s390/predicates.md: Use s390_rel_address_ok_p.
15333 * config/s390/s390-protos.h: Add prototype of
15334 s390_rel_address_ok_p.
15335 * config/s390/s390.c (s390_got_symbol): New function.
15336 (s390_rel_address_ok_p): New function.
15337 (legitimize_pic_address): Use s390_rel_address_ok_p.
15338 (s390_load_got): Use s390_got_symbol.
15339 (s390_option_override): Issue error if
15340 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
15341 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
15343 * config/s390/s390.opt: New option mpic-data-is-text-relative.
15345 2017-06-27 Andrew Pinski <apinski@cavium.com>
15347 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
15348 (X * copysign (1.0, X)): New pattern.
15349 (X * copysign (1.0, -X)): New pattern.
15350 (copysign (-1.0, CST)): New pattern.
15352 2017-06-27 Joseph Myers <joseph@codesourcery.com>
15354 * genmultilib (combination_space): Remove variable.
15355 Validate reuse rules against regular expression for any sequence
15356 of multilib options in any order.
15358 2017-06-27 Michael Collison <michael.collison@arm.com>
15360 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
15361 call aarch64_split_simd_combine.
15362 * (aarch64_combine_internal<mode>): Delete pattern.
15363 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
15364 Allow register and subreg operands.
15366 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15368 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
15369 specific need, just fallback on defaults.
15370 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
15372 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15373 Olivier Hainque <hainque@adacore.com>
15375 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
15377 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
15378 targets. Pick a default if no particular attempt applied.
15379 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
15382 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15384 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
15385 (x86_64-wrs-vxworks7): Likewise.
15387 2017-06-27 Marek Polacek <polacek@redhat.com>
15390 * ubsan.c (instrument_null): Check get_base_address's result for null.
15392 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15394 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
15396 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15398 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
15399 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
15400 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
15401 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
15402 New function types.
15403 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
15404 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
15405 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
15406 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
15407 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
15408 BUILT_IN_FEUPDATEENV): New builtins.
15409 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
15410 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
15411 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
15412 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
15414 (builtin_structptr_types): Adjust size.
15415 * tree.c (builtin_structptr_types): Add four entries.
15417 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15418 Olivier Hainque <hainque@adacore.com>
15420 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
15421 (TLS_SYM): New local macro, forcing reference to __tls__ on
15422 link command lines for VxWorks 7 RTPs, triggering initialization
15424 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
15425 OS features TLS support, true for RTPs on VxWorks 7.
15426 * config/vxworks.c (vxworks_override_options): Setup emutls
15429 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15431 * predict.c (test_prediction_value_range): Use -1U instead of -1
15432 to avoid narrowing conversion warning.
15433 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
15434 to avoid narrowing conversion warning.
15435 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
15437 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
15439 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15441 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
15442 64bit configurations.
15443 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
15444 (SIZE_TYPE): Likewise.
15445 * config/vxworks.c (vxworks_emutls_var_fields): Use
15446 long_unsigned_type_node instead of unsigned_type_node as the offset
15447 field type, which is "pointer" mode in emutls.c.
15449 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15452 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
15454 PR middle-end/81207
15455 * gimple-fold.c (replace_call_with_call_and_fold): Handle
15456 gimple_vuse copying separately from gimple_vdef copying.
15458 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15460 * value-prof.c (free_hist): Remove call to memset and the enclosing if
15463 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15464 Olivier Hainque <hainque@adacore.com>
15466 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
15467 for all vxworks7 targets.
15468 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
15469 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
15470 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
15471 variations for VX6/VX7 and 32/64bits later on in ...
15472 (VXWORKS_LIB_SPEC): Leverage new macros.
15473 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
15474 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
15476 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15478 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
15479 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
15481 2017-06-26 Carl Love <cel@us.ibm.com>
15483 * config/rs6000/rs6000-c.c: Add support for built-in functions
15484 vector bool char vec_reve (vector bool char);
15485 vector signed char vec_reve (vector signed char);
15486 vector unsigned char vec_reve (vector unsigned char);
15487 vector bool int vec_reve (vector bool int);
15488 vector signed int vec_reve (vector signed int);
15489 vector unsigned int vec_reve (vector unsigned int);
15490 vector bool long long vec_reve (vector bool long long);
15491 vector signed long long vec_reve (vector signed long long);
15492 vector unsigned long long vec_reve (vector unsigned long long);
15493 vector bool short vec_reve (vector bool short);
15494 vector signed short vec_reve (vector signed short);
15495 vector double vec_reve (vector double);
15496 vector float vec_reve (vector float);
15497 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
15498 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
15499 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
15500 (altivec_vreve): New pattern.
15501 * config/rs6000/altivec.h (vec_reve): New define.
15502 * doc/extend.texi (vec_rev): Update the built-in documentation file
15503 for the new built-in functions.
15505 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15507 PR tree-optimization/71815
15508 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
15510 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
15512 (slsr_process_phi): Likewise.
15513 (replace_uncond_cands_and_profitable_phis): Don't replace a
15514 multiply candidate with a stride of 1 (copy or cast).
15515 (phi_incr_cost): Call uses_consumed_by_stmt rather than
15517 (lowest_cost_path): Likewise.
15518 (total_savings): Likewise.
15520 2017-06-26 Richard Biener <rguenther@suse.de>
15523 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
15524 Use def_builtin_pure for all gather builtins.
15526 2017-06-26 Richard Biener <rguenther@suse.de>
15528 PR tree-optimization/81203
15529 * tree-tailcall.c (find_tail_calls): Do not move stmts into
15530 non-dominating BBs.
15532 2017-06-26 Marek Polacek <polacek@redhat.com>
15535 * doc/invoke.texi: Document -Wmultistatement-macros.
15537 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
15539 * doc/sourcebuild.texi (ARM-specific attributes): Document new
15540 arm_neon_ok_no_float_abi effective target.
15542 2017-06-26 Richard Biener <rguenther@suse.de>
15544 PR tree-optimization/80928
15545 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
15546 (copy_bbs): Set BB_DUPLICATED flag early.
15547 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
15549 (execute_on_shrinking_pred): Likewise.
15550 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
15551 BB_DUPLICATED blocks.
15552 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
15553 iterate over all PHIs considering removal of *gsi.
15555 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
15557 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
15560 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15562 * config/rs6000/rs6000-string.c: (expand_block_clear,
15563 do_load_for_compare, select_block_compare_mode,
15564 compute_current_alignment, expand_block_compare,
15565 expand_strncmp_align_check, expand_strn_compare,
15566 expand_block_move, rs6000_output_load_multiple)
15567 Move functions related to string/block move/compare
15568 to a separate file.
15569 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
15570 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
15571 for this function which is now used in two files.
15572 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
15573 * config.gcc: Add rs6000-string.o to extra_objs for
15574 targets powerpc*-*-* and rs6000*-*-*.
15576 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
15579 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
15580 32-bit, since indexed is not valid for DImode.
15581 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
15582 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
15583 (define_peephole2 for Altivec d-form load): Add 32-bit support.
15584 (define_peephole2 for Altivec d-form store): Likewise.
15587 * multiple_target.c (create_dispatcher_calls): Only create the
15588 dispatcher call if the function is the default clone of a
15589 versioned function.
15591 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
15593 PR middle-end/80902
15594 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
15595 a call, force the call to not be a tail call.
15597 2017-06-23 Jeff Law <law@redhat.com>
15599 * doc/contrib.texi: Add entry for Steven Pemberton's work on
15602 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
15604 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
15605 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
15606 handling for early expansion of vector shifts (sl,sr,sra,rl).
15607 (builtin_function_type): Add vector shift right instructions
15608 to the unsigned argument list.
15610 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
15612 rtl-optimizatoin/79286
15613 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
15614 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
15615 trap. PIC register plus a const unspec without offset can never trap.
15617 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
15619 * tree.h (builtin_structptr_type): New type.
15620 (builtin_structptr_types): Declare new array.
15621 * tree.c (builtin_structptr_types): New array.
15622 (free_lang_data, build_common_tree_nodes): Use it.
15624 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
15627 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
15630 2017-06-22 Matt Turner <mattst88@gmail.com>
15632 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
15633 Lake models to skylake case. Assume skylake for unknown
15634 models with clflushopt.
15636 2017-06-22 Jeff Law <law@redhat.com>
15638 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
15639 frame sizes that do not satisfy aarch64_uimm12_shift.
15641 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
15643 * profile-count.h (apply_probability,
15644 apply_scale, probability_in): Fix checks for zero.
15646 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15648 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
15649 * doc/cppdiropts.texi (-I @var{dir}): Document it.
15651 2016-06-22 Richard Biener <rguenther@suse.de>
15653 * tree-vect-loop.c (vect_model_reduction_cost): Handle
15654 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
15655 REDUC_MAX_EXPR support.
15656 (vectorizable_reduction): Likewise.
15657 (vect_create_epilog_for_reduction): Likewise.
15659 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
15661 * match.pd (A / (1 << B) -> A >> B): New.
15662 * generic-match-head.c: Include optabs-tree.h.
15663 * gimple-match-head.c: Likewise.
15664 * optabs-tree.h (target_supports_op_p): New.
15665 * optabs-tree.c (target_supports_op_p): New.
15667 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15669 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
15670 $gcc_cv_ld --help output.
15671 (gcc_cv_ld_demangle): Likewise.
15672 (gcc_cv_ld_eh_frame_hdr): Likewise.
15673 (gcc_cv_ld_pie): Likewise.
15674 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
15675 (gcc_cv_ld_buildid): Likewise.
15676 (gcc_cv_ld_sysroot): Likewise.
15677 (ld_bndplt_support): Likewise.
15678 (ld_pushpopstate_support): Likewise.
15679 * configure: Regenerate.
15680 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
15682 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15685 * config/i386/sse.md (round<mode>2): Renumber match_dup and
15686 operands indexes to avoid gap between operands and match_dups.
15688 2017-06-21 Andrew Pinski <apinski@cavium.com>
15690 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
15691 Increment Arith_shift and Arith_shift_reg by 1.
15692 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
15694 * config/aarch64/aarch64.c (thunderx_tunings): Enable
15695 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
15696 (aarch64_strip_extend): Add new argument and test for it.
15697 (aarch64_cheap_mult_shift_p): New function.
15698 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
15699 add a cost if it is true.
15700 Update calls to aarch64_strip_extend.
15701 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
15703 2017-06-21 Andrew Pinski <apinski@cavium.com>
15705 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
15707 (thunderxt88): Likewise.
15708 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
15709 (thunderx_prefetch_tune): New variable.
15710 (thunderx2t99_prefetch_tune): Update for the correct values.
15711 (thunderxt88_tunings): New variable.
15712 (thunderx_tunings): Use thunderx_prefetch_tune instead of
15713 generic_prefetch_tune.
15714 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
15716 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15718 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
15719 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
15720 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
15721 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
15722 (aarch64_atomic_cas<mode>, GPI): Likewise.
15724 2017-06-21 Martin Liska <mliska@suse.cz>
15726 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
15727 statements on cold and hot labels.
15728 * predict.c (tree_estimate_probability_bb): Remove the
15729 prediction from this place.
15731 2017-06-21 Martin Liska <mliska@suse.cz>
15733 PR tree-optimization/79489
15734 * gimplify.c (maybe_add_early_return_predict_stmt): New
15736 (gimplify_return_expr): Call the function.
15737 * predict.c (tree_estimate_probability_bb): Remove handling
15739 * predict.def: Update comment about early return predictor.
15740 * gimple-predict.h (is_gimple_predict): New function.
15741 * predict.def: Change default value of early return to 66.
15742 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
15744 * passes.def: Put pass_strip_predict_hints to the beginning of
15747 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
15749 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
15750 FUNCTION_DECL declarations.
15751 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
15753 (dwaf2out_decl): Likewise.
15754 * godump.c (go_early_global_decl): Skip call to the real debug hook
15755 for FUNCTION_DECL declarations.
15756 * passes.c (rest_of_decl_compilation): Skip call to the
15757 early_global_decl debug hook for FUNCTION_DECL declarations, unless
15758 -fdump-go-spec is passed.
15760 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15762 * config/i386/i386.c (struct builtin_isa): New field pure_p.
15763 Reorder for compactness.
15764 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
15765 (def_builtin_pure, def_builtin_pure2): New functions.
15766 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
15768 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15770 * match.pd (nop_convert): New predicate.
15771 ((A +- CST1) +- CST2): Allow some NOP conversions.
15773 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15776 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
15777 with ctors/dtors if GOVD_SHARED is set.
15779 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15781 * config/aarch64/aarch64.md (movti_aarch64):
15782 Emit mov rather than orr.
15783 (movtf_aarch64): Likewise.
15784 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
15785 Emit mov rather than orr.
15787 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15789 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
15790 Swap alternatives, make integer dup more expensive.
15792 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15794 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
15795 Return true for non-tls symbols.
15797 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
15799 * config/aarch64/aarch64-cores.def (cortex-a55): New.
15800 (cortex-a75): Likewise.
15801 (cortex-a75.cortex-a55): Likewise.
15802 * config/aarch64/aarch64-tune.md: Regenerate.
15803 * doc/invoke.texi (-mtune): Document new values for -mtune.
15805 2017-06-21 Tom de Vries <tom@codesourcery.com>
15807 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
15808 stack_size feature.
15809 (Effective-Target Keywords, Other attributes): Suggest using
15810 dg-add-options stack_size feature to get stack limit in stack_size
15811 effective target documentation.
15813 2017-06-21 Julian Brown <julian@codesourcery.com>
15814 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15816 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
15817 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
15818 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
15820 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
15821 attribute type list for neon_multiply.
15822 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
15823 attribute type list for neon_multiply.
15824 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
15825 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
15826 attribute type list for neon_multiply.
15827 * config/arm/types.md (crypto_pmull): Add.
15828 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
15829 attribute type list.
15831 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
15833 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
15836 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15838 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
15839 to make sure not to dereference a NULL cost_classes_ptr pointer.
15841 2017-06-20 Carl Love <cel@us.ibm.com>
15843 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15844 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15845 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15846 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15847 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15848 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15849 VMULOSW): New enum "unspec" values.
15850 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15851 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15852 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15853 altivec_vmulosw): New patterns.
15854 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15855 VMULOSW): Add definitions.
15857 2017-06-20 Julia Koval <julia.koval@intel.com>
15859 * config/i386/i386.c: Fix rounding expand for new pattern.
15860 * config/i386/subst.md: Fix pattern (parallel -> unspec).
15862 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15864 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
15865 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
15867 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15869 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
15872 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15874 * config/aarch64/aarch64-cores.def: Rearrange to sort by
15875 architecture, then by implementer ID.
15876 * config/aarch64/aarch64-tune.md: Regenerate.
15878 2017-06-20 Richard Biener <rguenther@suse.de>
15880 PR middle-end/81097
15881 * fold-const.c (split_tree): Fold to type before negating.
15883 2017-06-20 David Malcolm <dmalcolm@redhat.com>
15885 * diagnostic-show-locus.c
15886 (selftest::test_fixit_deletion_affecting_newline): New function.
15887 (selftest::diagnostic_show_locus_c_tests): Call it.
15889 2017-06-20 Andreas Schwab <schwab@suse.de>
15892 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
15895 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15897 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
15898 __ARM_FEATURE_COPROC according to support.
15900 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15902 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15903 Rewritten to avoid overflow for > 32-bit pointers.
15906 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
15907 by removing enum keyword.
15908 (ubsan_type_descriptor): Likewise. Formatting fix.
15911 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
15912 splitter): Require TARGET_SSE2 in the condition.
15914 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
15917 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
15918 for doing vector set of SFmode on ISA 3.0.
15919 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
15920 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
15922 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
15923 SFmode value into a V4SF variable that was extracted from another
15924 V4SF variable without converting the element to double precision
15925 and back to single precision vector format.
15926 (vsx_insert_extract_v4sf_p9_2): Likewise.
15928 2017-06-19 Jakub Jelinek <jakub@redhat.com>
15930 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
15931 in UWHI to avoid undefined overflow.
15934 * ubsan.h (enum ubsan_encode_value_phase): New.
15935 (ubsan_encode_value): Change second argument to
15936 enum ubsan_encode_value_phase with default value of
15937 UBSAN_ENCODE_VALUE_GENERIC.
15938 * ubsan.c (ubsan_encode_value): Change second argument to
15939 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
15940 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
15941 create_tmp_var_raw instead of create_tmp_var and use a
15943 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
15944 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
15945 ubsan_encode_value callers.
15948 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
15949 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
15950 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
15952 2017-06-19 Richard Biener <rguenther@suse.de>
15954 PR middle-end/81118
15955 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
15956 estimates if we changed anything.
15958 2017-06-19 Richard Biener <rguenther@suse.de>
15960 PR tree-optimization/80887
15961 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
15962 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
15963 simplified lookups, then reset mprts_hook.
15964 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
15966 (try_to_simplify): Likewise.
15968 2017-06-19 Martin Liska <mliska@suse.cz>
15971 * gimplify.c (gimplify_switch_expr):
15972 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
15974 2017-06-19 Martin Liska <mliska@suse.cz>
15976 * doc/install.texi: Document that PGO runs in 4 stages.
15978 2017-06-19 Martin Liska <mliska@suse.cz>
15981 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
15982 to dispatcher function name.
15983 * multiple_target.c (replace_function_decl): New function.
15984 (create_dispatcher_calls): Redirect both edges and references.
15986 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
15988 * profile-count.c (profile_count::dump): Dump quality.
15989 (profile_count::differs_from_p): Update for unsigned val.
15990 * profile-count.h (profile_count_quality): New enum.
15991 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
15993 2017-06-19 Richard Biener <rguenther@suse.de>
15995 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
15996 struct function as arg.
15997 (estimate_numbers_of_iterations): Export overload with loop arg.
15998 (free_numbers_of_iterations_estimates_loop): Use an overload of
15999 free_numbers_of_iterations_estimates instead.
16000 * tree-cfg.c (remove_bb): Adjust.
16001 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
16002 * tree-parloops.c (gen_parallel_loop): Likewise.
16003 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
16005 (tree_unroll_loops_completely): Likewise.
16006 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
16007 Use an overload instead and export.
16008 (estimated_loop_iterations): Adjust.
16009 (max_loop_iterations): Likewise.
16010 (likely_max_loop_iterations): Likewise.
16011 (estimate_numbers_of_iterations): Take struct function as arg
16013 (loop_exits_before_overflow): Adjust.
16014 (free_numbers_of_iterations_estimates_loop): Use an overload.
16015 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
16016 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
16018 2017-06-19 Richard Biener <rguenther@suse.de>
16021 * ipa-prop.c (find_constructor_constant_at_offset): Handle
16022 RANGE_EXPR conservatively.
16024 2017-06-16 Carl Love <cel@us.ibm.com>
16026 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16027 definitions for vec_float, vec_float2, vec_floato,
16028 vec_floate built-ins.
16029 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
16030 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
16032 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
16033 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
16034 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
16035 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
16036 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
16037 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
16038 vec_floato): Add builtin defines.
16039 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
16040 Update the built-in documentation file for the new built-in
16043 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16045 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
16046 (mthumb): Mark as the negative of -marm.
16048 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16050 * doc/invoke.texi (ARM Options, -mcpu): Document supported
16052 (ARM Options, -mtune): Document that this accepts the same
16053 extension options as -mcpu.
16054 (ARM Options, -mfpu): Document addition of -mfpu=auto.
16056 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16058 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
16059 permitted extensions.
16061 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16063 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
16064 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
16065 (armv8-m.main): Add option +nodsp.
16066 * config/arm/arm-cpu-cdata.h: Regenerated.
16068 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16070 * config/arm/t-fuchsia: New file.
16071 * config.gcc (arm*-*-fuchsia*): Use it.
16073 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16075 * config/arm/t-symbian: Rewrite for new option infrastructure.
16077 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16079 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
16080 (MULTILIB_REQUIRED): Likewise.
16082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16084 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
16085 (MULTILIB_RESUE): Likewise.
16086 (MULTILIB_MATCHES): Likewise.
16087 (MULTLIB_REQUIRED): Likewise.
16089 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16091 * config/arm/t-rtems: Rewrite for new option framework.
16093 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16095 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
16096 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
16097 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
16098 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
16099 * config/arm/t-multilib: ... here.
16100 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
16101 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
16102 armv7-a and armv8*-a when A-profile libraries have not been built.
16103 * config/arm/t-rmprofile: Rewrite.
16105 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16107 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
16108 with a backslash. Remove the backslash after substituting unescaped
16110 * doc/fragments.texi (MULTILIB_REUSE): Document it.
16112 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16114 * config.gcc: (arm*-*-*): When building a-profile libraries, force
16115 the driver to pass through the default setting of -mfloat-abi.
16116 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
16118 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
16119 (all_feat_combs): New rule.
16120 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
16122 * config/arm/t-aprofile: Rewrite.
16124 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16126 * config/arm/arm.h (FPUTYPE_AUTO): Define.
16127 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
16128 fpu is not specified by the user/command-line.
16129 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
16130 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
16131 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
16132 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
16133 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
16134 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
16136 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16138 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
16139 * config/arm/t-arm-elf: Rewritten.
16141 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16143 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
16144 have some floating-point instructions.
16145 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
16146 (TARGET_MAYBE_HARD_FLOAT): New macro.
16147 * config/arm/arm-builtins.c (arm_init_builtins): Use
16148 TARGET_MAYBE_HARD_FLOAT.
16149 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
16151 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16153 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
16154 (configargs.h): Include it.
16155 (arm_print_hint_for_fpu_option): New function.
16156 (arm_parse_fpu_option): New function.
16157 (candidate_extension): New class.
16158 (arm_canon_for_multilib): New function.
16159 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
16160 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
16161 (ARCH_CANONICAL_SPECS): New macro.
16162 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
16164 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16166 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
16167 are set after handling multilib fragments. Set target_cpu_default2
16170 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16172 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
16174 (arm*-*-*): Set target_cpu_default2 to a quoted string.
16175 * config/arm/parsecpu.awk (check_cpu): Validate any extension
16177 (check_arch): Likewise.
16178 * config/arm/arm.c (arm_configure_build_target): Handle
16179 TARGET_CPU_DEFAULT being a string constant. Scan any feature
16180 options in the default.
16182 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16184 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
16185 when an option is an alias of another.
16186 * config/arm/parsecpu.awk (optalias): New parser token.
16187 (gen_comm_data): Mark non-alias options as such. Emit entries
16188 for extension aliases.
16189 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
16190 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
16191 (armv6kz, armv6zk, armv6t2): Likewise.
16192 (armv7): Make vfpv3-d16 an alias.
16193 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
16195 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
16196 Sort in canonical order.
16197 (armv8-a): Sort in canonical order.
16198 (armv8.1-a, armv8.2-a): Likewise.
16199 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
16201 (cortex-a9): Sort in canonical order.
16202 * config/arm/arm.c (selftests.h): Include it.
16203 (arm_test_cpu_arch_data): New function.
16204 (arm_run_self_tests): New function.
16205 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
16206 (targetm): Move declaration to the end of the file.
16207 * arm-cpu-cdata.h: Regenerated.
16209 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16211 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
16212 call to target_mode_check describing the type of option passed.
16213 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
16214 (arm_target_thumb_only): Use arm_parse_arch_option_name or
16215 arm_parse_cpu_option_name to match parameters against list of
16217 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
16218 arm_arch_core_flags data structure.
16219 * config/arm/arm-cpu_cdata.h: Regenerated.
16221 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16223 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
16225 (arm_print_hint_for_cpu_option): Likewise.
16226 (arm_print_hint_for_arch_option): Likewise.
16227 (arm_parse_cpu_option_name): Likewise.
16228 (arm_parse_arch_option_name): Likewise.
16229 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
16230 of entries in the all_fpus list.
16231 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
16232 (arm_parse_cpu_option_name): Declare.
16233 (arm_parse_arch_option_name): Declare.
16234 (arm_parse_option_features): Declare.
16235 (arm_intialize_isa): Declare.
16236 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
16238 (gen_comm_data): ... here. Make definitions non-static.
16239 * config/arm/arm-cpu-data.h: Regenerated.
16240 * config/arm/arm-cpu-cdata.h: Regenerated.
16242 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16244 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
16245 (cpu_arch_extension): New structure.
16246 (cpu_arch_option, arch_option, cpu_option): New structures.
16247 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
16248 architecture types.
16249 (gen_data): Generate new format data tables.
16250 * config/arm/arm.c (cpu_tune): New structure.
16251 (cpu_option, processors): Delete.
16252 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
16253 (arm_print_hint_for_cpu_option): ... this and ...
16254 (arm_print_hint_for_arch_option): ... this.
16255 (arm_parse_arch_cpu_name): Delete. Replace with ...
16256 (arm_parse_cpu_option_name): ... this and ...
16257 (arm_parse_arch_option_name): ... this.
16258 (arm_unrecognized_feature): Change type of target parameter to
16260 (arm_parse_arch_cpu_features): Delete. Replace with ...
16261 (arm_parse_option_features): ... this.
16262 (arm_configure_build_target): Rework to use new configuration data
16264 (arm_print_tune_info): Rework for new configuration data tables.
16265 * config/arm/arm-cpu-data.h: Regenerated.
16266 * config/arm/arm-cpu.h: Regenerated.
16268 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16270 * Makefile.in (OBJS): Move sbitmap.o from here ...
16271 (OBJS-libcommon): ... to here.
16273 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16275 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
16276 (ISA_ALL_CRYPTO): New macro.
16277 (ISA_ALL_SIMD): New macro
16278 (ISA_ALL_FP): New macro.
16279 * config/arm/arm.c (fpu_bitlist): Update initializer.
16280 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
16282 (arm9e): Add fpu. Add option for nofp
16283 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
16284 (arm926ej-s, arm1026ej-s): Likewise.
16285 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
16286 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
16287 neon-fp16, neon-vfpv4, nofp and nosimd.
16288 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
16289 (cortex-a8): Add fpu. Add option for nofp.
16290 (cortex-a9): Add fpu. Add options for nosimd and nofp.
16291 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
16292 (cortex-r4f): Add fpu.
16293 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
16294 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
16296 (cortex-r8): Likewise.
16297 (cortex-m4): Add fpu. Add option for nofp.
16298 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
16299 (cortex-a17.cortex-a7): Likewise.
16300 (cortex-a32): Add fpu. Add options for crypto and nofp.
16301 (cortex-a35, cortex-a53): Likewise.
16302 (cortex-a57): Add fpu. Add option for crypto.
16303 (cortex-a72, cortex-a73): Likewise.
16304 (exynos-m1): Likewise.
16305 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
16306 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
16307 (cortex-m33): Add fpu. Add option for nofp.
16308 * config/arm/arm-cpu-cdata.h: Regenerated
16309 * config/arm/arm-cpu-data.h: Regenerated.
16311 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16313 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
16314 (armv5te, armv5tej): Likewise.
16315 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
16316 (armv7): Add options fp and vfpv3-d16.
16317 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
16318 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
16320 (armv7ve): Likewise.
16321 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
16322 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
16323 (armv8-a): Add nocrypto option.
16324 (armv8.1-a, armv8.2-a): Likewise.
16325 (armv8-m.main): add options fp, fp.dp and nofp.
16327 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16329 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
16331 (armv8-a+crc): Delete.
16332 (armv8.1-a): Add options simd, crypto and nofp.
16333 (armv8.2-a): Add options fp16, simd, crypto and nofp.
16334 (armv8.2-a+fp16): Delete.
16335 (armv8-m.main): Add option dsp.
16336 (armv8-m.main+dsp): Delete.
16337 (cortex-a8): Add fpu. Add nofp option.
16338 (cortex-a9): Add fpu. Add nofp and nosimd options.
16339 * config/arm/parsecpu.awk (gen_data): Generate option tables and
16340 link to main cpu and architecture data structures.
16341 (gen_comm_data): Only put isa attributes from the main architecture
16343 (option): New statement for architecture and CPU entries.
16344 * arm.c (struct cpu_option): New structure.
16345 (struct processors): Add entry for options.
16346 (arm_unrecognized_feature): New function.
16347 (arm_parse_arch_cpu_name): Ignore any characters after the first
16349 (arm_parse_arch_cpu_feature): New function.
16350 (arm_configure_build_target): Separate out any CPU and architecture
16351 features and parse separately. Don't error out if -mfpu=auto is
16352 used with only an architecture string.
16353 (arm_print_asm_arch_directives): New function.
16354 (arm_file_start): Call it.
16355 * config/arm/arm-cpu-cdata.h: Regenerated.
16356 * config/arm/arm-cpu-data.h: Likewise.
16357 * config/arm/arm-tables.opt: Likewise.
16359 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16361 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
16362 assembler when it is not -mfpu=auto.
16364 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16366 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
16367 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
16368 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
16369 (ASM_CPU_SPEC): Rewrite.
16370 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
16371 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
16372 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
16373 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
16374 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
16375 copied string is NUL-terminated. Also strip any characters prefixed
16377 (arm_rewrite_selected_arch): New function.
16378 (arm_rewrite_march): New function.
16380 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16382 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
16383 (x_arm_cpu_string, x_arm_tune_string): Likewise.
16384 (march, mcpu, mtune): Convert to string-based options.
16385 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
16386 (arm_parse_arch_cpu_name): New function.
16387 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
16388 identify selected architecture or CPU.
16389 (arm_option_save): New function.
16390 (TARGET_OPTION_SAVE): Redefine.
16391 (arm_option_restore): Restore string options.
16392 (arm_option_print): Print string options.
16394 2017-06-16 Martin Sebor <msebor@redhat.com>
16396 PR tree-optimization/80933
16397 PR tree-optimization/80934
16398 * builtins.c (fold_builtin_3): Do not handle bcmp here.
16399 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
16400 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
16401 (gimple_fold_builtin): Call them.
16403 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16405 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
16406 as unlikely; update profile.
16408 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16410 * predict.c (force_edge_cold): Handle declaring edges impossible
16413 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16415 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
16417 (try_unroll_loop_completely): Fix reporting.
16419 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16421 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
16423 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
16426 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
16427 if given a non-constant argument for an intrinsic which requires a
16430 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16432 * profile.c (compare_freqs): New function.
16433 (branch_prob): Sort edge list.
16434 (find_spanning_tree): Assume that the list is priority sorted.
16436 2017-06-16 Richard Biener <rguenther@suse.de>
16438 PR tree-optimization/81090
16439 * passes.def (pass_record_bounds): Remove.
16440 * tree-pass.h (make_pass_record_bounds): Likewise.
16441 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
16442 make_pass_record_bounds): Likewise.
16443 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
16444 not free niter estimates at the beginning but at the end.
16445 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
16447 2017-06-16 Richard Biener <rguenther@suse.de>
16449 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
16450 initializer to workaround ICE in host GCC 4.8.
16452 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16454 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
16456 (clone_inlined_nodes): Update.
16458 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16460 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
16461 prefetch settings, and enable prefetching by default at -O3.
16463 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16465 * config/aarch64/aarch64.c (aarch64_override_options_internal):
16466 Set flag_prefetch_loop_arrays according to tuning data.
16468 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16470 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
16471 New tune structure.
16472 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
16473 [Unrelated to main purpose of the patch] Place the pointer field last
16474 to enable type checking errors when tune structure are wrongly merged.
16475 * config/aarch64/aarch64.c (generic_prefetch_tune,)
16476 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
16477 (thunderx2t99_prefetch_tune): New tune constants.
16478 (tune_params *_tunings): Update all tunings (no functional change).
16479 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
16480 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
16481 from tunings structures.
16483 2017-06-16 Jakub Jelinek <jakub@redhat.com>
16486 * ubsan.c (instrument_null): Add T argument, use it instead
16487 of computing it based on IS_LHS.
16488 (instrument_object_size): Likewise.
16489 (pass_ubsan::execute): Adjust instrument_null and
16490 instrument_object_size callers to pass gimple_get_lhs or
16491 gimple_assign_rhs1 result to it. Use instrument_null instead of
16492 calling get_base_address and instrument_mem_ref. Handle
16493 aggregate call arguments for object-size sanitization.
16495 2017-06-16 Yury Gribov <tetra2005@gmail.com>
16497 PR tree-optimization/81089
16498 * tree-vrp.c (is_masked_range_test): Validate operands of
16501 2017-06-15 Martin Sebor <msebor@redhat.com>
16504 * dumpfile.c (dump_register): Avoid calling memset to initialize
16505 a class with a default ctor.
16506 * gcc.c (struct compiler): Remove const qualification.
16507 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
16508 * hash-table.h: Ditto.
16509 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
16511 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
16512 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
16514 * params.h (struct param_info): Make struct members non-const.
16515 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
16516 with default initialization.
16517 * vec.h (vec_copy_construct, vec_default_construct): New helper
16519 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
16520 with vec_copy_construct.
16521 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
16522 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
16523 * doc/invoke.texi (-Wclass-memaccess): Document.
16525 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16527 * emit-rtl.h (is_leaf): Update comment about local
16528 register allocator.
16530 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16533 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
16534 for a variable to have a section before checking if the section has a
16536 Set section to.persistent if persistent attribute is set.
16537 Warn if .persistent attribute is used on an automatic variable.
16539 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
16541 PR rtl-optimization/80474
16542 * reorg.c (update_block): Do not ignore instructions in a delay slot.
16544 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
16546 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
16549 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
16551 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
16552 (casesi): Emit bounds checking as RTL.
16553 (casesi_internal_mips16_<mode>): Remove bounds checking.
16555 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
16557 * config/xtensa/xtensa.c (xtensa_option_override): Append
16558 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
16559 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
16560 xtensa_doloop_hooks): Define unconditionally.
16561 (xtensa_reorg_loops): Only call reorg_loops in the presence of
16563 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
16564 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
16565 for it in xtensa_option_override.
16566 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
16567 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
16569 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
16571 * doc/cppopts.texi: Document '-' special value to -MF.
16573 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
16575 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
16576 (cortex_a53_fconst): Likewise.
16577 (cortex_a53_fpmul): Likewise.
16578 (cortex_a53_f_load_64): Likewise.
16579 (cortex_a53_f_load_many): Likewise.
16580 (cortex_a53_advsimd_alu): Likewise.
16581 (cortex_a53_advsimd_alu_q): Likewise.
16582 (cortex_a53_advsimd_mul): Likewise.
16583 (cortex_a53_advsimd_mul_q): Likewise.
16584 (fpmac bypass): Add new bypass for fpmac-fpmac case.
16585 Add missing fmul, r2f_cvt and fconst cases.
16587 2017-06-14 Richard Biener <rguenther@suse.de>
16589 PR middle-end/81088
16590 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
16592 (fold_binary_loc): When associating do not treat pre-existing
16593 TREE_OVERFLOW on literal constants as a reason to allow
16594 TREE_OVERFLOW on associated literal constants.
16596 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16598 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
16599 (MASK_FEATURES): New macro.
16600 * config/sparc/sparc.c (sparc_option_override): Remove the special
16601 handling of -mfpu and generalize it to all MASK_FEATURES switches.
16603 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16605 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
16606 a division of 0 if non-call exceptions are enabled.
16608 2017-06-14 Andrew Pinski <apinski@cavium.com>
16609 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16612 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
16613 Improve vector initialization code gen for only variable case.
16615 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16617 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
16619 2017-06-14 Richard Biener <rguenther@suse.de>
16621 PR tree-optimization/81083
16622 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
16625 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16627 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
16628 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
16629 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
16630 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
16631 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
16632 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
16634 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16636 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
16637 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
16639 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16641 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
16643 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16645 * config/rs6000/t-rtems: Don't handle SPE.
16647 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16649 * config/rs6000/t-linux: Don't handle SPE.
16651 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16653 * config/rs6000/eabispe.h: Delete file.
16655 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16657 * config/rs6000/t-spe: Delete file.
16659 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16661 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
16662 (rs6000_legitimate_offset_address_p): Return false for anything in
16663 V2SImode or V2SFmode.
16665 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16667 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
16668 except V2SF and V2SI. Rearrange the vector modes, and add comments.
16669 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
16671 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
16672 (rs6000_legitimate_offset_address_p): Ditto.
16673 (rs6000_emit_move): Ditto.
16674 (rs6000_init_builtins): Remove V4HI_type_node.
16676 2017-06-13 Martin Liska <mliska@suse.cz>
16679 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
16680 (gate_asan): Likewise.
16681 * asan.h (asan_no_sanitize_address_p): Remove the function.
16682 (sanitize_flags_p): New function.
16683 * builtins.def: Fix coding style.
16684 * common.opt: Use renamed enum value.
16685 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
16686 * doc/extend.texi: Document no_sanitize attribute.
16687 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
16688 to SANITIZE_UNDEFINED_NONDEFAULT.
16689 * gcc.c (sanitize_spec_function): Use the renamed enum value.
16690 * gimple-fold.c (optimize_atomic_compare_exchange_p):
16691 Use sanitize_flags_p.
16692 * gimplify.c (gimplify_function_tree): Likewise.
16693 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
16694 * opts.c (parse_no_sanitize_attribute): New function.
16695 (common_handle_option): Use renamed enum value.
16696 * opts.h (parse_no_sanitize_attribute): Declare.
16697 * tree.c (sanitize_flags_p): New function.
16698 * tree.h: Declared here.
16699 * tsan.c: Use sanitize_flags_p.
16700 * ubsan.c (ubsan_expand_null_ifn): Likewise.
16701 (instrument_mem_ref): Likewise.
16702 (instrument_bool_enum_load): Likewise.
16703 (do_ubsan_in_current_function): Remove the function.
16704 (pass_ubsan::execute): Use sanitize_flags_p.
16705 * ubsan.h: Remove do_ubsan_in_current_function
16706 * tree-cfg.c (print_no_sanitize_attr_value): New function.
16707 (dump_function_to_file): Use it here.
16709 2017-06-13 Martin Jambor <mjambor@suse.cz>
16711 PR tree-optimization/80803
16712 PR tree-optimization/81063
16713 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
16714 (propagate_subaccesses_across_link): Enqueue subtree whenever
16715 necessary instead of relying on the caller.
16717 2017-06-13 Martin Jambor <mjambor@suse.cz>
16719 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
16720 that have a first_link.
16721 (sort_and_splice_var_accesses): Do not check first_link before
16723 (subtree_mark_written_and_enqueue): Likewise.
16724 (propagate_all_subaccesses): Likewise and do not stop at first
16725 parent with a first_link.
16727 2017-06-13 Martin Jambor <mjambor@suse.cz>
16729 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
16732 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16734 * match.pd: New pattern.
16736 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16738 * tree-vrp.c (is_masked_range_test): New function.
16739 (register_edge_assert_for): Determine ranges for
16742 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16744 PR tree-optimization/67328
16745 * fold-const.c (maskable_range_p): New function.
16746 (build_range_check): Generate bittests if possible.
16748 2017-06-13 Martin Liska <mliska@suse.cz>
16750 * gimple-pretty-print.c (dump_probability): Add new argument.
16751 (dump_edge_probability): Dump both probability and count.
16752 (dump_gimple_label): Likewise.
16753 (dump_gimple_bb_header): Likewise.
16755 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
16758 * config/avr/avr-devices.c: Fix indentation.
16759 * config/avr/gen-avr-mmcu-specs.c: Dito.
16761 2017-06-13 Richard Biener <rguenther@suse.de>
16763 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
16764 instead get vector type from stmt_info.
16765 (vectorizable_reduction): Adjust. Remove dead code.
16767 2017-06-13 Richard Biener <rguenther@suse.de>
16769 PR middle-end/81065
16770 * fold-const.c (extract_muldiv_1): Remove bogus distribution
16771 case of C * (x * C2 + C3).
16772 (fold_addr_of_array_ref_difference): Properly fold index difference.
16774 2017-06-12 David S. Miller <davem@davemloft.net>
16777 * config/sparc/sparc.md (return expander): Emit frame blockage if
16778 function uses alloca.
16780 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16782 * combine.c (make_field_assignment): Check len rather than the mode
16783 precision when calling force_to_mode.
16785 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
16787 Support multilibs and devices that see flash in RAM address range.
16790 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
16791 (avr_mcu_t) <flash_pm_offset>: New field.
16792 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
16793 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
16794 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
16795 (AVR_TINY_PM_OFFSET): Remove macro.
16796 * config/avr/avr.opt (-mshort-calls): New option.
16797 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16798 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
16799 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
16800 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
16801 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
16802 instead of avr_arch->have_jmp_call.
16803 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
16804 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
16805 avr_arch->flash_pm_offset to define.
16806 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
16807 new field flash_pm_offset. Add entry for avrxmega3.
16808 (avr_texinfo): Add entry for avrxmega3.
16809 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
16810 attiny212, attiny214,
16811 attiny412, attiny414, attiny416, attiny417,
16812 attiny814, attiny816, attiny817,
16813 attiny1614, attiny1616, attiny1617,
16814 attiny3214, attiny3216, attiny3217.
16815 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
16816 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
16817 (avr_print_operand_address) [AVR_TINY]: Same.
16818 (avr_asm_init_sections) <readonly_data_section>: Only patch
16819 callback if avr_arch->flash_pm_offset = 0.
16820 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
16821 for rodata if avr_arch->flash_pm_offset != 0.
16822 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
16823 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
16824 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
16825 (m_options): Append opt_rcall.
16826 (m_dirnames): Append dir_rcall.
16827 * config/avr/t-multilib: Regenerate.
16829 * configure.ac [target=avr]: Check whether avrxmega3 default
16830 linker description file works as needed.
16831 * configure: Regenerate.
16832 * doc/avr-mmcu.texi: Regenerate.
16833 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
16834 <__AVR_ARCH__>: Document avrxmega3 and 103.
16835 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
16836 <__AVR_SHORT_CALLS__>: Document it.
16837 <__AVR_PM_BASE_ADDRESS__>: Document it.
16838 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
16839 (AVR Variable Attributes) <progmem>: Document this is
16840 not needed for avrxmega3.
16841 (AVR Named Address Spaces) <__flash>: Dito.
16843 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
16845 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
16847 2017-06-12 Doug Rupp <rupp@adacore.com>
16849 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
16850 Append vxworks-stdint.h to the tm_file list.
16851 * config/vxworks-stdint.h: New file.
16853 2017-06-12 Martin Liska <mliska@suse.cz>
16855 PR tree-optimization/81041
16856 * tree-profile.c (gimple_gen_ic_func_profiler):
16857 Create an extra BB in profile-generate
16858 (gimple_gen_time_profiler): Likewise.
16860 2017-06-12 Jakub Jelinek <jakub@redhat.com>
16862 PR tree-optimization/81003
16863 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
16864 (update_range_test): Use it instead of force_gimple_operand_gsi.
16866 2017-06-12 Richard Biener <rguenther@suse.de>
16868 PR tree-optimization/81053
16869 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
16870 with backedge value not defined in loop. Simplify def stmt
16873 2017-06-11 Tom de Vries <tom@codesourcery.com>
16876 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
16878 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
16879 nvptx_cannot_force_const_mem.
16881 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16883 * opts.c (finish_options): Move test for flag_split_stack after
16884 it has been initialized.
16886 2017-06-11 Jason Merrill <jason@redhat.com>
16888 * tree.h (id_equal): New.
16889 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
16890 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
16891 instead of strcmp of IDENTIFIER_POINTER.
16893 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16895 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
16896 (mark_all_inlined_calls_cdtor): Fix formating.
16897 (inline_transform): Rescale profile before inlining.
16899 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16901 * cgraph.h (cgraph_edge::clone): Update prototype.
16902 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
16903 (cgraph_node::create_clone): Update.
16904 (cgraph_node::create_version_clone): Update.
16905 * tree-inline.c (copy_bb): Update.
16906 (expand_call_inline): Update.
16908 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16910 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
16911 factored out from ...
16912 (rs6000_emit_prologue): ... here.
16914 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16916 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
16917 factored out from ...
16918 (rs6000_emit_prologue): ... here.
16920 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16922 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
16924 (handle_missing_profiles): Fix computation of tp_first_run.
16925 (counts_to_freqs): Do not touch freqs when count is 0.
16927 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16929 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
16932 2017-06-10 Tom de Vries <tom@codesourcery.com>
16934 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
16935 attributes): Document signal effective target.
16937 2017-06-10 Tom de Vries <tom@codesourcery.com>
16939 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
16940 Document effective target stack_size.
16942 2017-06-09 David Malcolm <dmalcolm@redhat.com>
16944 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
16945 to the edit_context if they can be auto-applied.
16947 2017-06-9 Ian Lance Taylor <iant@golang.org>
16949 * opts.c (finish_options): If -fsplit-stack, disable implicit
16950 -forder-blocks-and-partition.
16951 * doc/invoke.texi (Optimize Options): Document that when using
16952 -fsplit-stack -forder-blocks-and-partition is not implicitly
16955 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16957 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
16958 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
16959 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
16960 * builtins.def (abort, trap, unreachable): Declare cold.
16961 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
16962 * tree-core.h (ECF_COLD): New.
16963 * tree.c (set_call_expr_flags): Handle ECF_COLD.
16964 (build_common_builtin_nodes): Mark unreachable and abort as cold.
16966 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16968 * predict.c (unlikely_executed_stmt_p): Cleanup.
16970 2017-06-09 Richard Biener <rguenther@suse.de>
16972 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
16973 model if the ref is always written to.
16975 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16977 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
16979 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16981 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
16984 2017-06-09 Tom de Vries <tom@codesourcery.com>
16987 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
16988 "target cannot support label values" when encountering LABEL_REF.
16990 2017-06-09 Martin Liska <mliska@suse.cz>
16992 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
16993 (gimple_gen_ic_func_profiler): Emit direct comparison
16994 of __gcov_indirect_call_callee with NULL.
16995 (gimple_gen_time_profiler): Change probability from
16996 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
16998 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17000 * profile.c (edge_gcov_counts): Turn to pointer.
17001 (compute_branch_probabilities, compute_branch_probabilities): Update.
17002 (branch_prob): Do not clear edge_gcov_count.
17003 * profile.h (edge_gcov_counts): Turn to pointer.
17004 (edge_gcov_count): Update.
17006 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17008 * gimple.h (gimple_check_failed): Mark cold.
17010 2017-06-09 Richard Biener <rguenther@suse.de>
17012 PR tree-optimization/66623
17013 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
17014 refactor check_reduction into two parts, properly computing
17015 whether we have to check reduction validity for outer loop
17018 2017-06-09 Richard Biener <rguenther@suse.de>
17020 PR tree-optimization/79483
17021 * graphite-scop-detection.c (order): New global.
17022 (get_order): Compute bb to order mapping that satisfies code
17023 generation constraints.
17024 (cmp_pbbs): New helper.
17025 (build_scops): Start domwalk at entry block, sort generated
17028 2017-06-09 Richard Biener <rguenther@suse.de>
17030 PR middle-end/81007
17031 * ipa-polymorphic-call.c
17032 (ipa_polymorphic_call_context::restrict_to_inner_class):
17033 Skip FIELD_DECLs with error_mark_node type.
17034 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
17037 2017-06-09 Martin Liska <mliska@suse.cz>
17039 * predict.c (struct branch_predictor): New struct.
17040 (test_prediction_value_range): New test.
17041 (predict_c_tests): New function.
17042 * selftest-run-tests.c (selftest::run_tests): Run the function.
17043 * selftest.h: Declare new tests.
17045 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
17048 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
17049 gen_add3_insn did not fail.
17050 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
17051 r0, construct that number in a temporary reg and add that reg to r0.
17052 If asked to put the result in r0 as well, fail.
17054 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
17056 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17057 for early expansion of vec_eqv.
17059 2017-06-08 Jakub Jelinek <jakub@redhat.com>
17061 PR middle-end/81005
17062 * ubsan.c (instrument_null): Avoid pointless code temporary.
17063 (pass_ubsan::execute): Instrument aggregate arguments of calls.
17065 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
17069 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
17072 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
17075 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17077 * predict.c (unlikely_executed_edge_p): Move ahead.
17078 (probably_never_executed_edge_p): Use it.
17080 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17082 PR middle-end/79988
17083 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
17084 gimple_call_builtin_p call.
17086 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17088 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
17089 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
17090 rtl_check_failed_type2, rtl_check_failed_code1,
17091 rtl_check_failed_code2, rtl_check_failed_code_mode,
17092 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
17093 rtvec_check_failed_bounds, rtl_check_failed_flag,
17094 _fatal_insn_not_found, _fatal_insn): Likewise.
17095 * tree.h (tree_contains_struct_check_failed,
17096 tree_check_failed, tree_not_check_failed,
17097 tree_class_check_failed, tree_range_check_failed,
17098 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
17099 tree_vec_elt_check_failed, phi_node_elt_check_failed,
17100 tree_operand_check_failed, omp_clause_check_failed,
17101 omp_clause_operand_check_failed, omp_clause_range_check_failed):
17104 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17106 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
17107 flag_branch_probabilities.
17108 * ipa-inline.c (edge_badness): Likewise.
17109 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17110 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
17111 * predict.c (maybe_hot_frequency_p): Likewise.
17112 (probably_never_executed): Likewise.
17113 * sched-ebb.c (schedule_ebbs): Likewise.
17114 * sched-rgn.c (find_single_block_region): Likewise.
17115 * tracer.c (tail_duplicate): Likewise.
17117 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17119 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
17120 longer requires x_flag_profile_use.
17122 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17124 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
17125 instead of flag_reorder_blocks_and_partition.
17126 * dbxout.c (dbxout_function_end): Likewise.
17127 * dwarf2out.c (gen_subprogram_die): Likewise.
17128 * haifa-sched.c (sched_create_recovery_edges): Likewise.
17129 * hw-doloop.c (reorg_loops): Likewise.
17130 * varasm.c (assemble_start_function,
17131 assemble_end_function): Likewise.
17132 (decide_function_section): Do not check for
17133 flag_reorder_blocks_and_partition.
17135 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17137 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17139 (chkp_get_hard_register_fake_addr_expr): Ditto.
17140 (chkp_build_addr_expr): Add check for hard reg case.
17141 (chkp_parse_array_and_component_ref): Ditto.
17142 (chkp_find_bounds_1): Ditto.
17143 (chkp_process_stmt): Don't generate bounds store for
17146 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17148 * predict.c (maybe_hot_bb_p): Do not check profile status.
17149 (maybe_hot_edge_p): Likewise.
17150 (probably_never_executed): Check for zero counts even if profile
17152 (unlikely_executed_edge_p): New function.
17153 (unlikely_executed_stmt_p): New function.
17154 (unlikely_executed_bb_p): New function.
17155 (set_even_probabilities): Use unlikely predicates.
17156 (combine_predictions_for_bb): Likewise.
17157 (predict_paths_for_bb): Likewise.
17158 (predict_paths_leading_to_edge): Likewise.
17159 (determine_unlikely_bbs): New function.
17160 (estimate_bb_frequencies): Use it.
17161 (compute_function_frequency): Use zero counts even if profile is
17163 * profile-count.h: Fix typo.
17165 2017-08-08 Julia Koval <julia.koval@intel.com>
17167 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
17168 _mm512_mask_cvtsepi16_storeu_epi8,
17169 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
17170 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
17171 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
17172 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
17173 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
17174 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
17175 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
17176 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
17177 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
17178 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
17179 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
17180 __builtin_ia32_pmovuswb256mem_mask,
17181 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
17182 __builtin_ia32_pmovwb512mem_mask): New builtins.
17184 2017-08-08 Julia Koval <julia.koval@intel.com>
17186 PR target/73350,80862
17187 * config/i386/subst.md (round): Fix round pattern.
17188 * config/i386/i386.c (ix86_erase_embedded_rounding):
17189 Fix erasing rounding for the fixed pattern.
17191 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17193 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
17195 2017-06-08 Martin Liska <mliska@suse.cz>
17197 PR gcov-profile/80911
17198 * gcov.c (block_info::block_info): New constructor.
17200 2017-06-07 Carl Love <cel@us.ibm.com>
17202 * config/rs6000/rs6000-c: The return type of the following
17203 built-in functions was implemented as int not long long. Fix sign
17204 of return value for the unsigned version of vec_mulo and vec_mule.
17205 vector unsigned long long vec_bperm (vector unsigned long long,
17206 vector unsigned char)
17207 vector signed long long vec_mule (vector signed int,
17209 vector unsigned long long vec_mule (vector unsigned int,
17210 vector unsigned int)
17211 vector signed long long vec_mulo (vector signed int,
17213 vector unsigned long long vec_mulo (vector unsigned int,
17214 vector unsigned int)
17215 * doc/extend.texi: Fix the documentation for the built-in
17218 2017-06-07 Carl Love <cel@us.ibm.com>
17221 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
17224 2017-06-07 Carl Love <cel@us.ibm.com>
17226 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
17227 support, Generate doublehv for signed int/float for BE case only.
17229 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
17231 * doc/invoke.texi (mcx16): Rewrite.
17233 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17235 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
17236 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
17237 *mov<mode>_softfloat, and an anonymous splitter): Use
17238 nonimmediate_operand instead of rs6000_nonimmediate_operand.
17240 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17242 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
17244 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17245 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
17246 (rs6000_debug_reg_global): Adjust.
17247 (rs6000_init_hard_regno_mode_ok): Adjust.
17248 (rs6000_dbx_register_number): Adjust.
17249 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
17250 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17251 Remove SPE_ACC and SPEFSCR.
17252 (REG_ALLOC_ORDER): Ditto.
17253 (FRAME_POINTER_REGNUM): Change to 111.
17254 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
17255 (REG_CLASS_NAMES): Ditto.
17256 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
17257 (REGISTER_NAMES): Ditto.
17258 (ADDITIONAL_REG_NAMES): Ditto.
17259 (rs6000_reg_names): Ditto.
17260 * config/rs6000/rs6000.md: Renumber some register number
17263 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17265 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
17267 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17268 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
17270 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
17271 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17272 Delete the SPE high registers.
17273 (REG_ALLOC_ORDER): Ditto.
17274 (enum reg_class): Remove SPE_HIGH_REGS.
17275 (REG_CLASS_NAMES): Ditto.
17276 (REG_CLASS_CONTENTS): Delete the SPE high registers.
17277 (REGISTER_NAMES): Ditto.
17278 (rs6000_reg_names): Ditto.
17279 * doc/tm.texi.in: Remove SPE as example.
17280 * doc/tm.texi: Regenerate.
17282 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17284 * config/rs6000/8540.md (ppc8540_brinc): Delete.
17285 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
17286 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
17287 * config/rs6000/rs6000.md (type): Remove "brinc".
17289 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17291 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
17292 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
17293 * config/rs6000/linuxspe.h: Delete file.
17294 * config/rs6000/rs6000.md: Don't include spe.md.
17295 * config/rs6000/spe.h: Delete file.
17296 * config/rs6000/spe.md: Delete file.
17297 * config/rs6000/t-rs6000: Remove spe.md.
17299 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17301 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
17302 (reg_or_none500mem_operand): Delete.
17303 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
17304 instead of reg_or_none500mem_operand.
17306 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17308 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17309 handling of SPE flags.
17310 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
17312 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17314 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17316 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
17317 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
17318 paired_divv2sf3): Similar.
17319 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
17320 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
17321 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
17323 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
17324 Rename the paired_* instruction patterns.
17325 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
17327 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
17328 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
17329 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
17330 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
17331 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
17332 PAIRED_VECTOR_MODE.
17333 (struct machine_function): Delete field spe_insn_chain_scanned_p.
17334 (spe_func_has_64bit_regs_p): Delete.
17335 (spe_expand_predicate_builtin): Delete.
17336 (spe_expand_evsel_builtin): Delete.
17337 (TARGET_DWARF_REGISTER_SPAN): Do not define.
17338 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
17339 (invalid_e500_subreg): Delete.
17340 (rs6000_legitimize_address): Always force_reg op2 as well, for
17341 paired single memory accesses.
17342 (rs6000_member_type_forces_blk): Delete.
17343 (rs6000_spe_function_arg): Delete.
17344 (rs6000_expand_unop_builtin): Delete SPE handling.
17345 (rs6000_expand_binop_builtin): Ditto.
17346 (spe_expand_stv_builtin): Delete.
17347 (bdesc_2arg_spe): Delete.
17348 (spe_expand_builtin): Delete.
17349 (spe_expand_predicate_builtin): Delete.
17350 (spe_expand_evsel_builtin): Delete.
17351 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
17352 (spe_init_builtins): Delete.
17353 (spe_func_has_64bit_regs_p): Delete.
17354 (savres_routine_name): Delete "info" parameter. Adjust callers.
17355 (rs6000_emit_stack_reset): Ditto.
17356 (rs6000_dwarf_register_span): Delete.
17357 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
17358 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
17359 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
17361 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
17363 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
17364 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
17365 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
17366 mulv2sf3, divv2sf3): Delete expanders.
17368 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17370 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
17372 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17374 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
17375 * config/rs6000/rs6000.c: Ditto.
17377 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17379 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
17380 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
17381 comparison_operator.
17383 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17385 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
17386 * config/rs6000/rs6000.opt: Ditto.
17387 * config/rs6000/t-rtems: Ditto.
17389 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17391 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
17392 TARGET_E500_SINGLE by 0, simplify.
17393 * config/rs6000/rs6000.c: Ditto.
17394 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
17395 (spe_build_register_parallel): Delete.
17396 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
17397 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
17398 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
17399 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
17400 (E500_CONVERT): Delete.
17401 * config/rs6000/spe.md: Remove many patterns and all define_constants.
17403 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17405 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
17406 * config/rs6000/dfp.md: Ditto.
17407 (negdd2, *negdd2_fpr): Merge.
17408 (absdd2, *absdd2_fpr): Merge.
17409 (negtd2, *negtd2_fpr): Merge.
17410 (abstd2, *abstd2_fpr): Merge.
17411 * config/rs6000/e500.h: Delete file.
17412 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
17413 TARGET_FPRS by 1 and simplify.
17414 * config/rs6000/rs6000-c.c: Ditto.
17415 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
17416 TARGET_DF_SPE by 0.
17417 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
17419 * config/rs6000/rs6000.md: Ditto.
17420 (floatdidf2, *floatdidf2_fpr): Merge.
17421 (move_from_CR_gt_bit): Delete.
17422 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
17423 (E500_CR_IOR_COMPARE): Delete.
17424 (All patterns that require !TARGET_FPRS): Delete.
17425 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
17427 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17429 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
17431 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17433 * graphds.c (add_edge): Intitialize edge's attached data.
17434 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
17435 pointer parameter. Call pointed function on each edge during
17436 graph traversing. Skip traversing the edge when the function
17438 (graphds_dfs, graphds_scc): Ditto.
17439 (for_each_edge): New parameter. Pass the new parameter to callback
17441 * graphds.h (skip_edge_callback): New function pointer type.
17442 (graphds_dfs, graphds_scc): New function pointer parameter.
17443 (graphds_edge_callback, for_each_edge): New parameter.
17445 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17447 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
17448 out code checking if runtime alias check is possible to below ...
17449 Call the new function.
17450 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
17451 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
17453 2017-06-07 Marek Polacek <polacek@redhat.com>
17456 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
17457 TYPE_OVERFLOW_WRAPS check.
17459 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17461 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
17462 if versioning is required.
17463 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
17464 peeling with the check for versioning.
17466 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17468 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
17469 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
17470 Set true to new parameter if new ssa variable is defined.
17471 (vect_gen_vector_loop_niters): Refactor. Set range information
17472 for the new vector loop bound variable.
17473 (vect_do_peeling): Ditto.
17475 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17477 * tree-affine.c (ssa.h): Include header file.
17478 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
17479 has wrapping overflow behavior.
17481 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17483 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
17485 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17487 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
17488 (tree_to_aff_combination): ... here.
17490 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17492 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
17493 reg_pressure model function.
17494 (ivopts_global_cost_for_size): Delete.
17495 (determine_set_costs, iv_ca_recount_cost): Call new model function
17496 ivopts_estimate_reg_pressure.
17498 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17500 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
17501 expensive than udiv. Remove floating point cases from mod.
17503 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17505 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
17506 Increase idiv cost.
17508 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17510 * config/aarch64/aarch64.md
17511 (copysignsf3): Fix mask generation.
17513 2017-06-07 Jakub Jelinek <jakub@redhat.com>
17515 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
17517 (class dump_manager): Add register_dumps method.
17518 * dumpfile.c: Include langhooks.h.
17519 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
17520 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
17521 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
17522 (dump_manager::dump_register): Start with 512 entries instead of 32.
17523 (dump_manager::register_dumps): New method.
17524 * toplev.c (general_init): Instead of invoking register_dumps
17525 langhook, invoke register_dumps method on the dump manager.
17526 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
17529 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
17531 * doc/md.texi: Clarify the restrictions on a define_insn condition.
17532 Say that # requires an associated define_split to exist, and that
17533 the define_split must be suitable for use after register allocation.
17535 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17537 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
17538 (compute_outgoing_frequencies): Also initialize zero counts.
17539 (find_many_sub_basic_blocks): Do not produce uninitialized profile
17540 around loops; preserve more of profile when nothing changes.
17542 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
17544 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
17546 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
17547 * config/arm/arm-cpu-cdata.h: Regenerate.
17548 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
17549 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
17550 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
17551 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
17552 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
17554 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
17556 * config/arm/t-rmprofile: Likewise.
17557 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
17559 2017-06-06 David S. Miller <davem@davemloft.net>
17562 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
17563 blockage if function uses alloca.
17565 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17567 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
17568 New "uid" fields to hold pretty-print IDs of group and ref.
17569 Memory references are now identified as <group_id>:<ref_id>
17570 instead of using [random] addresses.
17571 (dump_mem_details): Simplify, no functional change.
17572 (dump_mem_ref): Simplify and make output more concise.
17573 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
17574 (find_or_create_group): Initialize group uid.
17575 (record_ref): Initialize ref uid. Improve debug output.
17576 (prune_group_by_reuse, should_issue_prefetch_p,)
17577 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
17578 (mark_nontemporal_store, determine_loop_nest_reuse):
17579 Improve debug output.
17581 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17583 * dbgcnt.def (prefetch): New debug counter.
17584 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
17585 (schedule_prefetches): Stop issueing prefetches if debug counter
17588 2017-06-06 Tom de Vries <tom@codesourcery.com>
17590 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
17591 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
17593 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17595 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
17596 Use aarch64_reg_or_zero predicate for operand 4.
17597 (aarch64_compare_and_swap<mode> define_insn_and_split):
17598 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
17599 (aarch64_store_exclusive<mode>): Likewise for operand 2.
17601 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17603 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
17604 (arm_compute_save_core_reg_mask): This.
17605 (thumb1_compute_save_reg_mask): Rename into ...
17606 (thumb1_compute_save_core_reg_mask): This.
17607 (arm_compute_save_reg0_reg12_mask): Adapt comment.
17608 (arm_compute_frame_layout): Likewise.
17610 2017-06-06 Richard Biener <rguenther@suse.de>
17612 PR tree-optimization/80974
17613 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
17614 keep or clear leaders SSA info.
17616 2017-06-06 Tom de Vries <tom@codesourcery.com>
17618 * config/nvptx/nvptx.c (split_mode_p): New function.
17619 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
17621 2017-06-06 Tom de Vries <tom@codesourcery.com>
17623 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
17625 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17628 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
17631 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17633 * shrink-wrap.c (handle_simple_exit): Update profile.
17634 (try_shrink_wrapping): Upate profile.
17636 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17638 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
17639 (tree_guess_outgoing_edge_probabilities): New.
17640 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
17641 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
17643 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17645 * ipa-split.c (split_function): Initialize return bb profile.
17647 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17649 * profile.c (compute_branch_probabilities): Also initialize
17650 EXIT_BLOCK profile.
17652 2017-06-06 Richard Biener <rguenther@suse.de>
17654 PR tree-optimization/80928
17655 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
17656 (vect_analyze_loop_operations): Properly guard analysis for
17658 (vect_transform_loop): Likewise.
17659 (vect_analyze_loop_2): Also reset SLP type on PHIs.
17660 (vect_model_induction_cost): Do not cost for pure SLP.
17661 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
17662 of induction in inner loop vectorization.
17663 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
17664 (vect_get_and_check_slp_defs): Handle vect_induction_def.
17665 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
17667 (vect_analyze_slp_cost_1): Cost induction.
17668 (vect_detect_hybrid_slp_stmts): Handle PHIs.
17669 (vect_get_slp_vect_defs): Likewise.
17670 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
17671 (vect_transform_stmt): Handle SLP reductions.
17672 * tree-vectorizer.h (vectorizable_induction): Adjust.
17674 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17676 * config/rs6000/rs6000.c (make_resolver_func): Update
17677 init_lowered_empty_function call.
17679 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
17681 * doc/invoke.texi: Document the -fprofile-abs-path option.
17682 * common.opt (fprofile-abs-path): New option.
17683 * gcov-io.h (gcov_write_filename): Declare.
17684 * gcov-io.c (gcov_write_filename): New function.
17685 * coverage.c (coverage_begin_function): Use gcov_write_filename.
17686 * profile.c (output_location): Likewise.
17688 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17690 * shring-wrap.c: Revert accidental commit.
17692 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
17694 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
17696 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17698 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
17700 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
17701 profile in callgraph edge.
17702 * profile-count.h (apply_probability): If THIS is 0, then result is 0
17703 (apply_scale): Likewise.
17704 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
17705 Also scale profile when inlining function with zero profile.
17706 (initialize_cfun): Update exit block profile even when it is zero.
17707 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
17708 when profile is read.
17710 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17712 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
17713 (CLONE_*): New constants to define the processors we can generate
17714 code for with the target_clone attribute.
17715 (rs6000_clone_map): New array to identify which clone processors
17716 the current program is running on.
17717 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
17718 target_clone attribute.
17719 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
17720 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
17721 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
17722 (cpu_expand_builtin): Add support for target_clone attribute.
17723 (rs6000_valid_attribute_p): Allow "default" attribute.
17724 (get_decl_name): New debug function to simplify printing the
17725 current function name in debugging statements.
17726 (rs6000_clone_priority): New functions to support the target_clone
17727 attribute, and be able to generate code to switch between ISA 2.05
17728 through ISA 3.0 (power6 through power9).
17729 (rs6000_compare_version_priority): Likewise.
17730 (rs6000_get_function_versions_dispatcher): Likewise.
17731 (make_resolver_func): Likewise.
17732 (add_condition_to_bb): Likewise.
17733 (dispatch_function_versions): Likewise.
17734 (rs6000_generate_version_dispatcher_body): Likewise.
17735 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
17736 (fusion_gpr_load_p): Fix a spacing issue.
17737 * doc/extend.texi (Common Function Attributes): Document that the
17738 PowerPC supports the target_clone attribute.
17740 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17742 * config/arm/arm.h: explain F symbol found in description of ARM
17743 register allocation in its legend.
17745 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17747 * config/mips/frame-header-opt.c: Include profile-count.h.
17748 * config/riscv/riscv.c: Include profile-count.h
17750 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17752 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
17754 (sm_set_flag_if_changed): Add bbs field.
17755 (execute_sm_if_changed_flag_set): Pass BBS.
17756 (execute_sm): Update.
17758 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17760 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
17763 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17765 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
17766 (peephole2): New peephole2 to emit the above.
17767 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
17769 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17771 * config/aarch64/aarch64.c (define_peephole2 above
17772 *sub_<shift>_<mode>): New peephole.
17774 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17776 * config/i386/i386.c (make_resolver_func): Update.
17777 * Makefile.in: Add profile-count.h and profile-count.o
17778 * auto-profile.c (afdo_indirect_call): Update to new API.
17779 (afdo_set_bb_count): Update.
17780 (afdo_propagate_edge): Update.
17781 (afdo_propagate_circuit): Update.
17782 (afdo_calculate_branch_prob): Update.
17783 (afdo_annotate_cfg): Update.
17784 * basic-block.h: Include profile-count.h
17785 (struct edge_def): Turn count to profile_count.
17786 (struct basic_block_def): Likewie.
17787 (REG_BR_PROB_BASE): Move to profile-count.h
17788 (RDIV): Move to profile-count.h
17789 * bb-reorder.c (max_entry_count): Turn to profile_count.
17790 (find_traces): Update.
17791 (rotate_loop):Update.
17792 (connect_traces):Update.
17793 (sanitize_hot_paths):Update.
17794 * bt-load.c (migrate_btr_defs): Update.
17795 * cfg.c (RDIV): Remove.
17796 (init_flow): Use alloc_block.
17797 (alloc_block): Uninitialize count.
17798 (unchecked_make_edge): Uninitialize count.
17799 (check_bb_profile): Update.
17800 (dump_edge_info): Update.
17801 (dump_bb_info): Update.
17802 (update_bb_profile_for_threading): Update.
17803 (scale_bbs_frequencies_int): Update.
17804 (scale_bbs_frequencies_gcov_type): Update.
17805 (scale_bbs_frequencies_profile_count): New.
17806 * cfg.h (update_bb_profile_for_threading): Update.
17807 (scale_bbs_frequencies_profile_count): Declare.
17808 * cfgbuild.c (compute_outgoing_frequencies): Update.
17809 (find_many_sub_basic_blocks): Update.
17810 * cfgcleanup.c (try_forward_edges): Update.
17811 (try_crossjump_to_edge): Update.
17812 * cfgexpand.c (expand_gimple_tailcall): Update.
17813 (construct_exit_block): Update.
17814 * cfghooks.c (verify_flow_info): Update.
17815 (dump_bb_for_graph): Update.
17816 (split_edge): Update.
17817 (make_forwarder_block): Update.
17818 (duplicate_block): Update.
17819 (account_profile_record): Update.
17820 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
17821 (get_estimated_loop_iterations): Update.
17822 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
17823 (single_likely_exit): Update.
17824 * cfgloopmanip.c (scale_loop_profile): Update.
17826 (set_zero_probability): Update.
17827 (lv_adjust_loop_entry_edge): Update.
17828 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17829 (purge_dead_edges): Update.
17830 (rtl_account_profile_record): Update.
17831 * cgraph.c (cgraph_node::create): Uninitialize count.
17832 (symbol_table::create_edge): Uninitialize count.
17833 (cgraph_update_edges_for_call_stmt_node): Update.
17834 (cgraph_edge::dump_edge_flags): Update.
17835 (cgraph_node::dump): Update.
17836 (cgraph_edge::maybe_hot_p): Update.
17837 * cgraph.h: Include profile-count.h
17838 (create_clone), create_edge, create_indirect_edge): Update.
17839 (cgraph_node): Turn count to profile_count.
17840 (cgraph_edge0: Likewise.
17841 (make_speculative, clone): Update.
17842 (create_edge): Update.
17843 (init_lowered_empty_function): Update.
17844 * cgraphclones.c (cgraph_edge::clone): Update.
17845 (duplicate_thunk_for_node): Update.
17846 (cgraph_node::create_clone): Update.
17847 * cgraphunit.c (cgraph_node::analyze): Update.
17848 (cgraph_node::expand_thunk): Update.
17849 * final.c (dump_basic_block_info): Update.
17850 * gimple-streamer-in.c (input_bb): Update.
17851 * gimple-streamer-out.c (output_bb): Update.
17852 * graphite.c (print_global_statistics): Update.
17853 (print_graphite_scop_statistics): Update.
17854 * hsa-brig.c: Include basic-block.h.
17855 * hsa-dump.c: Include basic-block.h.
17856 * hsa-gen.c (T sum_slice): Update.
17857 (convert_switch_statements):Update.
17858 * hsa-regalloc.c: Include basic-block.h.
17859 * ipa-chkp.c (chkp_produce_thunks): Update.
17860 * ipa-cp.c (struct caller_statistics): Update.
17861 (init_caller_stats): Update.
17862 (gather_caller_stats): Update.
17863 (ipcp_cloning_candidate_p): Update.
17864 (good_cloning_opportunity_p): Update.
17865 (get_info_about_necessary_edges): Update.
17866 (dump_profile_updates): Update.
17867 (update_profiling_info): Update.
17868 (update_specialized_profile): Update.
17869 (perhaps_add_new_callers): Update.
17870 (decide_about_value): Update.
17871 (ipa_cp_c_finalize): Update.
17872 * ipa-devirt.c (struct odr_type_warn_count): Update.
17873 (struct decl_warn_count): Update.
17874 (struct final_warning_record): Update.
17875 (possible_polymorphic_call_targets): Update.
17876 (ipa_devirt): Update.
17877 * ipa-fnsummary.c (redirect_to_unreachable): Update.
17878 * ipa-icf.c (sem_function::merge): Update.
17879 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
17880 * ipa-inline.c (compute_uninlined_call_time): Update.
17881 (compute_inlined_call_time): Update.
17882 (want_inline_small_function_p): Update.
17883 (want_inline_self_recursive_call_p): Update.
17884 (edge_badness): Update.
17885 (lookup_recursive_calls): Update.
17886 (recursive_inlining): Update.
17887 (inline_small_functions): Update.
17888 (dump_overall_stats): Update.
17889 (dump_inline_stats): Update.
17890 * ipa-profile.c (ipa_profile_generate_summary): Update.
17891 (ipa_propagate_frequency): Update.
17892 (ipa_profile): Update.
17893 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17894 * ipa-utils.c (ipa_merge_profiles): Update.
17895 * loop-doloop.c (doloop_modify): Update.
17896 * loop-unroll.c (report_unroll): Update.
17897 (unroll_loop_runtime_iterations): Update.
17898 * lto-cgraph.c (lto_output_edge): Update.
17899 (lto_output_node): Update.
17900 (input_node): Update.
17901 (input_edge): Update.
17902 (merge_profile_summaries): Update.
17903 * lto-streamer-in.c (input_cfg): Update.
17904 * lto-streamer-out.c (output_cfg): Update.
17905 * mcf.c (create_fixup_graph): Update.
17906 (adjust_cfg_counts): Update.
17907 (sum_edge_counts): Update.
17908 * modulo-sched.c (sms_schedule): Update.
17909 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
17910 * predict.c (maybe_hot_count_p): Update.
17911 (probably_never_executed): Update.
17912 (dump_prediction): Update.
17913 (combine_predictions_for_bb): Update.
17914 (propagate_freq): Update.
17915 (handle_missing_profiles): Update.
17916 (counts_to_freqs): Update.
17917 (rebuild_frequencies): Update.
17918 (force_edge_cold): Update.
17919 * predict.h: Include profile-count.h
17920 (maybe_hot_count_p, counts_to_freqs): UPdate.
17921 * print-rtl-function.c: Do not include cfg.h
17922 * print-rtl.c: Include basic-block.h
17923 * profile-count.c: New file.
17924 * profile-count.h: New file.
17925 * profile.c (is_edge_inconsistent): Update.
17926 (correct_negative_edge_counts): Update.
17927 (is_inconsistent): Update.
17928 (set_bb_counts): Update.
17929 (read_profile_edge_counts): Update.
17930 (compute_frequency_overlap): Update.
17931 (compute_branch_probabilities): Update; Initialize and deinitialize
17933 (branch_prob): Update.
17934 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
17935 (edge_gcov_count): New.
17936 (bb_gcov_count): New.
17937 * shrink-wrap.c (try_shrink_wrapping): Update.
17938 * tracer.c (better_p): Update.
17939 * trans-mem.c (expand_transaction): Update.
17940 (ipa_tm_insert_irr_call): Update.
17941 (ipa_tm_insert_gettmclone_call): Update.
17942 * tree-call-cdce.c: Update.
17943 * tree-cfg.c (gimple_duplicate_sese_region): Update.
17944 (gimple_duplicate_sese_tail): Update.
17945 (gimple_account_profile_record): Update.
17946 (execute_fixup_cfg): Update.
17947 * tree-inline.c (copy_bb): Update.
17948 (copy_edges_for_bb): Update.
17949 (initialize_cfun): Update.
17950 (freqs_to_counts): Update.
17951 (copy_cfg_body): Update.
17952 (expand_call_inline): Update.
17953 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17954 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
17955 (try_unroll_loop_completely): Update.
17956 (try_peel_loop): Update.
17957 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17958 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
17959 * tree-ssa-loop-split.c (connect_loops): Update.
17960 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
17961 * tree-ssa-reassoc.c (branch_fixup): Update.
17962 * tree-ssa-tail-merge.c (replace_block_by): Update.
17963 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
17964 (compute_path_counts): Update.
17965 (update_profile): Update.
17966 (recompute_probabilities): Update.
17967 (update_joiner_offpath_counts): Update.
17968 (estimated_freqs_path): Update.
17969 (freqs_to_counts_path): Update.
17970 (clear_counts_path): Update.
17971 (ssa_fix_duplicate_block_edges): Update.
17972 (duplicate_thread_path): Update.
17973 * tree-switch-conversion.c (case_bit_test_cmp): Update.
17974 (struct switch_conv_info): Update.
17975 * tree-tailcall.c (decrease_profile): Update.
17976 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17977 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17978 * value-prof.c (check_counter): Update.
17979 (gimple_divmod_fixed_value): Update.
17980 (gimple_mod_pow2): Update.
17981 (gimple_mod_subtract): Update.
17982 (gimple_ic_transform): Update.
17983 (gimple_stringop_fixed_value): Update.
17984 * value-prof.h (gimple_ic): Update.
17986 2017-06-02 Carl Love <cel@us.ibm.com>
17988 * config/rs6000/rs6000-c: Add support for built-in functions
17989 vector double vec_doublee (vector signed int);
17990 vector double vec_doublee (vector unsigned int);
17991 vector double vec_doublee (vector float);
17992 vector double vec_doubleh (vector signed int);
17993 vector double vec_doubleh (vector unsigned int);
17994 vector double vec_doubleh (vector float);
17995 vector double vec_doublel (vector signed int);
17996 vector double vec_doublel (vector unsigned int);
17997 vector double vec_doublel (vector float);
17998 vector double vec_doubleo (vector signed int);
17999 vector double vec_doubleo (vector unsigned int);
18000 vector double vec_doubleo (vector float);.
18001 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
18002 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
18004 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
18005 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
18006 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
18008 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
18009 vec_doublel, vec_doubleh.
18010 * doc/extend.texi: Update the built-in documentation file for the
18011 new built-in functions.
18013 2017-06-02 David Malcolm <dmalcolm@redhat.com>
18016 * ipa-inline-analysis.c (free_growth_caches): Set
18017 edge_removal_hook_holder to NULL after removing it.
18019 2017-06-02 Sudakshina Das <sudi.das@arm.com>
18021 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
18022 comparision with zero.
18024 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18025 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18026 for early expansion of vec_min and vec_max builtins.
18027 (builtin_function_type): Add min/max unsigned variants to those
18028 identified as having unsigned arguments.
18030 2017-06-02 Olivier Hainque <hainque@adacore.com>
18032 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
18034 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18036 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
18037 Use VALL_F16 iterator rather than VALL.
18039 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18041 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
18042 Emit CBNZ inside loop when doing a strong exchange and comparing
18043 against zero. Generate the CC flags after the loop.
18045 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
18047 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
18048 (dl_section_ref): New.
18049 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
18050 On AIX, append an expression to subtract the size of the
18051 section length to dl_section_ref.
18053 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18055 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18056 for early expansion of vector absolute builtins.
18058 2017-06-02 Richard Biener <rguenther@suse.de>
18060 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
18061 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
18063 2017-06-02 Richard Biener <rguenther@suse.de>
18065 PR tree-optimization/80948
18066 * tree-tailcall.c (find_tail_calls): Track stmts to move in
18067 stmt order as well.
18069 2017-06-02 Richard Biener <rguenther@suse.de>
18071 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
18073 * tree-vect-stmts.c (process_use): Do not mark backedge defs
18074 for inductions as relevant.
18076 2017-06-02 Richard Biener <rguenther@suse.de>
18078 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
18079 (vectorizable_induction): ... this. Remove dead code.
18081 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
18083 * builtins. (expand_builtin_alloca): Remove second parameter and
18084 infer its value from the first parameter instead.
18085 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
18087 2017-06-02 Jakub Jelinek <jakub@redhat.com>
18089 PR rtl-optimization/80903
18090 * loop-doloop.c (add_test): Unshare sequence.
18092 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
18094 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
18096 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
18098 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
18100 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
18101 xlogue_layout::get_instance, logue_layout::xlogue_layout,
18102 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
18103 (xlogue_layout::get_stub_rtx): Make static.
18104 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
18105 (xlogue_layout::compute_stub_managed_regs): Rename to...
18106 (xlogue_layout::count_stub_managed_regs): ...this.
18107 (xlogue_layout::is_stub_managed_reg): New function.
18108 (xlogue_layout::m_stub_names): Rename to...
18109 (xlogue_layout::s_stub_names): ...this, make static.
18110 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
18111 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
18112 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
18113 xlogue_layout::s_stub_names): Instantiate statics.
18114 (stub_managed_regs): Remove.
18115 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
18116 (disable_call_ms2sysv_xlogues): Rename to...
18117 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
18118 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
18120 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
18121 change after reload_completed.
18122 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
18124 (ix86_expand_prologue): Likewise.
18125 (ix86_expand_epilogue): Likewise.
18126 (ix86_expand_split_stack_prologue): Likewise.
18127 (ix86_compute_frame_layout): Remove frame parameter ...
18128 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
18129 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
18131 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
18132 (ix86_frame): Move from here ...
18133 * config/i386/i386.h (ix86_frame): ... to here.
18134 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
18135 complete ix86_frame data structure instead. Remove some_ld_name.
18137 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18139 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
18140 symbols that hold a DECL_VALUE_EXPR.
18142 2017-06-01 Martin Jambor <mjambor@suse.cz>
18144 PR tree-optimization/80898
18145 * tree-sra.c (process_subtree_disqualification): Removed.
18146 (disqualify_candidate): Do not acll
18147 process_subtree_disqualification.
18148 (subtree_mark_written_and_enqueue): New function.
18149 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
18150 RHS has been disqualified and re-queue LHS if necessary. Apart
18151 from that, ignore disqualified RHS.
18153 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18155 * config/s390/s390.c (s390_emit_epilogue): Disable early return
18156 address fetch for z10 or later.
18158 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18160 * config/arc/arc.md (tst_movb): Add guard when splitting.
18162 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18164 * config/arc/arc.c (arc_can_eliminate): Test against
18165 arc_frame_pointer_needed.
18167 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18169 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
18170 to prevent store reordering.
18171 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
18172 (type): Add block type.
18173 (stack_tie): Define special instruction to be used in
18176 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18178 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
18179 constraint. It is not valid for the pattern.
18180 (noncommutative_binary_comparison): Likewise.
18182 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18184 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
18187 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18189 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
18190 be used by the reg-alloc.
18192 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18194 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
18195 reg-alloc when having mul64 or mul32x16 instructions.
18196 (mulsidi3): Likewise.
18197 (umulsidi3): Likewise.
18198 (mulsi32x16): New pattern.
18199 (mulsi64): Likewise.
18200 (mulsidi64): Likewise.
18201 (umulsidi64): Likewise.
18202 (MUL32x16_REG): Define.
18203 (mul64_600): Use MUL32x16_REG.
18204 (mac64_600): Likewise.
18205 (umul64_600): Likewise.
18206 (umac64_600): Likewise.
18208 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18210 * config/arc/arc.md (mulsi3_700): Make it commutative.
18212 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
18214 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
18216 (*sign_extendsidi2_insn): Likewise for movstosw.
18218 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18220 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
18221 the type of the input discriminant value. Convert the
18222 discriminant value of signedness vary.
18224 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
18226 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
18227 Add to -Wall section.
18229 2017-06-01 Richard Biener <rguenther@suse.de>
18231 PR middle-end/66313
18232 * fold-const.c (fold_plusminus_mult_expr): If the factored
18233 factor may be zero use a wrapping type for the inner operation.
18234 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
18235 and handle moved defs.
18236 (process_assignment): Properly guard the unary op case. Return a
18237 tri-state indicating that moving the stmt before the call may allow
18238 to continue. Pass through to_move.
18239 (find_tail_calls): Handle moving unrelated defs before
18242 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
18245 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
18246 splitter result in the canonical way.
18248 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
18250 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
18251 also for 32bit target. Update insn attributes.
18252 (zero-extendsidi2 splitter): Allow all registers for operand 1.
18254 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
18256 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
18257 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
18258 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
18259 (_mm_maskz_min_ss): New intrinsics.
18261 2017-05-31 Martin Liska <mliska@suse.cz>
18263 * tree-vect-loop.c (vect_create_epilog_for_reduction):
18264 Change comment style to one we normally use.
18265 (vectorizable_reduction): Likewise.
18266 (vectorizable_induction): Likewise.
18267 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18268 (vectorizable_call): Likewise.
18269 (vectorizable_simd_clone_call): Likewise.
18270 (vectorizable_conversion): Likewise.
18271 (vectorizable_assignment): Likewise.
18272 (vectorizable_shift): Likewise.
18273 (vectorizable_operation): Likewise.
18274 (vectorizable_store): Likewise.
18275 (vectorizable_load): Likewise.
18276 * tree-vectorizer.h: Likewise.
18278 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
18280 * passes.c (emergency_dump_function): New.
18281 * tree-pass.h (emergency_dump_function): Declare.
18282 * plugin.c (plugins_internal_error_function): Remove.
18283 * plugin.h (plugins_internal_error_function): Remove declaration.
18284 * toplev.c (internal_error_function): New static function. Use it...
18285 (general_init): ...here.
18287 2017-05-31 Graham Markall <graham.markall@embecosm.com>
18289 * config/arc/arc.c (arc_print_operand): Handle constant operands.
18290 (arc_rtx_costs): Add costs for new patterns.
18291 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
18292 * config/arc/predicates.md: Add _1_2_3_operand predicate.
18294 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18296 * tree-ssa-strlen.c (get_next_strinfo): New function.
18297 (get_stridx_plus_constant): Use it.
18298 (zero_length_string): Likewise.
18299 (adjust_related_strinfos): Likewise.
18300 (adjust_last_stmt): Likewise.
18302 2017-05-31 Richard Biener <rguenther@suse.de>
18305 * config/i386/i386.c (ix86_expand_builtin): Remove assert
18306 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
18308 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18310 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
18311 loop_vinfo argument and use of dependence distance vectors.
18312 Check instead whether the two references differ only in their
18313 initial value and assume that they have the same alignment if the
18314 difference is a multiple of the vector alignment.
18315 (vect_analyze_data_refs_alignment): Update call accordingly.
18317 2017-05-31 Martin Liska <mliska@suse.cz>
18320 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
18322 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18324 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
18325 (create_intersect_range_checks): Move from ...
18326 * tree-data-ref.c (create_intersect_range_checks_index)
18327 (create_intersect_range_checks): ... to here.
18328 (create_runtime_alias_checks): New function factored from ...
18329 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
18330 here. Call above function.
18331 * tree-data-ref.h (create_runtime_alias_checks): New function.
18333 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18335 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
18336 segment length for dr_b and compute it in wide_int.
18338 2017-05-31 Richard Biener <rguenther@suse.de>
18340 PR tree-optimization/80906
18341 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
18342 and pass through iv_map.
18343 (copy_bb_and_scalar_dependences): Adjust.
18344 (translate_pending_phi_nodes): Likewise.
18345 (copy_loop_close_phi_args): Handle code-generating IVs instead
18348 2017-05-30 David Malcolm <dmalcolm@redhat.com>
18350 * diagnostic-color.c (color_dict): Add "type-diff".
18351 (parse_gcc_colors): Update comment.
18352 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
18353 -fdiagnostics-show-template-tree and -fno-elide-type.
18354 (GCC_COLORS): Add type-diff to example.
18356 (-fdiagnostics-show-template-tree): New.
18357 (-fno-elide-type): New.
18358 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
18359 the pp_format_decoder callback. Call any m_format_postprocessor's
18361 (pretty_printer::pretty_printer): Initialize
18362 m_format_postprocessor.
18363 (pretty_printer::~pretty_printer): Delete any
18364 m_format_postprocessor.
18365 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
18366 (class format_postprocessor): New class.
18367 (struct pretty_printer::format_decoder): Document the new parameters.
18368 (struct pretty_printer::m_format_postprocessor): New field.
18369 * tree-diagnostic.c (default_tree_printer): Update for new
18370 bool and const char ** params.
18371 * tree-diagnostic.h (default_tree_printer): Likewise.
18373 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
18375 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
18376 (lwa_operand): Delete rs6000_gen_cell_microcode test.
18377 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18378 rs6000_gen_cell_microcode code.
18379 (rs6000_final_prescan_insn): Delete.
18380 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
18381 "warn-cell-microcode" entries.
18382 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
18383 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
18384 throughout. Change cc_reg_not_micro_cr0_operand to
18385 cc_reg_not_cr0_operand throughout.
18386 (*extendhi<mode>2_noload): Delete.
18387 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
18388 (mwarn-cell-microcode): Delete.
18389 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
18390 -mgen-cell-microcode and -mwarn-cell-microcode.
18392 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
18395 * config/i386/constraints.md (Yd): New constraint.
18397 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
18398 and (?Yd, r) alternatives. Update insn attributes.
18399 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
18400 and (?*Yd, r) alternatives. Update insn attributes.
18401 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
18403 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
18405 * gimplify.c (gimplify_modify_expr): Don't create a
18406 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
18409 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
18411 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
18413 2017-05-30 Richard Biener <rguenther@suse.de>
18415 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
18416 and reduc_def fields.
18417 (STMT_VINFO_REDUC_TYPE): New define.
18418 (STMT_VINFO_REDUC_DEF): Likewise.
18419 (vect_force_simple_reduction): Adjust prototype.
18420 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
18421 (vect_is_simple_reduction): Remove check_reduction argument.
18422 (vect_force_simple_reduction): Adjust and set
18423 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18424 (vectorizable_reduction): Do not re-do reduction analysis
18425 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18426 * tree-parloops.c (gather_scalar_reductions): Adjust.
18428 2017-05-30 Richard Biener <rguenther@suse.de>
18430 PR middle-end/80901
18431 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
18434 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
18436 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
18437 Introduce unknown_misalignment parameter and remove vf.
18438 (vect_peeling_hash_get_lowest_cost):
18439 Pass unknown_misalignment parameter.
18440 (vect_enhance_data_refs_alignment):
18441 Fix unsupportable data ref treatment.
18443 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18445 * tree-vect-data-refs.c (vect_get_data_access_cost):
18446 Workaround for SLP handling.
18447 (vect_enhance_data_refs_alignment):
18448 Compute costs for doing no peeling at all, compare to the best
18449 peeling costs so far and avoid peeling if cheaper.
18451 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18453 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
18454 Return peeling info and set costs to zero for unlimited cost
18456 (vect_enhance_data_refs_alignment): Also inspect all datarefs
18457 with unknown misalignment. Compute and costs for unknown
18458 misalignment, compare them to the costs for known misalignment
18459 and choose the cheapest for peeling.
18461 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18463 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
18464 (vect_get_peeling_costs_all_drs): Create function.
18465 (vect_peeling_hash_get_lowest_cost):
18466 Use vect_get_peeling_costs_all_drs.
18467 (vect_peeling_supportable): Create function.
18468 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
18470 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18472 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
18473 DR_HAS_NEGATIVE_STEP.
18474 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
18475 (vect_enhance_data_refs_alignment): Use.
18476 (vect_duplicate_ssa_name_ptr_info): Use.
18477 * tree-vectorizer.h (dr_misalignment): Use.
18478 (known_alignment_for_access_p): Use.
18480 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18483 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
18485 (has_section_name): New function.
18487 2017-05-30 Martin Liska <mliska@suse.cz>
18490 * auto-profile.c (get_function_decl_from_block): Fix
18493 2017-05-30 Richard Biener <rguenther@suse.de>
18495 PR middle-end/80876
18496 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
18498 2017-05-30 Martin Liska <mliska@suse.cz>
18500 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
18501 * dumpfile.h (struct dump_file_info): Remove ctors.
18503 2017-05-30 Martin Liska <mliska@suse.cz>
18505 * predict.def: Fix GNU coding style.
18507 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18509 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18510 Mark 'to' argument with ATTRIBUTE_UNUSED.
18512 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18514 * config/xtensa/xtensa.c (xtensa_emit_call): Use
18515 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
18516 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
18519 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
18521 * doc/install.texi (Options specification): Restore entry of
18522 --enable-sjlj-exceptions.
18524 2017-05-27 Michael Eager <eager@eagercon.com>
18527 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
18529 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
18531 * config/microblaze/microblaze.h
18532 (FIXED_REGISTERS): Update in macro.
18533 (CALL_USED_REGISTERS): Update in macro.
18535 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
18537 * doc/install.texi: Add links to macOS binary distributions.
18539 2017-05-27 Jakub Jelinek <jakub@redhat.com>
18543 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18545 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18547 2017-05-26 Martin Liska <mliska@suse.cz>
18549 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
18551 2017-05-26 Martin Liska <mliska@suse.cz>
18553 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
18554 always leading ';; '.
18555 (dump_bb_info): Likewise.
18556 (brief_dump_cfg): Likewise.
18557 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
18558 * dumpfile.c: Remove usage of TDF_VERBOSE.
18559 * dumpfile.h (enum dump_kind): Likewise.
18560 (dump_gimple_bb_header): Do not use TDF_COMMENT.
18561 * print-tree.c (debug_verbose): Remove.
18562 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
18563 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
18564 * tree-diagnostic.c (default_tree_printer): Replace
18565 TDF_DIAGNOSTIC with TDF_SLIM.
18567 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18569 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
18570 in parameter loop, rather than loop_vinfo.
18571 (create_intersect_range_checks): Ditto.
18572 (vect_create_cond_for_alias_checks): Update call to above functions.
18574 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18576 PR tree-optimization/80815
18577 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
18578 for merging runtime alias checks. Handle negative DR_STEPs.
18580 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18582 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
18584 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
18585 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
18586 out code pruning runtime alias checks.
18587 * tree-data-ref.c (prune_runtime_alias_test_list): New function
18588 factored out from above.
18589 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18591 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18593 (prune_runtime_alias_test_list): New decalaration.
18595 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18597 * tree-vect-data-refs.c (compare_tree): Rename and move ...
18598 * tree-data-ref.c (data_ref_compare_tree): ... to here.
18599 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
18600 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
18601 (operator==, comp_dr_with_seg_len_pair): Ditto.
18602 (vect_prune_runtime_alias_test_list): Ditto.
18604 2017-05-26 Martin Liska <mliska@suse.cz>
18607 * params.def: Bound partial-inlining-entry-probability param.
18609 2017-05-26 Marek Polacek <polacek@redhat.com>
18612 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
18615 2017-05-26 Richard Biener <rguenther@suse.de>
18617 PR tree-optimization/80842
18618 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
18621 2017-05-26 Richard Biener <rguenther@suse.de>
18623 PR tree-optimization/80844
18624 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
18626 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
18628 * doc/md.texi (Machine Constraints): Update x86 family
18629 machine constraints section to match 'config/i386/constraints.md'.
18631 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
18633 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
18635 2017-05-25 Nathan Sidwell <nathan@acm.org>
18637 * doc/invoke.texi (--enable-languages): Update documentation.
18639 2017-05-25 Martin Liska <mliska@suse.cz>
18641 * dumpfile.c: Add TDF_FOLDING.
18642 * dumpfile.h (enum dump_kind): Likewise.
18643 * genmatch.c (dt_simplify::gen_1): Use it.
18645 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18647 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
18649 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18651 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18652 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
18654 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18656 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
18657 * match.pd (X == C): Rewrite it here.
18658 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
18659 with_certain_nonzero_bits2): New predicates.
18660 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
18662 2017-05-24 Nathan Sidwell <nathan@acm.org>
18664 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
18667 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
18670 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18672 * config/powerpcspe: New port. Files are copied from the rs6000
18673 port, with "rs6000" in filenames replaced by "powerpcspe".
18675 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
18677 PR rtl-optimization/80754
18678 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
18680 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
18682 * config/sparc/sparc.md (length): Return the correct value for -mflat
18683 sibcalls to match output_sibcall.
18685 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18689 * config/rs6000/rs6000.c (struct machine_function): Add new field
18691 (rs6000_get_separate_components): Init that field, use it.
18692 (rs6000_components_for_bb): Use the field.
18694 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
18696 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
18698 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
18700 PR middle-end/80823
18701 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
18703 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18706 * config/s390/s390.c (s390_check_qrst_address): Check incoming
18707 address against address_operand predicate.
18708 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
18710 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
18712 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
18713 parameters passed indirectly.
18715 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
18717 * config/i386/i386.md (*movdi_internal): Remove SSE4
18718 alternative 18 (?r, *v). Update insn attributes.
18719 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
18720 Update insn attributes.
18721 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
18722 Update insn attributes.
18723 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
18724 alternative 1 (r, v). Remove isa attribute.
18725 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
18726 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
18727 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18729 2017-05-23 Tom de Vries <tom@codesourcery.com>
18731 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
18734 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18736 * cgraphunit.c (symbol_table::process_new_functions): Update.
18737 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
18738 (inline_generate_summary): Rename to ...
18739 (ipa_fn_summary_generate): ... this one.
18740 (inline_read_summary): Rename to ...
18741 (ipa_fn_summary_read): ... this one.
18742 (inline_write_summary): Rename to ...
18743 (ipa_fn_summary_write): ... this one.
18744 (inline_free_summary): Rename to ...
18745 (ipa_free_fn_summary): ... this one.
18746 (pass_data_local_fn_summary, pass_local_fn_summary,
18747 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
18748 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
18749 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
18750 make_pass_ipa_fn_summary): New.
18751 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
18752 inline_write_summary, inline_free_summary): Remove.
18753 (ipa_free_fn_summary) : New.
18754 * ipa-inline.c (ipa_inline): Update.
18755 (pass_ipa_inline): Do not generate summaries.
18756 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
18758 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
18759 and add pass_ipa_fn_summary.
18760 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
18762 (make_pass_inline_parameters): Remove.
18764 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
18766 * omp-low.c (struct omp_context): Remove "default_kind" member.
18769 * omp-offload.c (execute_oacc_device_lower): Remove the
18770 parallelism dimensions function attributes for unparallelized
18771 OpenACC kernels constructs.
18773 2017-05-23 Martin Liska <mliska@suse.cz>
18775 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
18777 (cgraph_edge::make_speculative): Likewise.
18778 (cgraph_edge::resolve_speculation): Likewise.
18779 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
18780 (cgraph_node::dump): Likewise.
18781 * cgraph.h: Likewise.
18782 * cgraphunit.c (analyze_functions): Likewise.
18783 (symbol_table::compile): Likewise.
18784 * ipa-cp.c (print_all_lattices): Likewise.
18785 (determine_versionability): Likewise.
18786 (initialize_node_lattices): Likewise.
18787 (ipcp_verify_propagated_values): Likewise.
18788 (estimate_local_effects): Likewise.
18789 (update_profiling_info): Likewise.
18790 (create_specialized_node): Likewise.
18791 (perhaps_add_new_callers): Likewise.
18792 (decide_about_value): Likewise.
18793 (decide_whether_version_node): Likewise.
18794 (identify_dead_nodes): Likewise.
18795 (ipcp_store_bits_results): Likewise.
18796 * ipa-devirt.c (dump_targets): Likewise.
18797 (ipa_devirt): Likewise.
18798 * ipa-icf.c (sem_item::dump): Likewise.
18799 (sem_function::equals): Likewise.
18800 (sem_variable::equals): Likewise.
18801 (sem_item_optimizer::read_section): Likewise.
18802 (sem_item_optimizer::execute): Likewise.
18803 (congruence_class::dump): Likewise.
18804 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
18805 (dump_inline_summary): Likewise.
18806 (estimate_node_size_and_time): Likewise.
18807 (inline_analyze_function): Likewise.
18808 * ipa-inline-transform.c (inline_call): Likewise.
18809 * ipa-inline.c (report_inline_failed_reason): Likewise.
18810 (want_early_inline_function_p): Likewise.
18811 (edge_badness): Likewise.
18812 (update_edge_key): Likewise.
18813 (inline_small_functions): Likewise.
18814 * ipa-profile.c (ipa_profile): Likewise.
18815 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
18816 (ipa_make_edge_direct_to_target): Likewise.
18817 (remove_described_reference): Likewise.
18818 (ipa_impossible_devirt_target): Likewise.
18819 (propagate_controlled_uses): Likewise.
18820 (ipa_print_node_params): Likewise.
18821 (ipcp_transform_function): Likewise.
18822 * ipa-pure-const.c (pure_const_read_summary): Likewise.
18823 (propagate_pure_const): Likewise.
18824 * ipa-reference.c (generate_summary): Likewise.
18825 (read_write_all_from_decl): Likewise.
18826 (propagate): Likewise.
18827 (ipa_reference_read_optimization_summary): Likewise.
18828 * ipa-utils.c (ipa_merge_profiles): Likewise.
18829 * ipa.c (walk_polymorphic_call_targets): Likewise.
18830 (symbol_table::remove_unreachable_nodes): Likewise.
18831 (ipa_single_use): Likewise.
18832 * passes.c (execute_todo): Likewise.
18833 * predict.c (drop_profile): Likewise.
18834 * symtab.c (symtab_node::get_dump_name): New function.
18835 (symtab_node::dump_name): Likewise.
18836 (symtab_node::dump_asm_name): Likewise.
18837 (symtab_node::dump_references): Likewise.
18838 (symtab_node::dump_referring): Likewise.
18839 (symtab_node::dump_base): Likewise.
18840 (symtab_node::debug_symtab): Likewise.
18841 * tree-sra.c (convert_callers_for_node): Likewise.
18842 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
18843 * value-prof.c (init_node_map): Likewise.
18845 2017-05-23 Martin Liska <mliska@suse.cz>
18847 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
18848 and symtab_node::debug_symtab to symbol_table::debug.
18849 * cgraphunit.c (analyze_functions): Use the renamed function.
18850 (symbol_table::compile): Likewise.
18851 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
18852 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
18853 * passes.c (execute_todo): Likewise.
18854 * symtab.c (symbol_table::dump): New function.
18855 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
18857 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18859 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
18860 that nonconst implies exec.
18862 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18864 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18865 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18866 (inline_edge_summary_vec): Turn into ...
18867 (ipa_call_summaries): ... this one.
18868 (redirect_to_unreachable, edge_set_predicate,
18869 evaluate_properties_for_edge, inline_summary_alloc,
18870 reset_ipa_call_summary, reset_inline_summary,
18871 inline_summary_t::duplicate): Update.
18872 (inline_edge_duplication_hook): Turn to ...
18873 (ipa_call_summary_t::duplicate): ... this one.
18874 (inline_edge_removal_hook): Turn to ...
18875 (ipa_call_summary_t::remove): ... this one.
18876 (dump_inline_edge_summary): Turn to ...
18877 (dump_ipa_call_summary): ... this one.
18878 (estimate_function_body_sizes): Update.
18879 (inline_update_callee_summaries): Update.
18880 (remap_edge_change_prob): Update.
18881 (remap_edge_summaries): Update.
18882 (inline_merge_summary): Update.
18883 (do_estimate_edge_time): Update.
18884 (inline_generate_summary): Update.
18885 (inline_read_section): Update.
18886 (inline_read_summary): Update.
18887 (inline_free_summary): Update.
18888 * ipa-inline.c (can_inline_edge_p): Update.
18889 (compute_inlined_call_time): Update.
18890 (want_inline_small_function_p): Update.
18891 (edge_badness): Update.
18892 (early_inliner): Update.
18893 * ipa-inline.h (inline_edge_summary): Turn to ...
18894 (ipa_call_summary): ... this one.
18895 (ipa_call_summary_t): New class.
18896 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18897 (ipa_call_summaries): New.
18898 (inline_edge_summary): Remove.
18899 (estimate_edge_growth): Update.
18900 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18901 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18902 * ipa-split.c (execute_split_functions): Update.
18903 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18905 2017-05-23 Tom de Vries <tom@codesourcery.com>
18907 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18908 attributes): Document rdrand effective target.
18910 2017-05-23 Tom de Vries <tom@codesourcery.com>
18912 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18913 attributes): Sort alphabetically.
18915 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
18917 * config/avr/genmultilib.awk: Use gsub instead of gensub.
18919 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
18922 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
18923 V2DF/V2DI splat into two separate patterns, one that handles
18924 registers, and the other that only handles memory. Drop support
18925 for splatting from a GPR on ISA 2.07 and then splitting the
18926 splat into direct move and splat.
18927 (vsx_splat_<mode>_reg): Likewise.
18928 (vsx_splat_<mode>_mem): Likewise.
18930 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
18932 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
18934 2017-05-22 Jakub Jelinek <jakub@redhat.com>
18936 PR middle-end/80809
18937 * omp-low.c (finish_taskreg_remap): New function.
18938 (finish_taskreg_scan): If unit size of ctx->record_type
18939 is non-constant, unshare the size expression and replace
18940 decls in it with possible outer var refs.
18942 PR middle-end/80809
18943 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
18944 GOVD_SHARED rather than GOVD_PRIVATE with it.
18945 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
18946 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
18948 PR middle-end/80853
18949 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
18950 as last argument to build_outer_var_ref for pointer bases of array
18951 section reductions.
18953 2017-05-19 Martin Sebor <msebor@redhat.com>
18955 * print-tree.c (print_node): Print DECL_READ_P flag.
18957 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18959 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
18960 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
18961 * cgraph.c: Likewise.
18962 * cgraphunit.c: Likewise.
18963 * gengtype.c: Likewise.
18964 * ipa-cp.c: Likewise.
18965 * ipa-devirt.c: Likewise.
18966 * ipa-icf.c: Likewise.
18967 * ipa-predicate.c: Likewise.
18968 * ipa-profile.c: Likewise.
18969 * ipa-prop.c: Likewise.
18970 * ipa-split.c: Likewise.
18972 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
18973 edge_predicate_pool, dump_inline_hints,
18974 inline_summary::account_size_time, redirect_to_unreachable,
18975 edge_set_predicate, set_hint_predicate,
18976 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
18977 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
18978 inline_summary_t::remove, remap_hint_predicate_after_duplication,
18979 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
18980 ipa_call_summary_t::remove, initialize_growth_caches,
18981 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
18982 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
18983 mark_modified, unmodified_parm_1, unmodified_parm,
18984 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
18985 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
18986 compute_bb_predicates, will_be_nonconstant_expr_predicate,
18987 will_be_nonconstant_predicate, record_modified_bb_info,
18988 get_minimal_bb, record_modified, param_change_prob,
18989 phi_result_unknown_predicate, predicate_for_phi_result,
18990 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
18991 estimate_function_body_sizes, compute_inline_parameters,
18992 compute_inline_parameters_for_curren, pass_data_inline_parameters,
18993 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
18994 inline_update_callee_summaries, remap_edge_change_prob,
18995 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
18996 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
18997 inline_analyze_function, inline_summary_t::insert,
18998 inline_generate_summary, read_ipa_call_summary, inline_read_section,
18999 inline_read_summary, write_ipa_call_summary, inline_write_summary,
19000 inline_free_summary): Move to ipa-fnsummary.h
19001 (predicate_t): Remove.
19002 * ipa-fnsummary.c: New file.
19003 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
19004 (enum inline_hints_vals, inline_hints, agg_position_info,
19005 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
19006 inline_summaries, ipa_call_summary, ipa_call_summary_t,
19007 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
19008 dump_inline_summary, dump_inline_hints, inline_generate_summary,
19009 inline_read_summary, inline_write_summary, inline_free_summary,
19010 inline_analyze_function, initialize_inline_failed,
19011 inline_merge_summary, inline_update_overall_summary,
19012 compute_inline_parameters): Move to ipa-fnsummary.h
19013 * ipa-fnsummary.h: New file.
19014 * ipa-inline-transform.h: Include ipa-inline.h.
19015 * ipa-inline.c: LIkewise.
19017 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19019 * ipa-inline.c (edge_badness): Use inlined_time instead of
19020 inline_summaries->get.
19022 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19024 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
19026 2017-05-22 Nathan Sidwell <nathan@acm.org>
19028 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
19029 (fdump-lang): Document 'raw' option.
19030 * dumpfile.h (TDI_tu): Delete.
19031 * dumpfile.c (dump_files): Remove translation-unit.
19032 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
19034 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
19036 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
19037 command option from $(AWK) call.
19038 * config/avr/genmultilib.awk: Simplify and rewrite so that it
19039 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
19040 [FORMAT]: Remove handling of variable.
19041 * config/avr/t-multilib: Regenerate.
19043 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19045 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
19047 (dump_inline_summary): Do not print self_time.
19048 (estimate_function_body_sizes): Do not set self_time.
19049 (compute_inline_parameters): Likewise.
19050 (inline_read_section, inline_write_summary): Do not stream self_time.
19051 * ipa-inline.h (inline_summary): Drop self_time.
19053 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19055 * ipa-inline-analysis.c (account_size_time): Rename to ...
19056 (inline_summary::account_size_time): ... this one.
19057 (reset_ipa_call_summary): Turn to ...
19058 (ipa_call_summary::reset): ... this one.
19059 (reset_inline_summary): Turn to ...
19060 (inline_summary::reset): ... this one.
19061 (inline_summary_t::remove): Update.
19062 (inline_summary_t::duplicate): Update.
19063 (ipa_call_summary_t::remove): Update.
19064 (dump_inline_summary): Update.
19065 (estimate_function_body_sizes): Update.
19066 (compute_inline_parameters): Update.
19067 (estimate_node_size_and_time): Update.
19068 (inline_merge_summary): Update.
19069 (inline_update_overall_summary): Update.
19070 (inline_read_section): Update.
19071 (inline_write_summary): Update.
19072 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
19073 add account_size_time and reset member functions.
19074 (ipa_call_summary): Add reset function.
19075 * ipa-predicate.h (predicate::operator &): Constify.
19077 2017-05-22 Richard Biener <rguenther@suse.de>
19079 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
19081 2017-05-19 Jason Merrill <jason@redhat.com>
19083 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
19085 2017-05-19 Marek Polacek <polacek@redhat.com>
19088 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
19089 TYPE_OVERFLOW_WRAPS checks.
19091 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
19093 * tree-core.h (enum omp_clause_default_kind): Add
19094 "OMP_CLAUSE_DEFAULT_PRESENT".
19095 * tree-pretty-print.c (dump_omp_clause): Handle it.
19096 * gimplify.c (enum gimplify_omp_var_data): Add
19097 "GOVD_MAP_FORCE_PRESENT".
19098 (gimplify_adjust_omp_clauses_1): Map it to
19099 "GOMP_MAP_FORCE_PRESENT".
19100 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
19102 * gimplify.c (oacc_default_clause): Clarify.
19104 2017-05-19 Nathan Sidwell <nathan@acm.org>
19106 LANG_HOOK_REGISTER_DUMPS
19107 * toplev.c (general_init): Call register dump lang hook.
19108 * doc/invoke.texi: Document -fdump-lang option family.
19109 * dumpfile.c (dump_files): Remove class dump here.
19110 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
19111 * dumpfile.h (tree_dump_index): Remove TDI_class.
19112 * langhooks-def.h (lhd_register_dumps): Declare.
19113 (LANG_HOOKS_REGISTER_DUMPS): Define.
19114 (LANG_HOOKS_INITIALIZER): Add it.
19115 * langhooks.c (lhd_register_dumps): Define.
19116 * langhooks.h (struct lang_hooks): Add register_dumps.
19118 2017-05-19 Nathan Sidwell <nathan@acm.org>
19120 * context.h (context::set_passes): New.
19121 * context.c (context::context): Do not create pass manager.
19122 * toplev.c (general_init): Create pass manager here.
19124 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
19126 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
19127 use this splitter if two add or or instructions would also work for
19128 the constant we want to generate.
19130 2017-05-19 Richard Biener <rguenther@suse.de>
19133 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
19134 predicate evaluation.
19136 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19138 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
19140 * ipa-inline.c (want_inline_small_function_p): Do not cast to
19143 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19145 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19146 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19147 (inline_edge_summary_vec): Turn into ...
19148 (ipa_call_summaries): ... this one.
19149 (redirect_to_unreachable, edge_set_predicate,
19150 evaluate_properties_for_edge, inline_summary_alloc,
19151 reset_ipa_call_summary, reset_inline_summary,
19152 inline_summary_t::duplicate): Update.
19153 (inline_edge_duplication_hook): Turn to ...
19154 (ipa_call_summary_t::duplicate): ... this one.
19155 (inline_edge_removal_hook): Turn to ...
19156 (ipa_call_summary_t::remove): ... this one.
19157 (dump_inline_edge_summary): Turn to ...
19158 (dump_ipa_call_summary): ... this one.
19159 (estimate_function_body_sizes): Update.
19160 (inline_update_callee_summaries): Update.
19161 (remap_edge_change_prob): Update.
19162 (remap_edge_summaries): Update.
19163 (inline_merge_summary): Update.
19164 (do_estimate_edge_time): Update.
19165 (inline_generate_summary): Update.
19166 (inline_read_section): Update.
19167 (inline_read_summary): Update.
19168 (inline_free_summary): Update.
19169 * ipa-inline.c (can_inline_edge_p): Update.
19170 (compute_inlined_call_time): Update.
19171 (want_inline_small_function_p): Update.
19172 (edge_badness): Update.
19173 (early_inliner): Update.
19174 * ipa-inline.h (inline_edge_summary): Turn to ...
19175 (ipa_call_summary): ... this one.
19176 (ipa_call_summary_t): New class.
19177 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19178 (ipa_call_summaries): New.
19179 (inline_edge_summary): Remove.
19180 (estimate_edge_growth): Update.
19181 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19182 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19183 * ipa-split.c (execute_split_functions): Update.
19184 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19186 2017-05-19 Richard Biener <rguenther@suse.de>
19188 PR middle-end/80764
19189 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
19191 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
19193 * config/rs6000/rs6000.c (struct machine_function): Add field
19194 fpr_is_wrapped_separately.
19195 (rs6000_get_separate_components): Use 64 components. Handle the
19196 new FPR components.
19197 (rs6000_components_for_bb): Handle the FPR components.
19198 (rs6000_emit_prologue_components): Handle the FPR components.
19199 (rs6000_emit_epilogue_components): Handle the FPR components.
19200 (rs6000_set_handled_components): Handle the FPR components.
19201 (rs6000_emit_prologue): Don't output prologue code for those FPRs
19202 that are already separately shrink-wrapped.
19203 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
19204 that are already separately shrink-wrapped.
19206 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19209 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
19212 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
19213 (define_peephole2 for Altivec d-form load): Add peepholes to catch
19214 cases where the register allocator uses a move and an offsettable
19215 memory operation to/from a FPR register on ISA 2.06/2.07.
19216 (define_peephole2 for Altivec d-form store): Likewise.
19218 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
19221 * config/i386/mmx.md (*mov<mode>_internal): Enable
19222 alternatives 11, 12, 13 and 14 also for 32bit targets.
19223 Remove alternatives 15, 16, 17 and 18.
19224 * config/i386/sse.md (vec_concatv2di): Change
19225 alternative (!x, *y) to (x, ?!*Yn).
19227 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
19229 * dumpfile.h (enum dump_kind): Remove stray comma.
19231 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19233 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
19234 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
19235 predicate::num_conditions
19236 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
19237 (CHANGED): turn into predicate::changed.
19238 (agg_position_info): Move to ipa-predicate.h
19239 (add_condition, predicate::add_clause, predicate::operator &=,
19240 predicate::or_with, predicate::evaluate, predicate::probability,
19241 dump_condition, dump_clause, predicate::dump,
19242 predicate::remap_after_duplication, predicate::remap_after_inlining,
19243 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
19244 (evaluate_conditions_for_known_args): Update.
19245 (set_cond_stmt_execution_predicate): Update.
19246 * ipa-inline.h: Include ipa-predicate.h
19247 (condition, inline_param_summary, conditions, agg_position_info,
19248 predicate): Move to ipa-predicate.h
19249 * ipa-predicate.c: New file.
19250 * ipa-predicate.h: New file.
19252 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
19254 * final.c (leaf_function_p): Check we are not in a sequence.
19256 2017-05-18 Martin Liska <mliska@suse.cz>
19258 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
19259 * dumpfile.c (dump_register): Use new enum dump_kind.
19260 (get_dump_file_name): Likewise.
19261 (dump_enable_all): Likewise.
19262 (dump_switch_p_1): Likewise.
19263 (enable_rtl_dump_file): Remove usage of TDF_RTL.
19264 * dumpfile.h (enum dump_kind): New enum type.
19265 (struct dump_file_info): Create constructor and
19266 format fields and comments.
19267 * passes.c (pass_manager::register_one_dump_file):
19269 * statistics.c (statistics_early_init): Likewise.
19270 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
19271 TDF_TREE with TDF_SLIM.
19272 (gather_memory_references_ref): Likewise.
19274 2017-05-18 Martin Liska <mliska@suse.cz>
19276 * vec.h (struct vnull): Use it.
19278 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19280 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
19281 (true_predicate, false_predicate, true_predicate_p,
19282 false_predicate_p): Remove.
19283 (single_cond_predicate, not_inlined_predicate): Turn to member function
19285 (add_condition): Update.
19286 (add_clause): Turn to...
19287 (predicate::add_clause): ... this one; update; allow passing NULL
19289 (and_predicates): Turn to ...
19290 (predicate::operator &=): ... this one.
19291 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
19292 (or_predicates): Turn to ...
19293 (predicate::or_with): ... this one.
19294 (evaluate_predicate): Turn to ...
19295 (predicate::evaluate): ... this one.
19296 (predicate_probability): Turn to ...
19297 (predicate::probability): ... this one.
19298 (dump_condition): Update.
19299 (dump_predicate): Turn to ...
19300 (predicate::dump): ... this one.
19301 (account_size_time): Update.
19302 (edge_set_predicate): Update.
19303 (set_hint_predicate): UPdate.
19304 (evaluate_conditions_for_known_args): Update.
19305 (evaluate_properties_for_edge): Update.
19306 (remap_predicate_after_duplication): Turn to...
19307 (predicate::remap_after_duplication): ... this one.
19308 (remap_hint_predicate_after_duplication): Update.
19309 (inline_summary_t::duplicate): UPdate.
19310 (dump_inline_edge_summary): Update.
19311 (dump_inline_summary): Update.
19312 (set_cond_stmt_execution_predicate): Update.
19313 (set_switch_stmt_execution_predicate): Update.
19314 (compute_bb_predicates): Update.
19315 (will_be_nonconstant_expr_predicate): Update.
19316 (will_be_nonconstant_predicate): Update.
19317 (phi_result_unknown_predicate): Update.
19318 (predicate_for_phi_result): Update.
19319 (array_index_predicate): Update.
19320 (estimate_function_body_sizes): Update.
19321 (estimate_node_size_and_time): Update.
19322 (estimate_ipcp_clone_size_and_time): Update.
19323 (remap_predicate): Rename to ...
19324 (predicate::remap_after_inlining): ... this one.
19325 (remap_hint_predicate): Update.
19326 (inline_merge_summary): Update.
19327 (inline_update_overall_summary): Update.
19328 (estimate_size_after_inlining): Update.
19329 (read_predicate): Rename to ...
19330 (predicate::stream_in): ... this one.
19331 (read_inline_edge_summary): Update.
19332 (write_predicate): Rename to ...
19333 (predicate::stream_out): ... this one.
19334 (write_inline_edge_summary): Update.
19335 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
19336 (clause_t): Turn to uint32_t
19337 (predicate): Turn to class; implement constructor and operators
19339 (size_time_entry): Update.
19340 (inline_summary): Update.
19341 (inline_edge_summary): Update.
19343 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
19345 * fold-const.c (fold_binary_loc): Move transformation...
19346 * match.pd (C - X CMP X): ... here.
19348 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
19350 * config/sparc/sparc.c (sparc_option_override): Set function
19351 alignment for -mcpu=niagara7 to 64 to match the I$ line.
19352 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
19354 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
19356 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
19358 2017-05-18 Marek Polacek <polacek@redhat.com>
19361 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
19362 (pass_ubsan::execute): Call gimple_assign_single_p instead of
19363 gimple_assign_load_p.
19365 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
19367 PR middle-end/80692
19368 * real.c (do_compare): Give decimal_do_compare preference over
19369 comparing just the signs.
19371 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
19373 * doc/md.texi (Canonicalization of Instructions): Describe the
19374 canonical form of instructions that inherently set a condition
19377 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
19379 PR middle-end/80775
19380 * tree-cfg.c: Move deletion of unreachable case statements to after
19381 the merging of consecutive case labels.
19383 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
19385 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
19386 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
19387 restoring of callee-saved registers.
19389 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
19391 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
19392 * config/visium/visium.c (single_set_and_flags): Likewise.
19393 * config/visium/visium.md (Substitutions): Likewise.
19395 2017-05-17 Martin Liska <mliska@suse.cz>
19397 * cfg.c: Introduce dump_flags_t type and
19398 use it instead of int type.
19400 * cfghooks.c: Likewise.
19401 * cfghooks.h (struct cfg_hooks): Likewise.
19402 * cfgrtl.c: Likewise.
19403 * cfgrtl.h: Likewise.
19404 * cgraph.c (cgraph_node::get_body): Likewise.
19405 * coretypes.h: Likewise.
19406 * domwalk.c: Likewise.
19407 * domwalk.h: Likewise.
19408 * dumpfile.c (struct dump_option_value_info): Likewise.
19409 (dump_enable_all): Likewise.
19410 (dump_switch_p_1): Likewise.
19411 (opt_info_switch_p): Likewise.
19412 * dumpfile.h (enum tree_dump_index): Likewise.
19413 (struct dump_file_info): Likewise.
19414 * genemit.c: Likewise.
19415 * generic-match-head.c: Likewise.
19416 * gengtype.c (open_base_files): Likewise.
19417 * gimple-pretty-print.c: Likewise.
19418 * gimple-pretty-print.h: Likewise.
19419 * graph.c (print_graph_cfg): Likewise.
19420 * graphite-scop-detection.c (dot_all_sese): Likewise.
19421 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
19422 * loop-unroll.c (report_unroll): Likewise.
19423 * passes.c (pass_manager::register_one_dump_file): Likewise.
19424 * print-tree.c: Likewise.
19425 * statistics.c: Likewise.
19426 * tree-cfg.c: Likewise.
19427 * tree-cfg.h: Likewise.
19428 * tree-dfa.c: Likewise.
19429 * tree-dfa.h: Likewise.
19430 * tree-dump.c (dump_function): Likewise.
19431 * tree-dump.h (struct dump_info): Likewise.
19432 * tree-pretty-print.c: Likewise.
19433 * tree-pretty-print.h: Likewise.
19434 * tree-ssa-live.c: Likewise.
19435 * tree-ssa-live.h: Likewise.
19436 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
19437 * tree-vect-loop.c: Likewise.
19438 * tree-vect-slp.c: Likewise.
19440 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
19441 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19443 PR tree-optimization/80457
19444 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
19445 of all arguments to a statement as scalar_to_vec operations.
19446 (vectorizable_call): Adjust call to vect_model_simple_cost for
19448 (vectorizable_conversion): Likewise.
19449 (vectorizable_assignment): Likewise.
19450 (vectorizable_shift): Likewise.
19451 (vectorizable_operation): Likewise.
19452 (vectorizable_comparison): Likewise.
19453 (vect_is_simple_cond): Record the def types for operands.
19454 (vectorizable_condition): Likewise, call vect_model_simple_cost.
19455 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
19456 for statement argument count.
19458 2017-05-16 Carl Love <cel@us.ibm.com>
19460 * config/rs6000/rs6000-c: Add support for built-in functions
19461 vector unsigned long long vec_bperm (vector unsigned long long,
19462 vector unsigned char)
19463 vector signed long long vec_mule (vector signed int,
19465 vector unsigned long long vec_mule (vector unsigned int,
19466 vector unsigned int)
19467 vector signed long long vec_mulo (vector signed int,
19469 vector unsigned long long vec_mulo (vector unsigned int,
19470 vector unsigned int)
19471 vector signed char vec_sldw (vector signed char,
19472 vector signed char,
19474 vector unsigned char vec_sldw (vector unsigned char,
19475 vector unsigned char,
19477 vector signed short vec_sldw (vector signed short,
19478 vector signed short,
19480 vector unsigned short vec_sldw (vector unsigned short,
19481 vector unsigned short,
19483 vector signed int vec_sldw (vector signed int,
19486 vector unsigned int vec_sldw (vector unsigned int,
19487 vector unsigned int,
19489 vector signed long long vec_sldw (vector signed long long,
19490 vector signed long long,
19492 vector unsigned long long vec_sldw (vector unsigned long long,
19493 vector unsigned long long,
19495 * config/rs6000/rs6000-c: Add support for built-in functions
19496 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
19497 * config/rs6000/altivec.h: Add defintion for vec_sldw.
19498 * doc/extend.texi: Update the built-in documentation for the
19499 new built-in functions.
19501 2017-05-16 Marek Polacek <polacek@redhat.com>
19505 * tree.c (save_expr): Don't fold the expression.
19507 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
19509 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
19510 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
19511 and (?*y,m). Update insn attributes.
19513 2017-05-16 Martin Liska <mliska@suse.cz>
19515 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
19516 flags argument of print_gimple_stmt, print_gimple_expr,
19517 print_generic_stmt and print_generic_expr.
19518 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
19519 * coretypes.h: Likewise.
19520 * except.c (dump_eh_tree): Likewise.
19521 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
19522 * gimple-pretty-print.h: Likewise.
19523 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
19524 (backprop::push_to_worklist): Likewise.
19525 (backprop::pop_from_worklist): Likewise.
19526 (backprop::process_use): Likewise.
19527 (backprop::intersect_uses): Likewise.
19528 (note_replacement): Likewise.
19529 * gimple-ssa-store-merging.c
19530 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
19531 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
19532 (pass_store_merging::execute): Likewise.
19533 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
19534 (ssa_base_cand_dump_callback): Likewise.
19535 (dump_incr_vec): Likewise.
19536 (replace_refs): Likewise.
19537 (replace_mult_candidate): Likewise.
19538 (create_add_on_incoming_edge): Likewise.
19539 (create_phi_basis): Likewise.
19540 (insert_initializers): Likewise.
19541 (all_phi_incrs_profitable): Likewise.
19542 (introduce_cast_before_cand): Likewise.
19543 (replace_one_candidate): Likewise.
19544 * gimplify.c (gimplify_expr): Likewise.
19545 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
19546 (set_rename): Likewise.
19547 (rename_uses): Likewise.
19548 (copy_loop_phi_nodes): Likewise.
19549 (add_close_phis_to_merge_points): Likewise.
19550 (copy_loop_close_phi_args): Likewise.
19551 (copy_cond_phi_args): Likewise.
19552 (graphite_copy_stmts_from_block): Likewise.
19553 (translate_pending_phi_nodes): Likewise.
19554 * graphite-poly.c (print_pdr): Likewise.
19555 (dump_gbb_cases): Likewise.
19556 (dump_gbb_conditions): Likewise.
19557 (print_scop_params): Likewise.
19558 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
19559 (build_cross_bb_scalars_use): Likewise.
19560 (gather_bbs::before_dom_children): Likewise.
19561 * hsa-dump.c (dump_hsa_immed): Likewise.
19562 * ipa-cp.c (print_ipcp_constant_value): Likewise.
19563 (get_replacement_map): Likewise.
19564 * ipa-inline-analysis.c (dump_condition): Likewise.
19565 (estimate_function_body_sizes): Likewise.
19566 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
19567 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
19568 * ipa-prop.c (ipa_dump_param): Likewise.
19569 (ipa_print_node_jump_functions_for_edge): Likewise.
19570 (ipa_modify_call_arguments): Likewise.
19571 (ipa_modify_expr): Likewise.
19572 (ipa_dump_param_adjustments): Likewise.
19573 (ipa_dump_agg_replacement_values): Likewise.
19574 (ipcp_modif_dom_walker::before_dom_children): Likewise.
19575 * ipa-pure-const.c (check_stmt): Likewise.
19576 (pass_nothrow::execute): Likewise.
19577 * ipa-split.c (execute_split_functions): Likewise.
19578 * omp-offload.c (dump_oacc_loop_part): Likewise.
19579 (dump_oacc_loop): Likewise.
19580 * trans-mem.c (tm_log_emit): Likewise.
19581 (tm_memopt_accumulate_memops): Likewise.
19582 (dump_tm_memopt_set): Likewise.
19583 (dump_tm_memopt_transform): Likewise.
19584 * tree-cfg.c (gimple_verify_flow_info): Likewise.
19585 (print_loop): Likewise.
19586 * tree-chkp-opt.c (chkp_print_addr): Likewise.
19587 (chkp_gather_checks_info): Likewise.
19588 (chkp_get_check_result): Likewise.
19589 (chkp_remove_check_if_pass): Likewise.
19590 (chkp_use_outer_bounds_if_possible): Likewise.
19591 (chkp_reduce_bounds_lifetime): Likewise.
19592 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
19593 (chkp_mark_completed_bounds): Likewise.
19594 (chkp_register_incomplete_bounds): Likewise.
19595 (chkp_mark_invalid_bounds): Likewise.
19596 (chkp_maybe_copy_and_register_bounds): Likewise.
19597 (chkp_build_returned_bound): Likewise.
19598 (chkp_get_bound_for_parm): Likewise.
19599 (chkp_build_bndldx): Likewise.
19600 (chkp_get_bounds_by_definition): Likewise.
19601 (chkp_generate_extern_var_bounds): Likewise.
19602 (chkp_get_bounds_for_decl_addr): Likewise.
19603 * tree-chrec.c (chrec_apply): Likewise.
19604 * tree-data-ref.c (dump_data_reference): Likewise.
19605 (dump_subscript): Likewise.
19606 (dump_data_dependence_relation): Likewise.
19607 (analyze_overlapping_iterations): Likewise.
19608 * tree-inline.c (expand_call_inline): Likewise.
19609 (tree_function_versioning): Likewise.
19610 * tree-into-ssa.c (dump_defs_stack): Likewise.
19611 (dump_currdefs): Likewise.
19612 (dump_names_replaced_by): Likewise.
19613 (dump_update_ssa): Likewise.
19614 (update_ssa): Likewise.
19615 * tree-object-size.c (pass_object_sizes::execute): Likewise.
19616 * tree-parloops.c (build_new_reduction): Likewise.
19617 (try_create_reduction_list): Likewise.
19618 (ref_conflicts_with_region): Likewise.
19619 (oacc_entry_exit_ok_1): Likewise.
19620 (oacc_entry_exit_single_gang): Likewise.
19621 * tree-pretty-print.h: Likewise.
19622 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
19623 (get_scalar_evolution): Likewise.
19624 (add_to_evolution): Likewise.
19625 (get_loop_exit_condition): Likewise.
19626 (analyze_evolution_in_loop): Likewise.
19627 (analyze_initial_condition): Likewise.
19628 (analyze_scalar_evolution): Likewise.
19629 (instantiate_scev): Likewise.
19630 (number_of_latch_executions): Likewise.
19631 (gather_chrec_stats): Likewise.
19632 (final_value_replacement_loop): Likewise.
19633 (scev_const_prop): Likewise.
19634 * tree-sra.c (dump_access): Likewise.
19635 (disqualify_candidate): Likewise.
19636 (create_access): Likewise.
19637 (reject): Likewise.
19638 (maybe_add_sra_candidate): Likewise.
19639 (create_access_replacement): Likewise.
19640 (analyze_access_subtree): Likewise.
19641 (analyze_all_variable_accesses): Likewise.
19642 (sra_modify_assign): Likewise.
19643 (initialize_constant_pool_replacements): Likewise.
19644 (find_param_candidates): Likewise.
19645 (decide_one_param_reduction): Likewise.
19646 (replace_removed_params_ssa_names): Likewise.
19647 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
19648 * tree-ssa-copy.c (dump_copy_of): Likewise.
19649 (copy_prop_visit_cond_stmt): Likewise.
19650 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
19651 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
19652 (record_equivalences_from_stmt): Likewise.
19653 * tree-ssa-dse.c (compute_trims): Likewise.
19654 (delete_dead_call): Likewise.
19655 (delete_dead_assignment): Likewise.
19656 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
19657 (forward_propagate_into_cond): Likewise.
19658 (pass_forwprop::execute): Likewise.
19659 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
19660 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
19662 (move_computations_worker): Likewise.
19663 (execute_sm): Likewise.
19664 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
19665 (remove_exits_and_undefined_stmts): Likewise.
19666 (remove_redundant_iv_tests): Likewise.
19667 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
19668 (adjust_iv_update_pos): Likewise.
19669 * tree-ssa-math-opts.c (bswap_replace): Likewise.
19670 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
19671 (value_replacement): Likewise.
19672 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
19673 * tree-ssa-pre.c (print_pre_expr): Likewise.
19674 (get_representative_for): Likewise.
19675 (create_expression_by_pieces): Likewise.
19676 (insert_into_preds_of_block): Likewise.
19677 (eliminate_insert): Likewise.
19678 (eliminate_dom_walker::before_dom_children): Likewise.
19679 (eliminate): Likewise.
19680 (remove_dead_inserted_code): Likewise.
19681 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
19682 * tree-ssa-reassoc.c (get_rank): Likewise.
19683 (eliminate_duplicate_pair): Likewise.
19684 (eliminate_plus_minus_pair): Likewise.
19685 (eliminate_not_pairs): Likewise.
19686 (undistribute_ops_list): Likewise.
19687 (eliminate_redundant_comparison): Likewise.
19688 (update_range_test): Likewise.
19689 (optimize_range_tests_var_bound): Likewise.
19690 (optimize_vec_cond_expr): Likewise.
19691 (rewrite_expr_tree): Likewise.
19692 (rewrite_expr_tree_parallel): Likewise.
19693 (linearize_expr): Likewise.
19694 (break_up_subtract): Likewise.
19695 (linearize_expr_tree): Likewise.
19696 (attempt_builtin_powi): Likewise.
19697 (attempt_builtin_copysign): Likewise.
19698 (transform_stmt_to_copy): Likewise.
19699 (transform_stmt_to_multiply): Likewise.
19700 (dump_ops_vector): Likewise.
19701 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
19702 (print_scc): Likewise.
19703 (set_ssa_val_to): Likewise.
19704 (visit_reference_op_store): Likewise.
19705 (visit_use): Likewise.
19706 (sccvn_dom_walker::before_dom_children): Likewise.
19707 (run_scc_vn): Likewise.
19708 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
19710 (expr_hash_elt::print): Likewise.
19711 (const_and_copies::pop_to_marker): Likewise.
19712 (const_and_copies::record_const_or_copy_raw): Likewise.
19713 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
19714 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
19715 (dump_predicates): Likewise.
19716 (find_uninit_use): Likewise.
19717 (warn_uninitialized_phi): Likewise.
19718 (pass_late_warn_uninitialized::execute): Likewise.
19719 * tree-ssa.c (verify_vssa): Likewise.
19720 (verify_ssa): Likewise.
19721 (maybe_optimize_var): Likewise.
19722 * tree-vrp.c (dump_value_range): Likewise.
19723 (dump_all_value_ranges): Likewise.
19724 (dump_asserts_for): Likewise.
19725 (register_edge_assert_for_2): Likewise.
19726 (vrp_visit_cond_stmt): Likewise.
19727 (vrp_visit_switch_stmt): Likewise.
19728 (vrp_visit_stmt): Likewise.
19729 (vrp_visit_phi_node): Likewise.
19730 (simplify_cond_using_ranges_1): Likewise.
19731 (fold_predicate_in): Likewise.
19732 (evrp_dom_walker::before_dom_children): Likewise.
19733 (evrp_dom_walker::push_value_range): Likewise.
19734 (evrp_dom_walker::pop_value_range): Likewise.
19735 (execute_early_vrp): Likewise.
19737 2017-05-16 Richard Biener <rguenther@suse.de>
19739 * dwarf2out.c (loc_list_from_tree_1): Do not create
19740 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
19742 2017-05-16 Richard Biener <rguenther@suse.de>
19744 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
19746 (note_variable_value_in_expr): If we resolved the decl ref
19747 do not push to the stack.
19749 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
19751 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
19752 operations in fast-math mode.
19753 (vaddq_f16): Likewise.
19754 (vmul_f16): Likewise.
19755 (vmulq_f16): Likewise.
19756 (vsub_f16): Likewise.
19757 (vsubq_f16): Likewise.
19758 * config/arm/neon.md (add<mode>3): New.
19760 (fma:<VH:mode>3): New. Also remove outdated comment.
19763 2017-05-16 Martin Liska <mliska@suse.cz>
19767 * ipa-devirt.c (warn_types_mismatch): Fix typo.
19768 (odr_types_equivalent_p): Likewise.
19770 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
19772 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
19774 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
19777 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
19778 non-interunit SSE move alternatives with '?'.
19779 (zero-extendsidi peephole2): New peephole to skip intermediate
19780 general register in SSE zero-extend sequence.
19782 2017-05-15 Jeff Law <law@redhat.com>
19784 * reorg.c (relax_delay_slots): Create a new variable to hold
19785 the temporary target rather than clobbering TARGET_LABEL.
19787 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
19788 missing argument to extract_bit_field call.
19789 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
19791 2017-05-15 Martin Liska <mliska@suse.cz>
19794 * gcc.c (process_command): Do not allow empty argument of -o option.
19796 2017-05-15 Renlin Li <renlin.li@arm.com>
19798 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
19799 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
19800 * config/aarch64/constraints.md (Usf): Add long call check.
19801 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
19802 (call_value): Likewise.
19803 (sibcall): Likewise.
19804 (sibcall_value): Likewise.
19806 (call_value_insn): New.
19807 (sibcall_insn): Update rtx pattern.
19808 (sibcall_value_insn): Likewise.
19809 (call_internal): Remove.
19810 (call_value_internal): Likewise.
19811 (sibcall_internal): Likewise.
19812 (sibcall_value_internal): Likewise.
19813 (call_reg): Likewise.
19814 (call_symbol): Likewise.
19815 (call_value_reg): Likewise.
19816 (call_value_symbol): Likewise.
19818 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
19821 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
19823 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
19825 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
19826 compatible with CCGOCmode and with CCZmode.
19828 2017-05-14 Martin Sebor <msebor@redhat.com>
19830 PR middle-end/77671
19831 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
19832 (gimple_fold_builtin_snprintf): Same.
19833 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
19834 (gimple_fold_builtin_snprintf): Same.
19835 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
19836 of character types.
19837 (is_call_safe): New function.
19838 (try_substitute_return_value): Call it.
19839 (try_simplify_call): New function.
19840 (pass_sprintf_length::handle_gimple_call): Call it.
19842 2017-05-14 Martin Sebor <msebor@redhat.com>
19844 PR middle-end/80669
19845 * builtins.c (expand_builtin_stpncpy): Simplify.
19847 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
19849 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
19850 * config/i386/i386.h
19851 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
19852 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
19853 (struct machine_function): Add new members call_ms2sysv,
19854 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
19855 (struct machine_frame_state): New fields sp_realigned and
19856 sp_realigned_offset.
19857 * config/i386/i386.c
19858 (enum xlogue_stub): New enum.
19859 (enum xlogue_stub_sets): New enum.
19860 (class xlogue_layout): New class.
19861 (struct ix86_frame): New fields stack_realign_allocate_offset,
19862 stack_realign_offset and outlined_save_offset. Modify comments to
19863 detail stack layout when using out-of-line stubs.
19864 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
19865 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
19866 -mcall-ms2sysv-xlogues.
19867 (stub_managed_regs): New static variable.
19868 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
19869 registers managed by out-of-line stub.
19870 (disable_call_ms2sysv_xlogues): New function.
19871 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
19872 m->call_ms2sysv when appropriate and compute frame layout for
19874 (sp_valid_at, fp_valid_at): New inline functions.
19875 (choose_basereg): New function.
19876 (choose_baseaddr): Add align parameter, use choose_basereg and modify
19878 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
19879 Use align parameter of choose_baseaddr to generated aligned SSE movs
19881 (pro_epilogue_adjust_stack): Modify to track
19882 machine_frame_state::sp_realigned.
19883 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
19884 (ix86_nsaved_sseregs): Likewise.
19885 (ix86_emit_save_regs): Likewise.
19886 (ix86_emit_save_regs_using_mov): Likewise.
19887 (ix86_emit_save_sse_regs_using_mov): Likewise.
19888 (get_scratch_register_on_entry): Likewise.
19889 (gen_frame_set): New function.
19890 (gen_frame_load): Likewise.
19891 (gen_frame_store): Likewise.
19892 (emit_outlined_ms2sysv_save): Likewise.
19893 (emit_outlined_ms2sysv_restore): Likewise.
19894 (ix86_expand_prologue): Modify stack re-alignment code and call
19895 emit_outlined_ms2sysv_save when appropriate.
19896 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
19897 parameter rtx_insn *insn, which allows the function to be used to only
19898 generate the notes.
19899 (ix86_expand_epilogue): Modify validity checks of frame and stack
19900 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
19901 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
19902 * config/i386/predicates.md
19903 (save_multiple): New predicate.
19904 (restore_multiple): Likewise.
19905 * config/i386/sse.md
19906 (save_multiple<mode>): New pattern.
19907 (save_multiple_realign<mode>): Likewise.
19908 (restore_multiple<mode>): Likewise.
19909 (restore_multiple_and_return<mode>): Likewise.
19910 (restore_multiple_leave_return<mode>): Likewise.
19911 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
19913 2017-05-14 Julia Koval <julia.koval@intel.com>
19915 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
19916 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
19917 (__builtin_ia32_xsetbv): New builtins.
19918 * config/i386/i386.c (ix86_expand_special_args_builtin):
19920 (ix86_expand_builtin): Special expand for new intrinsics.
19921 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
19922 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
19923 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
19925 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19927 * cfganal.c (inverted_post_order_compute): Change argument type
19929 * cfganal.h (inverted_post_order_compute): Adjust prototype.
19930 * df-core.c (rest_of_handle_df_initialize): Adjust.
19931 (rest_of_handle_df_finish): Likewise.
19932 (df_analyze_1): Likewise.
19933 (df_analyze): Likewise.
19934 (loop_inverted_post_order_compute): Change argument to be a vec *.
19935 (df_analyze_loop): Adjust.
19936 (df_get_n_blocks): Likewise.
19937 (df_get_postorder): Likewise.
19938 * df.h (struct df_d): Change field to be a vec.
19939 * lcm.c (compute_laterin): Adjust.
19940 (compute_available): Likewise.
19941 * lra-lives.c (lra_create_live_ranges_1): Likewise.
19942 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
19943 * tree-ssa-pre.c (compute_antic): Likewise.
19945 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19947 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
19948 (depth_first_search::depth_first_search): Change structure init
19949 function to this constructor.
19950 (depth_first_search::add_bb): Rename function to this member.
19951 (depth_first_search::execute): Likewise.
19952 (flow_dfs_compute_reverse_finish): Adjust.
19954 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19956 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
19957 (longest_simple_path): Likewise.
19958 * shrink-wrap.c (spread_components): Likewise.
19959 (disqualify_problematic_components): Likewise.
19960 (emit_common_heads_for_components): Likewise.
19961 (emit_common_tails_for_components): Likewise.
19962 (insert_prologue_epilogue_for_components): Likewise.
19964 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19966 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
19969 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19971 * df-core.c (df_set_blocks): Start using auto_bitmap.
19972 (df_compact_blocks): Likewise.
19973 * df-problems.c (df_rd_confluence_n): Likewise.
19974 * df-scan.c (df_insn_rescan_all): Likewise.
19975 (df_process_deferred_rescans): Likewise.
19976 (df_update_entry_block_defs): Likewise.
19977 (df_update_exit_block_uses): Likewise.
19978 (df_entry_block_bitmap_verify): Likewise.
19979 (df_exit_block_bitmap_verify): Likewise.
19980 (df_scan_verify): Likewise.
19981 * lra-constraints.c (lra_constraints): Likewise.
19982 (undo_optional_reloads): Likewise.
19983 (lra_undo_inheritance): Likewise.
19984 * lra-remat.c (calculate_gen_cands): Likewise.
19985 (do_remat): Likewise.
19986 * lra-spills.c (assign_spill_hard_regs): Likewise.
19987 (spill_pseudos): Likewise.
19988 * tree-ssa-pre.c (bitmap_set_and): Likewise.
19989 (bitmap_set_subtract_values): Likewise.
19991 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19993 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
19994 management with auto_bitmap.
19995 (fix_inter_tick): Likewise.
19996 (fix_recovery_deps): Likewise.
19997 * ira.c (add_store_equivs): Likewise.
19998 (find_moveable_pseudos): Likewise.
19999 (split_live_ranges_for_shrink_wrap): Likewise.
20000 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
20001 (rtx_reuse_manager::seen_def_p): Likewise.
20002 (rtx_reuse_manager::set_seen_def): Likewise.
20003 * print-rtl.h (class rtx_reuse_manager): Likewise.
20005 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20007 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
20009 (migrate_btr_def): Likewise.
20010 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
20011 * df-core.c (loop_post_order_compute): Likewise.
20012 (loop_inverted_post_order_compute): Likewise.
20013 * hsa-common.h: Likewise.
20014 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
20015 * init-regs.c (initialize_uninitialized_regs): Likewise.
20016 * ipa-inline.c (resolve_noninline_speculation): Likewise.
20017 (inline_small_functions): Likewise.
20018 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
20019 * ira.c (combine_and_move_insns): Likewise.
20020 (build_insn_chain): Likewise.
20021 * loop-invariant.c (find_invariants): Likewise.
20022 * lower-subreg.c (propagate_pseudo_copies): Likewise.
20023 * predict.c (tree_predict_by_opcode): Likewise.
20024 (predict_paths_leading_to): Likewise.
20025 (predict_paths_leading_to_edge): Likewise.
20026 (estimate_loops_at_level): Likewise.
20027 (estimate_loops): Likewise.
20028 * shrink-wrap.c (try_shrink_wrapping): Likewise.
20029 (spread_components): Likewise.
20030 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
20031 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
20032 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
20033 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
20034 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
20035 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
20036 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
20037 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
20038 (mark_threaded_blocks): Likewise.
20039 (thread_through_all_blocks): Likewise.
20040 * tree-ssa.c (verify_ssa): Likewise.
20041 (execute_update_addresses_taken): Likewise.
20042 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
20044 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20046 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
20048 (post_order_compute): Likewise.
20049 (inverted_post_order_compute): Likewise.
20050 (pre_and_rev_post_order_compute_fn): Likewise.
20052 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20054 * genrecog.c (int_set::int_set): Explicitly construct our
20055 auto_vec base class.
20056 * vec.h (auto_vec::auto_vec): New constructor.
20058 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20060 * bitmap.h (class auto_bitmap): New constructor taking
20061 bitmap_obstack * argument.
20063 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20065 * bitmap.h (class auto_bitmap): Change type of m_bits to
20066 bitmap_head, and adjust ctor / dtor and member operators.
20068 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
20070 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
20071 when returned register mode doesn't match original mode.
20073 2017-05-12 Jeff Law <law@redhat.com>
20074 Jakub Jelinek <jakub@redhat.com>
20076 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
20077 we look for cc setter after the compare-elim changes.
20078 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
20079 within the vector to match what compare-elim now expects.
20080 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
20081 (xorsi3_flags, one_cmplsi2_flags): Likewise.
20083 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
20084 after the compare-elim changes.
20085 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
20086 the vector to match what compare-elim now expects.
20087 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
20088 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
20089 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
20090 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
20091 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
20093 * config/visium/visium.c (single_set_and_flags): Fix where
20094 we look for cc setter after the compare-elim changes.
20095 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
20096 with the vector to match what compare-elim now expects.
20097 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
20098 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
20099 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
20100 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
20101 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
20102 (neg<mode>2_insn_set_overflow): Likewise.
20104 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
20106 PR middle-end/79794
20107 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
20108 maybe_expand_insn call, set ops[0].target. If still set after call,
20109 set alt_rtl. Add extra arg to recursive calls.
20110 (extract_bit_field): Add alt_rtl argument. Pass to
20112 * expmed.h (extract_bit_field): Fix prototype.
20113 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
20114 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
20115 to extract_bit_field_calls.
20116 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
20117 Pass alt_rtl to extract_bit_field calls.
20118 * calls.c (store_unaligned_arguments_into_psuedos)
20119 load_register_parameters): Pass extra NULL to extract_bit_field calls.
20120 * optabs.c (maybe_legitimize_operand): Clear op->target when call
20122 * optabs.h (struct expand_operand): Add target bitfield.
20124 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20126 * compare-elim.c (try_eliminate_compare): Canonicalize
20127 operation with embedded compare to
20128 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
20129 (set (reg) (operation)].
20131 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
20133 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20136 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
20137 cost of adding a carry flag for ADC instruction.
20138 [case MINUS]: Ignore the cost of subtracting a carry flag
20139 for SBB instruction.
20141 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
20143 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
20145 * config/rs6000/bmiintrin.h: New file.
20146 * config/rs6000/bmi2intrin.h: New file.
20147 * config/rs6000/x86intrin.h: New file.
20149 2017-05-12 Jeff Law <law@redhat.com>
20151 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
20154 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
20156 PR middle-end/80707
20157 * tree-cfg.c: Remove cfg edges of unreachable case statements.
20159 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20161 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20162 early expansion of vector divide builtins.
20163 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
20164 builtins identified as having unsigned arguments.
20166 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20168 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
20169 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
20170 expansion of vector logical operations (and, andc, or, xor,
20173 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20175 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
20176 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
20178 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20180 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20181 early GIMPLE expansion of vector multiplies.
20183 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20185 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
20186 TARGET_HAVE_MOVT conditional.
20187 (movt splitter): Likewise.
20189 2017-05-12 Richard Biener <rguenther@suse.de>
20191 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
20192 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20193 Fold all stmts not inplace.
20195 2017-05-12 Richard Biener <rguenther@suse.de>
20197 PR tree-optimization/80713
20198 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
20199 inserted_exprs bit for not removed stmts.
20201 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
20203 PR middle-end/69921
20204 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
20205 parallelized" attribute for parallelized OpenACC kernels.
20206 * omp-offload.c (execute_oacc_device_lower): Use it.
20208 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
20209 Set "oacc kernels" attribute.
20210 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
20211 parameter. Adjust all users.
20212 (oacc_fn_attrib_kernels_p): Remove function.
20213 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
20214 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
20215 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
20216 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
20217 assert "oacc kernels" attribute is set.
20219 2017-05-11 Carl Love <cel@us.ibm.com>
20221 * config/rs6000/rs6000-c: Add support for built-in functions
20222 vector unsigned char vec_popcnt (vector signed char)
20223 vector unsigned char vec_popcnt (vector unsigned char)
20224 vector unsigned short vec_popcnt (vector signed short)
20225 vector unsigned short vec_popcnt (vector unsigned short)
20226 vector unsigned int vec_popcnt (vector signed int)
20227 vector unsigned int vec_popcnt (vector unsigned int)
20228 vector unsigned long long vec_popcnt (vector signed long long)
20229 vector unsigned long long vec_popcnt (vector unsigned long long)
20230 vector signed long long vec_slo (vector signed long long,
20231 vector signed char)
20232 vector signed long long vec_slo (vector signed long long,
20233 vector unsigned char)
20234 vector unsigned long long vec_slo (vector unsigned long long,
20235 vector signed char)
20236 vector unsigned long long vec_slo (vector unsigned long long,
20237 vector unsigned char)
20238 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
20239 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
20240 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
20241 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
20242 * doc/extend.texi: Update the built-in documentation file for the
20243 new built-in functions.
20245 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20247 * attribs.h (sorted_attr_string): Move machine independent
20248 functions for target clone support from the i386 port to common
20249 code. Rename ix86_function_versions to common_function_versions.
20250 Rename make_name to make_unique_name.
20251 (common_function_versions): Likewise.
20252 (make_unique_name): Likewise.
20253 (make_dispatcher_decl): Likewise.
20254 (is_function_default_version): Likewise.
20255 * attribs.c (attr_strcmp): Likewise.
20256 (sorted_attr_string): Likewise.
20257 (common_function_versions): Likewise.
20258 (make_unique_name): Likewise.
20259 (make_dispatcher_decl): Likewise.
20260 (is_function_default_version): Likewise.
20261 * config/i386/i386.c (attr_strcmp): Likewise.
20262 (sorted_attr_string): Likewise.
20263 (ix86_function_versions): Likewise.
20264 (make_name): Likewise.
20265 (make_dispatcher_decl): Likewise.
20266 (is_function_default_version): Likewise.
20267 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
20269 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20272 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20273 Account for direct move costs for vec_construct of integer
20276 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
20279 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
20280 (UNSPEC_STX_ATOMIC): Ditto.
20281 (loaddi_via_sse): New insn.
20282 (storedi_via_sse): Ditto.
20283 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
20284 Update corresponding peephole2 patterns.
20285 (atomic_storedi_fpu): Ditto.
20287 2017-05-11 Julia Koval <julia.koval@intel.com>
20289 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
20290 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
20292 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
20293 (__builtin_ia32_rsqrt14ss_mask): New builtins.
20294 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
20296 2017-05-11 Nathan Sidwell <nathan@acm.org>
20298 * graphite-poly.c: Include dumpfile.h.
20300 * dumpfle.h (dump_function): Declare here ...
20301 * tree-dump.h (dump_function): ... not here.
20302 * dumpfile.c: #include tree-cfg.h.
20303 (dump_function): Move here from ...
20304 * tree-dump.c (dump_function): ... here.
20305 * gimplify.c: #include splay-tree.h, not tree-dump.h.
20306 * graphite-poly.c: Don't include tree-dump.h.
20307 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
20308 * print-tree.c: Likewise.
20309 * stor-layout.c: Likewise.
20310 * tree-nested.c: Likewise.
20312 * dumpfile.c (dump_start): Use TDF_FLAGS.
20313 (dump_enable_all): Fix TDF_KIND check thinko.
20315 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
20317 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20318 array entries to represent two legal parameterizations of the
20319 overloaded __builtin_cmpb function, as represented by the
20320 P6_OV_BUILTIN_CMPB constant.
20321 (altivec_resolve_overloaded_builtin): Add special case handling
20322 for the __builtin_cmpb function, as represented by the
20323 P6_OV_BUILTIN_CMPB constant.
20324 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
20325 (BU_P6_64BIT_2): New macro.
20326 (BU_P6_OVERLOAD_2): New macro
20327 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
20328 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
20329 (CMPB): Add overload support to represent both 32-bit and 64-bit
20330 compare-bytes function.
20331 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
20332 support for TARGET_CMPB.
20333 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
20334 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20335 documentation of the __builtin_cmpb overloaded built-in function.
20337 2017-05-11 Richard Biener <rguenther@suse.de>
20339 PR tree-optimization/80705
20340 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
20341 bases are not vectorizable.
20343 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20345 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
20346 when counting register pressure.
20348 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20350 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
20351 (iv_ca_more_deps): Renamed to ...
20352 (iv_ca_compare_deps): ... this.
20353 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
20355 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20357 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
20359 (determine_group_iv_costs): ... here.
20360 (find_inv_vars_cb): Record inv var if it's not recorded before.
20362 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20364 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
20365 (get_shiftadd_cost): Ditto.
20367 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20369 * tree-ssa-address.c: Include header file.
20370 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
20372 (add_to_parts): Refactor.
20373 (addr_to_parts): New parameter. Update use of move_hint_to_base.
20374 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
20377 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20379 PR tree-optimization/53090
20380 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
20382 (extract_cond_operands): Detect condition with IV on both sides
20383 and return COMP_IV_EXPR_2.
20384 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
20385 (rewrite_use_compare): Simplify by removing call to function
20386 extract_cond_operands.
20388 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20390 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
20391 (extract_cond_operands): Detect condition comparing against non-
20392 invariant bound and return appropriate enum value.
20393 (find_interesting_uses_cond): Update use of extract_cond_operands.
20394 Handle its return value accordingly.
20395 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
20397 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20399 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
20400 nonlinear iv_use computation in loop invariant sensitive way.
20402 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20404 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
20405 (find_iv_candidates): Call relate_compare_use_with_all_cands.
20407 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20409 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
20410 (dump_cand): Support iv_cand.inv_exprs.
20411 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
20413 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
20416 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20418 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
20420 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
20421 as local function. Include necessary header files.
20422 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
20424 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20426 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
20428 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20430 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
20431 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
20432 RSHIFT_EXPR and BIT_NOT_EXPR.
20434 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20436 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
20437 (adjust_setup_cost): New parameter supporting round up adjustment.
20438 (struct address_cost_data): Delete.
20439 (force_expr_to_var_cost): Don't bound cost with spill_cost.
20440 (split_address_cost, ptr_difference_cost): Delete.
20441 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
20442 (struct ainc_cost_data): New struct.
20443 (get_address_cost_ainc): New function.
20444 (get_address_cost, get_computation_cost): Reimplement.
20445 (determine_group_iv_cost_address): Record inv_expr for all uses of
20447 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
20448 (iv_ca_has_deps): Reimplemented to ...
20449 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
20451 (iv_ca_extend): Call iv_ca_more_deps.
20453 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20455 * tree-ssa-address.c (struct mem_address): Move to header file.
20456 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
20457 * tree-ssa-address.h (struct mem_address): Move from C file.
20458 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
20460 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20462 * tree-affine.h (aff_combination_type): New interface.
20463 (aff_combination_zero_p): Remove static.
20464 (aff_combination_const_p): New interface.
20465 (aff_combination_singleton_var_p): New interfaces.
20467 2017-05-11 Richard Biener <rguenther@suse.de>
20469 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20470 Skip unreachable blocks and destinations.
20471 (eliminate): Move stmt removal and fixup ...
20472 (fini_eliminate): ... here. Skip inserted exprs.
20473 (pass_pre::execute): Move fini_pre after fini_eliminate.
20474 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
20475 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
20476 PRE to get rid of dead code that has invalid SSA form and
20477 split critical edges again.
20479 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20481 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20483 2017-05-11 Richard Biener <rguenther@suse.de>
20485 * passes.c (execute_function_todo): Verify loops if they are
20486 said to be up-to-date.
20487 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
20488 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
20490 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
20493 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
20494 handle calling assemble_external ourself.
20497 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
20498 modes with zero size. Enhance comment.
20500 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20502 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
20503 built-ins for vec_xl and vec_xst with short and char pointer
20506 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20508 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
20509 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
20510 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
20511 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
20512 (_mm_maskz_min_round_ss): New intrinsics.
20513 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
20514 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
20515 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
20516 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
20517 (__builtin_ia32_minss_mask_round): New builtins.
20518 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20519 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20520 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
20522 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
20523 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
20525 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
20528 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20530 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
20531 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
20532 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
20533 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
20534 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
20535 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
20536 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
20537 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20538 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20539 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
20540 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
20541 (__builtin_ia32_mulss_mask_round): New builtins.
20542 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20543 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20544 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
20546 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
20547 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20549 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20552 2017-05-10 Julia Koval <julia.koval@intel.com>
20554 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
20555 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
20556 (_mm256_setr_m128i): New intrinsics.
20558 2017-05-10 Julia Koval <julia.koval@intel.com>
20560 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
20561 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
20562 (_mm_maskz_rcp14_ss): New intrinsics.
20563 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
20564 (__builtin_ia32_rcp14ss_mask): New builtins.
20565 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
20567 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
20569 PR tree-optimization/51513
20570 * tree-cfg.c (gimple_seq_unreachable_p): New function.
20571 (assert_unreachable_fallthru_edge_p): Use it.
20572 (group_case_labels_stmt): Likewise.
20573 * tree-cfg.h: Prototype it.
20574 * stmt.c: Include cfghooks.h and tree-cfg.h.
20575 (emit_case_dispatch_table) <gap_label>: New local variable.
20576 Use it to fill dispatch table gaps.
20577 Test for default_label before updating probabilities.
20578 (expand_case) <default_label>: Remove unneeded initialization.
20579 Test for unreachable default case statement and remove its edge.
20580 Set default_label accordingly.
20581 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
20583 2017-05-10 Carl Love <cel@us.ibm.com>
20585 * config/rs6000/rs6000-c: Add support for built-in functions
20586 vector signed char vec_neg (vector signed char)
20587 vector signed short int vec_neg (vector short int)
20588 vector signed int vec_neg (vector signed int)
20589 vector signed long long vec_neg (vector signed long long)
20590 vector float vec_neg (vector float)
20591 vector double vec_neg (vector double)
20592 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
20594 * config/rs6000/altivec.h: Add define for vec_neg
20595 * doc/extend.texi: Update the built-in documentation for the
20596 new built-in functions.
20598 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20600 PR tree-optimization/77644
20601 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
20603 2017-05-10 Nathan Sidwell <nathan@acm.org>
20605 * dumpfile.h (TDI_lang_all): New.
20606 (TDF_KIND): New. Renumber others
20607 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
20609 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
20611 (get_dump_file_name): Adjust suffix generation.
20612 (dump_enable_all): Use TDF_KIND.
20613 * doc/invoke.texi (-fdump-lang-all): Document.
20615 * dumpfile.h: Tabify.
20617 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
20620 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
20621 Move member access before delete.
20623 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
20625 * tree-inline.c (expand_call_inline): Split block at stmt
20628 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
20631 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
20632 are now unused after splitting mov{sf,sd}_hardfloat.
20633 (f32_lr2): Likewise.
20634 (f32_lm): Likewise.
20635 (f32_lm2): Likewise.
20636 (f32_li): Likewise.
20637 (f32_li2): Likewise.
20638 (f32_lv): Likewise.
20639 (f32_sr): Likewise.
20640 (f32_sr2): Likewise.
20641 (f32_sm): Likewise.
20642 (f32_sm2): Likewise.
20643 (f32_si): Likewise.
20644 (f32_si2): Likewise.
20645 (f32_sv): Likewise.
20646 (f32_dm): Likewise.
20647 (f32_vsx): Likewise.
20648 (f32_av): Likewise.
20649 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
20650 For movsf, order stores so the VSX stores occur before the GPR
20651 store which encourages the register allocator to use a traditional
20652 FPR instead of a GPR. For movsd, order the stores so that the GPR
20653 store comes before the VSX stores to allow the power6 to work.
20654 This is due to the power6 not having a 32-bit integer store
20655 instruction from a FPR.
20656 (movsf_hardfloat): Likewise.
20657 (movsd_hardfloat): Likewise.
20659 2017-05-09 Martin Sebor <msebor@redhat.com>
20661 PR translation/80280
20662 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
20665 PR translation/80280
20666 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
20667 data member added in r247778.
20668 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
20670 2017-05-09 Nathan Sidwell <nathan@acm.org>
20672 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
20674 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
20677 2017-05-09 Marek Polacek <polacek@redhat.com>
20679 * doc/invoke.texi: Fix typo.
20681 2017-05-09 Richard Biener <rguenther@suse.de>
20683 * tree-vrp.c (vrp_val_is_max): Adjust comment.
20684 (vrp_val_is_min): Likewise.
20685 (set_value_range_to_value): Likewise.
20686 (set_value_range_to_nonnegative): Likewise.
20687 (gimple_assign_nonzero_p): Likewise.
20688 (gimple_stmt_nonzero_p): Likewise.
20689 (vrp_int_const_binop): Likewise. Remove unreachable case.
20690 (adjust_range_with_scev): Adjust comments.
20691 (compare_range_with_value): Likewise.
20692 (extract_range_from_phi_node): Likewise.
20693 (test_for_singularity): Likewise.
20695 2017-05-09 Richard Biener <rguenther@suse.de>
20697 * tree-vrp.c (get_single_symbol): Add assert that we don't
20698 get overflowed constants as invariant part.
20699 (compare_values_warnv): Add comment before the TREE_NO_WARNING
20700 checks. Use wi::cmp instead of recursing for integer constants.
20701 (compare_values): Just ignore whether we assumed undefined
20702 overflow instead of failing the compare.
20703 (extract_range_for_var_from_comparison_expr): Add comment before the
20704 TREE_NO_WARNING sets.
20705 (test_for_singularity): Likewise.
20706 (extract_range_from_comparison): Do not disable optimization
20707 when we assumed undefined overflow.
20708 (extract_range_basic): Remove init of unused var.
20710 2017-05-09 Richard Biener <rguenther@suse.de>
20712 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
20713 (extract_range_from_multiplicative_op_1): Adjust.
20714 (extract_range_from_binary_expr_1): Use int_const_binop.
20716 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20719 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
20720 rs6000_store_data_bypass_p in seven define_bypass directives and
20721 in several comments.
20722 * config/rs6000/rs6000-protos.h: Add prototype for
20723 rs6000_store_data_bypass_p function.
20724 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
20725 function implements slightly different (rs6000-specific) semantics
20726 than store_data_bypass_p, returning false rather than aborting
20727 with assertion error when arguments do not satisfy the
20728 requirements of store data bypass.
20729 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
20730 rs6000_store_data_bypass_p.
20732 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
20734 * config/xtensa/xtensa-protos.h
20735 (xtensa_initial_elimination_offset): New declaration.
20736 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20737 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
20738 macro definition, add case for FRAME_POINTER_REGNUM when
20739 FRAME_GROWS_DOWNWARD.
20740 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
20741 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
20742 xtensa_initial_elimination_offset.
20744 2017-05-08 Nathan Sidwell <nathan@acm.org>
20746 * doc/invoke.texi: Alphabetize -fdump options.
20748 2017-05-08 Martin Sebor <msebor@redhat.com>
20750 PR translation/80280
20751 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
20753 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
20755 * target.def (compute_frame_layout): New optional target hook.
20756 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
20757 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
20758 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
20760 * reload1.c (verify_initial_elim_offsets): Likewise.
20761 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
20762 (use_simple_return_p): Call arm_compute_frame_layout if needed.
20763 (arm_get_frame_offsets): Split up into this ...
20764 (arm_compute_frame_layout): ... and this function.
20766 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
20768 * config/aarch64/constraints.md (Usa): New constraint.
20769 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
20771 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
20773 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
20774 with_multilib_list after it has been checked.
20776 2017-05-08 Richard Biener <rguenther@suse.de>
20778 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
20779 (bitmap_set_subtract_values): Likewise.
20781 2017-05-08 Richard Biener <rguenther@suse.de>
20783 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
20784 (gimple_assign_nonzero): ... this and remove strict_overflow_p
20786 (gimple_stmt_nonzero_warnv_p): Rename to ...
20787 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
20789 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
20790 (extract_range_basic): Adjust, do not disable propagation on
20791 strict overflow sensitive simplification.
20792 (vrp_visit_cond_stmt): Likewise.
20794 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
20796 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
20797 body size unconditionally.
20799 2017-05-07 Jeff Law <law@redhat.com>
20802 2017-05-06 Jeff Law <law@redhat.com>
20803 PR tree-optimization/78496
20804 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
20807 PR tree-optimization/78496
20808 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
20809 (simplify_stmt_using_ranges): Call it.
20810 (vrp_dom_walker::before_dom_children): Extract equivalences
20811 from an ASSERT_EXPR with an equality comparison against a
20814 2017-05-06 Jeff Law <law@redhat.com>
20816 PR tree-optimization/78496
20817 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
20820 PR tree-optimization/78496
20821 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
20822 (simplify_stmt_using_ranges): Call it.
20823 (vrp_dom_walker::before_dom_children): Extract equivalences
20824 from an ASSERT_EXPR with an equality comparison against a
20827 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
20829 * lra-constraints.c (lra_copy_reg_equiv): New function.
20830 (split_reg): Use it to copy equivalence information from the
20831 original register to the spill register.
20833 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
20835 PR rtl-optimization/75964
20836 * simplify-rtx.c (simplify_const_relational_operation): Remove
20837 invalid handling of comparisons of integer ABS.
20839 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
20841 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
20842 initialize to zero.
20843 (init_regs): Remove declaration.
20844 (function_arg_advance_32): Initialize error_p as boolean variable.
20846 2017-05-05 Nathan Sidwell <nathan@acm.org>
20848 * store-motion.c (remove_reachable_equiv_notes): Reformat long
20849 lines. Use for (;;).
20851 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20853 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
20854 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
20855 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
20856 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
20857 VF=2 that require versioning.
20859 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20861 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
20864 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20866 * diagnostic.h (diagnostic_override_option_index): Convert from
20867 macro to inline function.
20869 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20871 * diagnostic.c (last_module_changed_p): New function.
20872 (set_last_module): New function.
20873 (diagnostic_report_current_module): Convert macro usage to
20874 the above functions.
20875 * diagnostic.h (diagnostic_context::last_module): Strengthen
20876 from const line_map * to const line_map_ordinary *.
20877 (diagnostic_last_module_changed): Delete macro.
20878 (diagnostic_set_last_module): Delete macro.
20880 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20882 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
20883 with diagnostic_report_diagnostic.
20884 (diagnostic_n_impl_richloc): Likewise.
20885 * diagnostic.h (report_diagnostic): Delete macro.
20886 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
20887 with diagnostic_report_diagnostic.
20888 * substring-locations.c (format_warning_va): Likewise.
20890 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20892 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
20893 save/restor of format_spec. Move option-printing code to...
20894 (print_option_information): ...this new function, and
20895 reimplement by simply printing to the pretty_printer,
20896 rather than appending to the format string.
20898 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20900 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
20901 handling logic into...
20902 (update_effective_level_from_pragmas): ...this new function.
20904 2017-05-04 Andrew Waterman <andrew@sifive.com>
20906 * config/riscv/riscv.opt (mstrict-align): New option.
20907 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
20908 (SLOW_UNALIGNED_ACCESS): Define.
20909 (riscv_slow_unaligned_access): Declare.
20910 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
20912 (riscv_slow_unaligned_access): New variable.
20913 (rocket_tune_info): Set slow_unaligned_access to true.
20914 (optimize_size_tune_info): Set slow_unaligned_access to false.
20915 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
20916 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
20917 (riscv_option_override): Set riscv_slow_unaligned_access.
20918 * doc/invoke.texi: Add -mstrict-align to RISC-V.
20920 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
20922 * config/riscv/riscv.md: Unify indentation.
20924 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20929 * config/rs6000/rs6000.md (u code attribute): Add FIX and
20931 (extendsi<mode>2): Add support for doing sign extension via
20932 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
20933 don't have ISA 3.0 instructions.
20934 (extendsi<mode>2 splitter): Likewise.
20935 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
20936 generate the normal insns since SImode can now go in vector
20937 registers. Disallow the special UNSPECs needed for previous
20938 machines to hide SImode being used. Add new insns
20939 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
20940 (fix_trunc<mode>si2_stfiwx): Likewise.
20941 (fix_trunc<mode>si2_internal): Likewise.
20942 (fixuns_trunc<mode>si2): Likewise.
20943 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20944 (fctiw<u>z_<mode>_smallint): Likewise.
20945 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
20946 of floating point to 32-bit integer from doing a direct move to
20947 the GPR registers to do a store.
20948 (fctiwz_<mode>): Break long line.
20950 2017-05-05 Bin Cheng <bin.cheng@arm.com>
20952 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
20953 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
20954 (addr_list, addr_offset_valid_p): New.
20955 (split_address_groups): Check offset validity with above function.
20956 (gt-tree-ssa-loop-ivopts.h): Include header file.
20958 2017-05-05 Nathan Sidwell <nathan@acm.org>
20960 * config.gcc (arm*-*-*): Add missing 'fi'.
20962 2017-05-05 Steve Ellcey <sellcey@cavium.com>
20964 * doc/invoke.texi (-fopt-info): Explicitly say order of options
20965 included in -fopt-info does not matter.
20966 * doc/optinfo.texi (-fopt-info): Fix description of default
20967 behavour. Explicitly say order of options included in -fopt-info
20970 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20972 * config.gcc: Allow combinations of aprofile and rmprofile values for
20973 --with-multilib-list.
20974 * config/arm/t-multilib: New file.
20975 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
20976 variables. Remove setting of ISA and floating-point ABI in
20977 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
20978 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
20979 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
20980 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
20982 * config/arm/t-rmprofile: Likewise except for the matches changes.
20983 * doc/install.texi (--with-multilib-list): Document the combination of
20984 aprofile and rmprofile values and warn about pitfalls in doing that.
20986 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20988 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
20989 (movdi_aarch64): Likewise.
20991 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20993 PR tree-optimization/80632
20994 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
20996 (build_arrays): Initialize it for virtual phis.
20997 (fix_phi_nodes): Use it for virtual phis.
20999 PR tree-optimization/80558
21000 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
21001 [x, y] op z into [x op, y op z] for op & or | if conditions
21004 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
21005 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21008 * config/arm/arm.md (use_literal_pool): Remove.
21009 (64-bit immediate split): No longer takes cost into consideration
21010 if arm_disable_literal_pool is enabled.
21011 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
21012 used when arm_disable_literal_pool is enabled.
21013 (arm_max_const_double_inline_cost): Remove use of
21014 arm_disable_literal_pool.
21015 (push_minipool_fix): Add assert.
21016 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
21017 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
21018 (no_literal_pool_sf_immediate): New.
21020 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21022 PR tree-optimization/80613
21023 * tree-ssa-dce.c (propagate_necessity): Remove cases for
21024 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
21026 2017-05-05 Richard Biener <rguenther@suse.de>
21028 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
21030 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
21032 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
21033 of this flag from insn conditions due to removal from r247495.
21035 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21037 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
21039 (arm_early_store_addr_dep_ptr): Likewise.
21040 * config/arm/aarch-common-protos.h
21041 (arm_early_load_addr_dep_ptr): Add prototype.
21042 (arm_early_store_addr_dep_ptr): Likewise.
21043 * config/arm/cortex-a53.md: Add new bypasses.
21045 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21047 * tree.c (next_type_uid): Change type to unsigned.
21048 (type_hash_canon): Decrement back next_type_uid if
21049 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
21050 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
21053 2017-05-04 Martin Sebor <msebor@redhat.com>
21055 * builtins.c: Fix a trivial typo in a comment.
21057 PR middle-end/79234
21058 * builtins.c (check_sizes): Adjust to handle reading past the end.
21059 Avoid printing excessive upper bound of ranges. Use %E to print
21060 tree nodes instead of converting them to %wu.
21061 (expand_builtin_memchr): New function.
21062 (compute_dest_size): Rename...
21063 (compute_objsize): ...to this.
21064 (expand_builtin_memcpy): Adjust.
21065 (expand_builtin_mempcpy): Adjust.
21066 (expand_builtin_strcat): Adjust.
21067 (expand_builtin_strcpy): Adjust.
21068 (check_strncat_sizes): Adjust.
21069 (expand_builtin_strncat): Adjust.
21070 (expand_builtin_strncpy): Adjust and simplify.
21071 (expand_builtin_memset): Adjust.
21072 (expand_builtin_bzero): Adjust.
21073 (expand_builtin_memcmp): Adjust.
21074 (expand_builtin): Handle memcmp.
21075 (maybe_emit_chk_warning): Check strncat just once.
21077 2017-05-04 Martin Sebor <msebor@redhat.com>
21079 PR preprocessor/79214
21080 PR middle-end/79222
21081 PR middle-end/79223
21082 * builtins.c (check_sizes): Add inlining context and issue
21083 warnings even when -Wno-system-headers is set.
21084 (check_strncat_sizes): Same.
21085 (expand_builtin_strncat): Same.
21086 (expand_builtin_memmove): New function.
21087 (expand_builtin_stpncpy): Same.
21088 (expand_builtin): Handle memmove and stpncpy.
21090 2017-05-04 Bin Cheng <bin.cheng@arm.com>
21092 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
21093 which is not used any more.
21095 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21097 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
21099 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21101 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
21102 (cortexa53_tunings): Likewise.
21103 (cortexa57_tunings): Likewise.
21104 (cortexa72_tunings): Likewise.
21105 (cortexa73_tunings): Likewise.
21107 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21109 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
21110 Set loop alignment to 8.
21112 2017-05-04 Martin Sebor <msebor@redhat.com>
21114 PR translation/80280
21115 * builtins.c (expand_builtin_object_size): Add missing quoting to
21116 %D and like directives.
21117 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
21118 (hsa_type_for_tree_type): Same.
21119 (verify_function_arguments): Same.
21120 * symtab.c (symbol_table::change_decl_assembler_name): Same.
21121 * varasm.c (get_section): Same.
21124 2017-05-04 Martin Sebor <msebor@redhat.com>
21126 PR translation/80280
21127 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
21129 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21131 * config/aarch64/aarch64.c (generic_addrcost_table):
21132 Change HI/TI mode setting.
21134 2017-05-04 Martin Jambor <mjambor@suse.cz>
21136 PR tree-optimization/80622
21137 * tree-sra.c (comes_initialized_p): New function.
21138 (build_accesses_from_assign): Only set write lazily when
21139 comes_initialized_p is false.
21140 (analyze_access_subtree): Use comes_initialized_p.
21141 (propagate_subaccesses_across_link): Assert !comes_initialized_p
21142 instead of testing for PARM_DECL.
21144 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21146 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
21147 constraint on operand 0 to allow more general addressing modes.
21148 Adjust output template.
21149 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
21151 * config/aarch64/aarch64-protos.h
21152 (aarch64_address_valid_for_prefetch_p): Declare prototype.
21153 * config/aarch64/constraints.md (Dp): New address constraint.
21154 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
21157 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
21159 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
21160 update use of estimate_ipcp_clone_size_and_time.
21161 (estimate_local_effects): Update use of
21162 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
21163 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
21164 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
21165 Return nonspecialized time.
21167 2017-05-04 Richard Biener <rguenther@suse.de>
21169 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
21170 for the last VUSE which def dominates the PHI. Directly call
21172 (get_continuation_for_phi_1): Remove.
21174 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
21176 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
21177 to explain the use of truncating division. Cap the number of
21178 iterations to the maximum given by nb_iterations_upper_bound,
21181 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21183 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
21184 * configure: Regenerate.
21185 * config.in: Regenerate.
21186 * config/i386/driver-mingw32.c: new file.
21187 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
21188 * config.host: Link driver-mingw32.o on MinGW host.
21189 * doc/install.texi: Document new --enable-mingw-wildcard configure
21192 2017-05-04 Marek Polacek <polacek@redhat.com>
21194 PR tree-optimization/80612
21195 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
21197 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21198 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
21200 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
21201 (movt splitter): Likewise.
21202 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
21203 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
21204 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
21205 block for Thumb-1 with MOVT.
21206 (thumb2_legitimate_address_p): Move code block ...
21207 (can_avoid_literal_pool_for_label_p): ... into this new function.
21208 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
21210 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
21211 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
21212 "M-profile targets with the MOVT instruction".
21214 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21216 * config/arm/arm-builtins.c (arm_init_builtins): Rename
21217 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
21218 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
21220 2017-05-04 Martin Liska <mliska@suse.cz>
21222 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
21223 variable cond_code.
21225 2017-05-04 Richard Biener <rguenther@suse.de>
21227 * tree.c (array_at_struct_end_p): Handle arrays at struct
21228 end with flexarrays more conservatively. Refactor and treat
21229 arrays of arrays or aggregates more strict. Fix
21230 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
21231 * tree.c (array_at_struct_end_p): Adjust prototype.
21232 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
21233 * gimple-fold.c (get_range_strlen): Likewise.
21234 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
21236 2017-05-04 Richard Biener <rguenther@suse.de>
21238 PR tree-optimization/31130
21239 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
21241 (supports_overflow_infinity): Likewise.
21242 (is_negative_overflow_infinity): Likewise.
21243 (is_positive_overflow_infinity): Likewise.
21244 (is_overflow_infinity): Likewise.
21245 (stmt_overflow_infinity): Likewise.
21246 (overflow_infinity_range_p): Likewise.
21247 (usable_range_p): Remove as always returning true.
21248 (make_overflow_infinity): Remove.
21249 (negative_overflow_infinity): Likewise.
21250 (positive_overflow_infinity): Likewise.
21251 (avoid_overflow_infinity): Likewise.
21252 (set_value_range): Adjust accordingly.
21253 (set_value_range_to_nonnegative): Likewise, remove now unused
21254 overflow_infinity arg.
21255 (vrp_operand_equal_p): Adjust.
21256 (update_value_range): Likewise.
21257 (range_int_cst_singleton_p): Likewise.
21258 (operand_less_p): Likewise.
21259 (compare_values_warnv): Likewise.
21260 (extract_range_for_var_from_comparison_expr): Likewise.
21261 (vrp_int_const_binop): Likewise.
21262 (zero_nonzero_bits_from_vr): Likewise.
21263 (extract_range_from_multiplicative_op_1): Likewise.
21264 (extract_range_from_binary_expr_1): Likewise.
21265 (extract_range_from_unary_expr): Likewise.
21266 (extract_range_from_comparison): Likewise.
21267 (extract_range_basic): Likewise.
21268 (adjust_range_with_scev): Likewise.
21269 (compare_ranges): Likewise.
21270 (compare_range_with_value): Likewise.
21271 (dump_value_range): Likewise.
21272 (test_for_singularity): Likewise, remove strict_overflow_p parameter
21274 (simplify_cond_using_ranges): Adjust.
21276 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21278 * brig-builtins.def: Added a builtin for class_f64.
21279 * builtin-types.def: Added a builtin type needed by class_f64.
21281 2017-05-03 Jason Merrill <jason@redhat.com>
21283 * timevar.def: Add TV_CONSTEXPR.
21285 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21287 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
21289 2017-05-03 Martin Jambor <mjambor@suse.cz>
21291 * ipa-prop.c (ipa_update_after_lto_read): Removed.
21292 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
21293 * ipa-cp.c (ipcp_propagate_stage): Do not call
21294 ipa_update_after_lto_read.
21295 * ipa-inline.c (ipa_inline): Likewise.
21297 2017-05-03 Martin Jambor <mjambor@suse.cz>
21299 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
21300 tag. Added a default constructor and a destructor.
21301 (ipa_edge_args_sum_t): New class;
21302 (ipa_edge_args_sum): Declare.
21303 (ipa_edge_args_vector): Remove declaration.
21304 (IPA_EDGE_REF): Use ipa_edge_args_sum.
21305 (ipa_free_edge_args_substructures): Remove declaration.
21306 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
21307 (ipa_edge_args_info_available_for_edge_p): Likewise.
21308 * ipa-prop.c (ipa_edge_args_vector): Removed.
21309 (edge_removal_hook_holder): Likewise.
21310 (edge_duplication_hook_holder): Likewise.
21311 (ipa_edge_args_sum): New variable.
21312 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
21313 ipa_edge_args_vector.
21314 (ipa_free_edge_args_substructures): Likewise.
21315 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
21316 ipa_edge_args_vector.
21317 (ipa_edge_removal_hook): Turned into method
21318 ipa_edge_args_sum_t::remove.
21319 (ipa_edge_duplication_hook): Turned into method
21320 ipa_edge_args_sum_t::duplicate.
21321 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
21322 registering edge hooks.
21323 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
21324 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
21325 ipa_edge_args_sum instead of ipa_edge_args_vector.
21326 * ipa-profile.c (ipa_profile): Likewise.
21328 2017-05-03 Martin Jambor <mjambor@suse.cz>
21330 * symbol-summary.h (function_summary): New method exists.
21331 (function_summary::symtab_removal): Deallocate through release.
21332 (call_summary): New class.
21333 (gt_ggc_mx): New overload.
21334 (gt_pch_nx): Likewise.
21335 (gt_pch_nx): Likewise.
21337 2017-05-03 Jeff Law <law@redhat.com>
21339 PR tree-optimization/78496
21340 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
21341 from simplify_cond_using_ranges. Split off code to walk
21342 backwards through casts into ...
21343 (simplify_cond_using_ranges_2): New function.
21344 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
21345 (execute_vrp): After identifying jump threads, call
21346 simplify_cond_using_ranges_2.
21348 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
21351 * ipa-inline.h (inline_summary): Add ctor.
21352 (create_ggc): Do not use ggc_cleared_alloc.
21354 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
21355 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21357 * gcc.c (handle_braces): Support escaping in switch matching
21359 * doc/invoke.texi (Spec Files): Document it.
21360 Remove superfluous @code markup in items.
21362 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21364 * diagnostic-show-locus.c (struct column_range): New struct.
21365 (get_affected_columns): New function.
21366 (get_printed_columns): New function.
21367 (struct correction): New struct.
21368 (correction::ensure_capacity): New function.
21369 (correction::ensure_terminated): New function.
21370 (struct line_corrections): New struct.
21371 (line_corrections::~line_corrections): New dtor.
21372 (line_corrections::add_hint): New function.
21373 (layout::print_trailing_fixits): Reimplement in terms of the new
21375 (selftest::test_overlapped_fixit_printing): New function.
21376 (selftest::diagnostic_show_locus_c_tests): Call it.
21378 2017-05-03 Nathan Sidwell <nathan@acm.org>
21380 Canonicalize canonical type hashing
21381 * tree.h (type_hash_canon_hash): Declare.
21382 * tree.c (type_hash_list, attribute_hash_list): Move into
21383 type_hash_canon_hash.
21384 (build_type_attribute_qual_variant): Break out hash code calc into
21385 type_hash_canon_hash.
21386 (type_hash_canon_hash): New. Generic type hash computation.
21387 (build_range_type_1, build_array_type_1, build_function_type,
21388 build_method_type_directly, build_offset_type, build_complex_type,
21389 make_vector_type): Call it.
21391 2017-05-03 Richard Biener <rguenther@suse.de>
21393 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
21394 When all DRs have unknown misaligned do not always peel
21395 when there is a store but apply the same costing model as if
21396 there were only loads.
21398 2017-05-03 Richard Biener <rguenther@suse.de>
21401 PR tree-optimization/80492
21402 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21403 compare_base_decls returning dont-know properly.
21405 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
21407 * config/arm/iterators.md (CCSI): New mode iterator.
21408 (arch): New mode attribute.
21409 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
21410 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
21411 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
21412 code iterator for success result mode.
21413 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
21414 the corresponding new insn generators.
21416 2017-05-03 Bin Cheng <bin.cheng@arm.com>
21419 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21420 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21422 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
21424 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
21425 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
21426 (DDR_A): Wrap DDR argument in brackets.
21427 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
21428 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
21429 (DDR_REVERSED_P): Likewise.
21431 2017-05-03 Jakub Jelinek <jakub@redhat.com>
21433 PR tree-optimization/79472
21434 * tree-switch-conversion.c (struct switch_conv_info): Add
21435 contiguous_range and default_case_nonstandard fields.
21436 (collect_switch_conv_info): Compute contiguous_range and
21437 default_case_nonstandard fields, don't clear final_bb if
21438 contiguous_range and only the default case doesn't have the required
21440 (check_all_empty_except_final): Set default_case_nonstandard instead
21441 of failing if contiguous_range and the default case doesn't have empty
21443 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
21444 and only the default case doesn't have the required constants. Skip
21446 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
21447 if default_case_nonstandard.
21448 (build_constructors): Build constant 1 just once. Assert that default
21449 values aren't inserted in between cases if contiguous_range. Skip
21451 (build_arrays): Skip virtual phis.
21452 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
21453 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
21454 Handle virtual phis.
21455 (gen_inbound_check): Handle default_case_nonstandard case.
21456 (process_switch): Adjust check_final_bb caller. Call
21457 gather_default_values with the first non-default case instead of
21458 default case if default_case_nonstandard.
21460 2017-05-02 Nathan Sidwell <nathan@acm.org>
21462 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
21463 check. Fix formatting.
21465 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
21467 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
21468 errors when comparing specialized and unspecialized times.
21470 2017-05-02 David Malcolm <dmalcolm@redhat.com>
21472 * diagnostic-show-locus.c
21473 (layout::should_print_annotation_line_p): Make private.
21474 (layout::print_annotation_line): Make private.
21475 (layout::annotation_line_showed_range_p): Make private.
21476 (layout::show_ruler): Make private.
21477 (layout::print_source_line): Make private. Pass in line and
21478 line_width, rather than calling location_get_source_line. Drop
21480 (layout::print_leading_fixits): New method.
21481 (layout::print_any_fixits): Rename to...
21482 (layout::print_trailing_fixits): ...this, and make private.
21483 Don't print newline fixits.
21484 (diagnostic_show_locus): Move logic for printing one row into...
21485 (layout::print_line): ...this new function. Move the
21486 location_get_source_line call and error-handling from
21487 print_source_line to here. Call print_leading_fixits, and rename
21488 print_any_fixits to print_trailing_fixits.
21489 (selftest::test_fixit_insert_containing_newline): Update now that
21490 newlines are partially supported.
21491 (selftest::test_fixit_insert_containing_newline_2): New test.
21492 (selftest::test_fixit_replace_containing_newline): Update comments.
21493 (selftest::diagnostic_show_locus_c_tests): Call the new test.
21494 * edit-context.c (class added_line): New class.
21495 (class edited_line): Describe newline handling in comment.
21496 (edited_line::actually_edited_p): New method.
21497 (edited_line::print_content): Delete redundant decl.
21498 (edited_line::m_predecessors): New field.
21499 (edited_file::print_content): Call edited_line::print_content.
21500 (edited_file::print_diff): Update to support newlines.
21501 (edited_file::print_diff_hunk): Likewise.
21502 (edited_file::print_run_of_changed_lines): New function.
21503 (edited_file::print_diff_line): Convert to...
21504 (print_diff_line): ...this.
21505 (edited_file::get_effective_line_count): New function.
21506 (edited_line::edited_line): Initialize new field m_predecessors.
21507 (edited_line::~edited_line): Clean up m_predecessors.
21508 (edited_line::apply_fixit): Handle newlines.
21509 (edited_line::get_effective_line_count): New function.
21510 (edited_line::print_content): New function.
21511 (edited_line::print_diff_lines): New function.
21512 (selftest::test_applying_fixits_insert_containing_newline): New
21514 (selftest::test_applying_fixits_replace_containing_newline): New
21516 (selftest::insert_line): New function.
21517 (selftest::test_applying_fixits_multiple_lines): Add example of
21519 (selftest::edit_context_c_tests): Call the new tests.
21521 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21523 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
21524 parameter cand. Update dump information.
21525 (get_computation_cost): Update uses.
21527 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21529 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
21530 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
21531 (get_computation_at, rewrite_use_address): Update use of
21532 get_computation_aff.
21534 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21536 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
21537 (get_computation): Delete.
21538 (get_computation_cost): Implement like get_computation_cost_at.
21539 Use get_computation_at.
21540 (get_computation_cost_at): Delete.
21541 (rewrite_use_nonlinear_expr): Use get_computation_at.
21542 (rewrite_use_compare, remove_unused_ivs): Ditto.
21544 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21546 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
21548 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21550 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
21551 (ivopts_global_cost_for_size): Rename parameter and update uses.
21552 (iv_ca_recount_cost): Update uses.
21553 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
21554 candidates seperately in n_invs and n_cands.
21555 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
21557 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21559 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
21560 (find_inv_vars_cb): New.
21561 (find_depends): Renamed to ...
21562 (find_inv_vars): ... this.
21563 (add_candidate_1, force_var_cost): Call find_inv_vars.
21564 (split_address_cost, determine_group_iv_cost_cond): Ditto.
21566 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21568 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
21569 inv_vars. Add inv_exprs.
21570 (struct iv_cand): Rename depends_on to inv_vars.
21571 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
21572 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
21573 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
21574 (dump_cand): Dump inv_vars.
21575 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
21576 (record_invariant, find_depends, add_candidate_1): Ditto.
21577 (set_group_iv_cost, force_var_cost): Ditto.
21578 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
21579 (get_computation_cost_at, get_computation_cost): Ditto.
21580 (determine_group_iv_cost_generic): Ditto.
21581 (determine_group_iv_cost_address): Ditto.
21582 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
21583 (determine_group_iv_costs): Ditto.
21584 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
21585 (iv_ca_set_remove_invariants): Renamed to ...
21586 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
21587 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
21588 (iv_ca_set_add_invariants): Renamed to ...
21589 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
21590 (iv_ca_set_cp): Use iv_ca_set_add_invs.
21591 (iv_ca_has_deps): Support inv_vars and inv_exprs.
21592 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
21593 (create_new_ivs): Remove useless dump.
21595 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21597 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
21599 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
21600 (iv_ca_set_no_cp, create_new_iv): Ditto.
21602 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21604 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21606 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21608 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
21609 function tree_check2.
21611 2017-05-02 Martin Liska <mliska@suse.cz>
21613 * doc/gcov.texi: Add missing preposition.
21614 * gcov.c (function_info::function_info): Properly fill up
21615 all member variables.
21617 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21619 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
21621 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21623 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
21625 2017-05-02 Martin Liska <mliska@suse.cz>
21628 * lto-streamer-in.c (lto_read_tree_1): Remove
21629 LTO_STREAMER_DEBUG.
21630 * lto-streamer.c (struct tree_hash_entry): Likewise.
21631 (struct tree_entry_hasher): Likewise.
21632 (tree_entry_hasher::hash): Likewise.
21633 (tree_entry_hasher::equal): Likewise.
21634 (lto_streamer_init): Likewise.
21635 (lto_orig_address_map): Likewise.
21636 (lto_orig_address_get): Likewise.
21637 (lto_orig_address_remove): Likewise.
21638 * lto-streamer.h: Likewise.
21639 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
21640 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
21642 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
21644 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
21645 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
21646 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
21647 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
21648 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
21649 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
21650 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
21651 (mm_maskz_sub_ss): New intrinsics.
21652 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21653 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21654 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
21655 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
21656 (__builtin_ia32_subss_mask_round): New builtins.
21657 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21658 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21659 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
21661 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
21662 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21664 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21667 2017-05-02 Martin Jambor <mjambor@suse.cz>
21669 PR tree-optimization/78687
21670 * tree-sra.c (access): New field parent.
21671 (process_subtree_disqualification): New function.
21672 (disqualify_candidate): Call it.
21673 (build_accesses_from_assign): Reset write flag if creating an
21675 (build_access_subtree): Fill in parent field and also prpagate
21676 down grp_write flag.
21677 (create_artificial_child_access): New parameter set_grp_write, set
21678 grp_write to its value.
21679 (propagate_subaccesses_across_link): Also propagate grp_write flag
21681 (propagate_all_subaccesses): Push the closest parent back to work
21682 queue if add_access_to_work_queue returned true.
21684 2017-05-02 Richard Biener <rguenther@suse.de>
21686 * common.opt (fstrict-overflow): Alias negative to fwrapv.
21687 * doc/invoke.texi (fstrict-overflow): Remove all traces of
21688 -fstrict-overflow documentation.
21689 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
21690 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
21691 flag_strict_overflow.
21692 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
21693 * lto-opts.c (lto_write_options): Do not stream it.
21694 * lto-wrapper.c (merge_and_complain): Do not handle it.
21695 * opts.c (default_options_table): Do not set -fstrict-overflow.
21696 (finish_options): Likewise do not clear it when sanitizing.
21697 * simplify-rtx.c (simplify_const_relational_operation): Do not
21698 test flag_strict_overflow.
21700 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21702 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
21703 using enabled attribute.
21704 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
21705 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
21706 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
21707 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
21708 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
21709 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
21710 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
21711 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
21712 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
21713 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
21715 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21717 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
21719 2017-05-02 Richard Biener <rguenther@suse.de>
21721 PR tree-optimization/80591
21723 2017-04-10 Richard Biener <rguenther@suse.de>
21725 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21728 2017-05-02 Richard Biener <rguenther@suse.de>
21730 PR tree-optimization/80549
21731 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
21732 (cleanup_tree_cfg_noloop): Create forwarders to known loop
21733 headers if they do not have a preheader.
21735 2017-05-02 Martin Liska <mliska@suse.cz>
21738 * common.opt: Fix typo.
21739 * doc/invoke.texi: Likewise.
21741 2017-05-01 Jan Beulich <jbeulich@suse.com>
21743 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
21744 swapping, add (x,x,m,x,n) alternative.
21746 2017-05-01 Nathan Sidwell <nathan@acm.org>
21748 * calls.c (combine_pending_stack_adjustment_and_call): Remove
21749 unnecessary unadjusted_alignment check.
21751 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21754 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
21756 * gimplify.c (gimplify_cilk_detach): New function.
21757 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
21758 * tree-core.h: Document EXPR_CILK_SPAWN.
21759 * tree.h (EXPR_CILK_SPAWN): Define.
21761 2017-05-01 David Malcolm <dmalcolm@redhat.com>
21763 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
21764 to use new fixit_hint representation, using the "replace" logic.
21765 (get_line_span_for_fixit_hint): Likewise.
21766 (layout::print_any_fixits): Likewise.
21767 (selftest::test_one_liner_many_fixits): Rename to...
21768 (selftest::test_one_liner_many_fixits_1): ...this, and update
21769 comment and expected output to reflect that the multiple fix-it
21770 hints are now consolidated into one insertion.
21771 (selftest::test_one_liner_many_fixits_2): New test.
21772 (selftest::test_diagnostic_show_locus_one_liner): Update for
21774 (selftest::test_fixit_consolidation): Update for fix-it API
21776 * diagnostic.c (print_parseable_fixits): Likewise.
21777 * edit-context.c (edited_line::m_line_events): Convert from
21778 auto_vec <line_event *> to auto_vec <line_event>.
21779 (class line_event): Convert from abstract base class to a concrete
21780 class, taking over the role of replace_event.
21781 (class insert_event): Delete.
21782 (class replace_event): Rename to class line_event. Convert to
21784 (edit_context::add_fixits): Reimplement.
21785 (edit_context::apply_insert): Delete.
21786 (edit_context::apply_replace): Rename to...
21787 (edit_context::apply_fixit): ...this. Convert to half-open range.
21788 (edited_file::apply_insert): Delete.
21789 (edited_file::apply_replace): Rename to...
21790 (edited_file::apply_fixit): ...this.
21791 (edited_line::~edited_line): Drop deletion of events.
21792 (edited_line::apply_insert): Delete.
21793 (edited_line::apply_replace): Rename to...
21794 (edited_line::apply_fixit): ...this. Convert to half-open range.
21795 Update for change to type of m_line_events.
21796 * edit-context.h (edit_context::apply_insert): Delete.
21797 (edit_context::apply_replace): Rename to...
21798 (edit_context::apply_fixit): ...this.
21800 2017-05-01 Martin Sebor <msebor@redhat.com>
21802 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
21805 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
21808 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
21809 __get_cpuid_max returns 0.
21810 (__get_cpuid_count): Ditto.
21812 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
21814 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
21815 replacement expression is another instance of one of its arguments.
21817 2017-05-01 Jakub Jelinek <jakub@redhat.com>
21820 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
21821 check for stack push/pop autoinc.
21822 * config/i386/i386.c (ix86_agi_dependent): Return false
21823 if the only reason why modified_in_p returned true is that
21824 addr is SP based and set_insn is a push or pop.
21826 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
21828 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
21831 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
21834 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
21835 (account_size_time): Use two predicates - exec_pred and
21837 (evaluate_conditions_for_known_args): Compute both clause and
21839 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
21840 (inline_summary_t::duplicate): Update.
21841 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
21843 (compute_inline_parameters): Likewise.
21844 (estimate_edge_size_and_time): Update caluclation of time.
21845 (estimate_node_size_and_time): Compute both time and nonspecialized
21847 (estimate_ipcp_clone_size_and_time): Update.
21848 (inline_merge_summary): Update.
21849 (do_estimate_edge_time): Update.
21850 (do_estimate_edge_size): Update.
21851 (do_estimate_edge_hints): Update.
21852 (inline_read_section, inline_write_summary): Stream both new predicates.
21853 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
21855 (compute_inlined_call_time): Cleanup.
21856 (big_speedup_p): Update.
21857 (edge_badness): Update.
21858 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
21859 (size_time_entry): Replace predicate by exec_predicate and
21860 nonconst_predicate.
21861 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
21862 (estimate_edge_time): Return also nonspec_time.
21863 (reset_edge_growth_cache): Update.
21865 2017-04-29 Jakub Jelinek <jakub@redhat.com>
21867 PR rtl-optimization/80491
21868 * ifcvt.c (noce_process_if_block): When looking for x setter
21869 with missing else_bb, don't check only the insn right before
21870 cond_earliest, but look for the last insn that x is modified in
21871 within the same bb.
21873 PR rtl-optimization/80491
21874 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
21876 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
21878 PR tree-optimization/80487
21879 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
21881 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21883 PR tree-optimization/79697
21884 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
21885 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
21886 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
21888 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
21889 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
21891 2017-04-28 Martin Sebor <msebor@redhat.com>
21893 PR tree-optimization/80523
21894 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
21895 (init_target_to_host_charmap, target_to_host, target_strtol10): New
21897 (maybe_warn, format_directive, parse_directive): Use new functions.
21898 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
21900 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
21902 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
21904 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
21906 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
21907 target_header_dir): Set correctly.
21908 * configure: Regenerated.
21909 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21910 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21911 instead of SYSTEM_HEADER_DIR.
21913 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
21915 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
21916 (estimate_local_effects): Likewise.
21917 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
21918 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
21919 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
21920 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
21921 do_estimate_edge_time, estimate_edge_time): Likewise.
21922 * ipa-inline-analysis.c (estimate_node_size_and_time,
21923 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
21924 (estimate_time_after_inlining): Remove.
21926 2017-04-28 Martin Liska <mliska@suse.cz>
21928 * doc/gcov.texi: Enhance documentation of gcov.
21930 2017-04-28 Martin Liska <mliska@suse.cz>
21932 * doc/gcov.texi: Sort options in alphabetic order.
21933 * doc/gcov-dump.texi: Likewise.
21934 * doc/gcov-tool.texi: Likewise.
21935 * gcov.c (print_usage): Likewise.
21936 * gcov-dump.c (print_usage): Likewise.
21937 * gcov-tool.c (print_merge_usage_message): Likewise.
21938 (print_rewrite_usage_message): Likewise.
21939 (print_overlap_usage_message): Likewise.
21941 2017-04-28 Martin Liska <mliska@suse.cz>
21943 PR gcov-profile/53915
21944 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
21946 2017-04-28 Martin Liska <mliska@suse.cz>
21948 PR gcov-profile/79891
21949 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
21950 is marked by compiler as living on a line.
21951 (get_cycles_count): Remove usage of the union.
21952 (output_intermediate_file): Likewise.
21953 (find_source): Fix GNU coding style.
21954 (accumulate_line_counts): Remove old non-all block mode.
21955 (output_lines): Remove usage of the union.
21956 * profile.c (output_location): Include all BBs, even if
21957 belonging to a same line (and file) as a previous BB.
21959 2017-04-28 Martin Liska <mliska@suse.cz>
21961 * gcov.c (process_args): Handle new argument 'w'.
21962 (read_graph_file): Assign ID to BBs.
21963 (output_branch_count): Display BB # if verbose flag is set.
21964 (output_lines): Likewise for arcs.
21965 (print_usage): Add '--verbose' option help.
21966 * doc/gcov.texi: Document --verbose (-w) option.
21968 2017-04-28 Martin Liska <mliska@suse.cz>
21970 * gcov.c (struct block_location_info): New struct.
21971 (process_file): Fill up the new structure.
21972 (read_graph_file): Replace usage of encoding by the newly added
21974 (add_line_counts): Likewise.
21975 (accumulate_line_counts): Remove usage of the union.
21976 (function_info::function_info): New function.
21977 (function_info::~function_info): Likewise.
21978 (process_file): Call delete instead of release_function.
21979 (release_function): Release the function.
21980 (release_structures): Call delete instead of release_function.
21981 (solve_flow_graph): Replace usage of num_blocks.
21982 (find_exception_blocks): Likewise.
21983 (output_lines): Fix GNU coding style.
21985 2017-04-28 Martin Liska <mliska@suse.cz>
21988 * coverage.c (coverage_remove_note_file): New function.
21989 * coverage.h: Declare the function.
21990 * toplev.c (finalize): Clean if an error has been seen.
21992 2017-04-28 Martin Liska <mliska@suse.cz>
21994 PR gcov-profile/80031
21995 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
21996 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
21997 * gcov.c (read_graph_file): Read just number of blocks.
21998 * profile.c (branch_prob): Do not stream 0 flags per a basic
22001 2017-04-28 Martin Liska <mliska@suse.cz>
22003 * gcov-dump.c (tag_*): Add new argument to declarations.
22004 (dump_gcov_file): Likewise.
22005 (tag_blocks): Add and use new argument depth.
22006 (tag_arcs): Likewise.
22007 (tag_lines): Likewise.
22008 (tag_counters): Likewise.
22009 (tag_summary): Likewise.
22010 (dump_working_sets): Use depth to do a proper indentation.
22012 2017-04-28 Jakub Jelinek <jakub@redhat.com>
22015 * cgraph.h (symtab_node::debug_symtab): No longer inline.
22016 * symtab.c (symtab_node::debug_symtab): Move definition here.
22018 2017-04-28 Richard Biener <rguenther@suse.de>
22020 * lto-streamer.h (LTO_major_version): Bump to 7.
22022 2017-04-28 Richard Biener <rguenther@suse.de>
22024 * tree-vrp.c (assert_info): New struct.
22025 (add_assert_info): New helper.
22026 (register_edge_assert_for_2): Refactor to add asserts to a vector
22028 (register_edge_assert_for_1): Likewise.
22029 (register_edge_assert_for): Likewise.
22030 (finish_register_edge_assert_for): New helper actually registering
22031 asserts where live on edge.
22032 (find_conditional_asserts): Adjust.
22033 (find_switch_asserts): Likewise.
22034 (evrp_dom_walker::try_find_new_range): Generalize.
22035 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
22037 2017-04-27 Marek Polacek <polacek@redhat.com>
22040 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
22041 arg10 and arg11 to itype.
22043 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
22045 * doc/extend.texi (Object Size Checking): Improve grammar.
22047 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
22050 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
22051 that the logic for permitting reciprocal estimates matches that
22054 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22057 * tree.c (type_cache_hasher::equal): Only compare
22058 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
22059 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
22060 non-aggregate element types.
22061 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
22062 about the flag on ARRAY_TYPEs in the comment, formatting fix.
22064 2017-04-27 Richard Biener <rguenther@suse.de>
22066 PR middle-end/80533
22067 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
22068 stripping ARRAY_REFs from MEM_EXPR make sure we're not
22069 keeping a reference to a trailing array.
22071 2017-04-27 Richard Biener <rguenther@suse.de>
22073 PR middle-end/80539
22074 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
22075 being in loop-closed SSA form conservatively.
22076 (chrec_fold_multiply_poly_poly): Likewise.
22078 2017-04-27 Tamar Christina <tamar.christina@arm.com>
22080 PR middle-end/79665
22081 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
22082 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
22084 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22087 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
22088 (aarch64_function_arg_alignment): Return unsigned int again, but still
22089 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
22090 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22091 Don't emit -Wpsabi note.
22092 (aarch64_function_arg_boundary): Likewise.
22093 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22096 2017-04-26 Nathan Sidwell <nathan@acm.org>
22098 * tree.h (crc32_unsigned_n): Declare.
22099 (crc32_unsigned, crc32_unsigned): Make inline.
22100 * tree.c (crc32_unsigned_bits): Replace with ...
22101 (crc32_unsigned_n): ... this.
22102 (crc32_unsigned, crc32_byte): Remove.
22103 (crc32_string): Remove unnecessary braces.
22105 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22107 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
22108 * ipa-inline-analysis.c (MAX_TIME): Remove.
22109 (account_size_time): Use sreal for time.
22110 (dump_inline_summary): Update.
22111 (estimate_function_body_sizes): Update.
22112 (estimate_edge_size_and_time): Update.
22113 (estimate_calls_size_and_time): Update.
22114 (estimate_node_size_and_time): Update.
22115 (inline_merge_summary): Update.
22116 (inline_update_overall_summary): Update.
22117 (estimate_time_after_inlining): Update.
22118 (inline_read_section): Update.
22119 (inline_write_summary): Update.
22120 * ipa-inline.c (compute_uninlined_call_time): Update.
22121 (compute_inlined_call_time): Update.
22122 (recursive_inlining): Update.
22123 (inline_small_functions): Update.
22124 (dump_overall_stats): Update.
22125 * ipa-inline.h: Include sreal.h.
22126 (size_time_entry): Turn time to sreal.
22127 (inline_summary): Turn self_time nad time to sreal.
22129 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22131 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
22133 (sreal::stream_out, sreal::stream_in): New.
22134 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
22136 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22138 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
22141 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
22144 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
22146 (dimode_scalar_chain::compute_convert_gain): Ditto.
22147 (dimode_scalar_chain::make_vector_copies): Ditto.
22148 (dimode_scalar_chain::convert_reg): Ditto.
22149 (dimode_scalar_chain::convert_insn): Ditto.
22150 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
22151 (VI248_AVX512BW_1): New mode iterator.
22152 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
22153 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
22156 2017-04-25 Martin Sebor <msebor@redhat.com>
22158 PR tree-optimization/80497
22159 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
22160 constants are representable in HOST_WIDE_INT.
22161 (parse_directive): Ditto.
22163 2017-04-25 Martin Sebor <msebor@redhat.com>
22166 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
22167 (new_zero_array): Adjust signature.
22168 (dom_info::dom_init): Used unsigned rather that size_t.
22169 (dom_info::dom_info): Same.
22171 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22172 Jakub Jelinek <jakub@redhat.com>
22175 * config/arm/arm.c: Include gimple.h.
22176 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22177 returns negative, increment ncrn only if it returned positive.
22178 (arm_needs_doubleword_align): Return int instead of bool,
22179 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
22180 members, but if there is any such non-FIELD_DECL
22181 > PARM_BOUNDARY aligned decl, return -1 instead of false.
22182 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22183 returns negative, increment nregs only if it returned positive.
22184 (arm_setup_incoming_varargs): Likewise.
22185 (arm_function_arg_boundary): Emit -Wpsabi note if
22186 arm_needs_doubleword_align returns negative, return
22187 DOUBLEWORD_ALIGNMENT only if it returned positive.
22189 2017-04-25 Marek Polacek <polacek@redhat.com>
22192 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22193 first argument to type.
22195 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
22198 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
22199 type checks to test for compatibility instead of equality.
22201 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22202 Jakub Jelinek <jakub@redhat.com>
22205 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
22207 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
22208 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
22209 the alignment computation, but return their maximum in warn_alignment.
22210 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22211 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
22213 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
22214 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22217 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22219 * config/arc/simdext.md (dmpyh): Fix typo.
22221 2017-04-25 Richard Biener <rguenther@suse.de>
22223 PR tree-optimization/80492
22224 * alias.c (compare_base_decls): Handle registers with asm
22225 specification conservatively.
22226 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22227 compare_base_decls returning dont-know properly.
22229 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22231 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
22232 (legitimate_offset_address_p): New function.
22233 (arc_legitimate_address_p): Use above function.
22235 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22237 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
22239 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22241 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
22242 ACCH registers whenever they are available.
22244 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22246 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
22247 double regs fix when not used.
22249 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22251 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
22253 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
22254 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
22256 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22258 * config/arc/arc.c (arc_output_addsi): Check for h-register class
22259 when emitting short ADD instructions.
22261 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22263 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
22265 (cmpsi_cc_c_insn): Likewise.
22266 (cbranchsi4_scratch): Compute proper instruction length using
22267 compact_hreg_operand.
22268 * config/arc/predicates.md (compact_hreg_operand): New predicate.
22270 2017-04-25 Richard Biener <rguenther@suse.de>
22272 PR middle-end/80509
22273 * passes.c (pass_manager::pass_manager): Initialize
22274 m_name_to_pass_map.
22276 2017-04-25 Richard Biener <rguenther@suse.de>
22278 PR tree-optimization/79201
22279 * tree-ssa-sink.c (statement_sink_location): Handle calls.
22281 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22284 * config/s390/vector.md: Split MEM->GPR vector moves for
22285 non-s_operand addresses.
22287 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22290 * config/s390/predicates.md (reload_const_wide_int_operand): New
22292 * config/s390/s390.md ("movti"): Remove d/P alternative.
22293 ("movti_bigconst"): New pattern definition.
22295 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22298 * s390-protos.h (s390_expand_cs_hqi): Removed.
22299 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
22300 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
22301 modes as well as CCZ1mode and CCZmode.
22302 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
22303 signature of s390_emit_compare_and_swap.
22304 (s390_expand_cs_hqi): Likewise, make static.
22305 (s390_expand_cs_tdsi): Generate an explicit compare before trying
22306 compare-and-swap, in some cases.
22307 (s390_expand_cs): Wrapper function.
22308 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
22310 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
22311 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
22312 patterns for small and large integers. Forbid symref memory operands.
22313 Move expander to s390.c. Require cc register.
22314 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
22315 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
22316 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
22317 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
22318 symref memory operands. Remove CC mode and call s390_match_ccmode
22320 ("atomic_exchange<mode>"): Allow and implement all integer modes.
22322 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22324 * config/s390/s390.md (define_peephole2): New peephole to help
22325 combining the load-and-test pattern with volatile memory.
22327 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22329 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
22330 with CCZmode for TARGET_Z196.
22332 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22334 PR rtl-optimization/80501
22335 * combine.c (make_compound_operation_int): Set subreg_code to SET
22336 even for AND with mask of the sign bit of mode.
22338 PR rtl-optimization/80500
22339 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
22340 sum's initial value.
22342 2017-04-25 Julian Brown <julian@codesourcery.com>
22343 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22345 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
22347 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
22349 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
22351 2017-04-25 Julian Brown <julian@codesourcery.com>
22352 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22354 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
22355 (thunderx2t99_sha): New Reservation.
22357 2017-04-25 Julian Brown <julian@codesourcery.com>
22358 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22360 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
22361 type for 1-element load.
22363 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
22365 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
22367 2017-04-24 Martin Jambor <mjambor@suse.cz>
22369 PR tree-optimization/80293
22370 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
22371 char arrays not totally scalarizable if it is false.
22372 (analyze_all_variable_accesses): Pass correct value in the new
22373 parameter. Add a statistics counter.
22375 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
22377 PR middle-end/79931
22378 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
22380 2017-04-24 Richard Biener <rguenther@suse.de>
22382 PR tree-optimization/80494
22383 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
22384 out for complex types.
22386 2017-04-24 Richard Biener <rguenther@suse.de>
22388 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
22389 * tree-ssa-sccvn.c (print_scc): Print SCC size.
22390 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
22391 (DFS): Adjust and never fail.
22392 (sccvn_dom_walker::fail): Remove.
22393 (sccvn_dom_walker::before_dom_children): Adjust.
22394 (run_scc_vn): Likewise and never fail.
22395 * tree-ssa-pre.c (pass_pre::execute): Adjust.
22396 (pass_fre::execute): Likewise.
22398 2017-04-24 Richard Biener <rguenther@suse.de>
22400 PR tree-optimization/79725
22401 * tree-ssa-sink.c (statement_sink_location): Return whether
22402 failure reason was zero uses. Move that check later.
22403 (sink_code_in_bb): Deal with zero uses by removing the stmt
22406 2017-04-24 Richard Biener <rguenther@suse.de>
22409 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
22410 pointer-based references.
22412 2017-04-24 Richard Biener <rguenther@suse.de>
22415 * pass_manager.h (pass_manager::operator new): Remove.
22416 (pass_manager::operator delete): Likewise.
22417 * passes.c (pass_manager::operator new): Remove.
22418 (pass_manager::operator delete): Likewise.
22419 (pass_manager::pass_manager): Zero individual pass members.
22421 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
22424 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
22425 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
22426 Check "XEXP (src, 1)" operand here.
22427 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
22428 Check "XEXP (src, 1)" operand here.
22429 (dimode_scalar_chain::make_vector_copies): Detect count register
22430 of a shift instruction. Zero extend count register from QImode
22431 to DImode to satisfy vector shift pattern count operand predicate.
22432 Substitute vector shift count operand with a DImode copy.
22433 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
22436 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
22438 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
22439 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22440 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
22441 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22442 (UNSPEC_NOREX_MEM): Remove definition.
22444 2017-04-21 Richard Biener <rguenther@suse.de>
22446 PR tree-optimization/79547
22447 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22448 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
22449 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
22450 without any constraints.
22452 2017-04-21 Richard Biener <rguenther@suse.de>
22454 PR tree-optimization/78847
22455 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
22457 2017-04-21 Richard Biener <rguenther@suse.de>
22459 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
22460 (build_distinct_type_copy): Likewise.
22461 (build_variant_type_copy): Likewise.
22462 * tree.c (build_qualified_type): Pass down mem-stat info.
22463 (build_distinct_type_copy): Likewise.
22464 (build_variant_type_copy): Likewise.
22466 2017-04-21 Richard Biener <rguenther@suse.de>
22468 PR tree-optimization/80237
22469 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
22471 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
22472 for a simplified result.
22474 2016-04-21 Richard Biener <rguenther@suse.de>
22476 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
22477 sth as strict as a simple_iv but a chrec without symbols and an
22478 operand defined in the loop we are peeling (and not some subloop).
22479 (propagate_constants_for_unrolling): Propagate all constants.
22481 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22484 * config/i386/i386.c (print_reg): Remove assert for disalowed
22485 regno values, call output_operand_lossage instead.
22487 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22490 * config/i386/constraints.md (Yc): New register constraint.
22491 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
22492 Use Yc constraint for alternative 2 of operand 0. Remove
22493 preferred_for_speed attribute.
22495 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22497 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
22498 lastprivate clauses in SIMT case.
22500 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
22502 * doc/invoke.texi (-Wextra-semi): Document new warning option.
22504 2017-04-20 Richard Biener <rguenther@suse.de>
22506 PR tree-optimization/57796
22507 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
22508 as N scalar stores.
22509 (vect_model_load_cost): Cost gathers as N scalar loads.
22511 2017-04-20 Richard Biener <rguenther@suse.de>
22513 * ggc-page.c (ggc_allocated_p): Rename to ...
22514 (safe_lookup_page_table_entry): ... this and return the lookup
22516 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
22518 2017-04-20 Richard Biener <rguenther@suse.de>
22520 PR tree-optimization/80453
22521 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
22522 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
22523 from the conditions.
22524 (vn_phi_eq): Pass them down.
22525 (vn_phi_lookup): Record them.
22526 (vn_phi_insert): Likewise.
22528 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
22530 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
22531 uninitialized variable warning to avoid buffer overrun.
22533 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22536 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
22537 is suppressed for '{ 0 }' in C.
22539 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22541 * BASE-VER: Set to 8.0.0.
22543 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
22545 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
22546 priority .init_array and .fini_array section with SECTION_NOTYPE
22549 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22551 PR middle-end/80423
22552 * tree.h (build_array_type): Add typeless_storage default argument.
22553 * tree.c (type_cache_hasher::equal): Also compare
22554 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
22555 (build_array_type): Add typeless_storage argument, set
22556 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
22558 (build_nonshared_array_type): Adjust build_array_type_1 caller.
22559 (build_array_type): Likewise. Add typeless_storage argument.
22561 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22562 Jakub Jelinek <jakub@redhat.com>
22564 PR tree-optimization/80426
22565 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
22566 operation on symbolic operands, also compute the overflow for the
22567 invariant part when the operation degenerates into a negation.
22569 2017-04-19 Jakub Jelinek <jakub@redhat.com>
22572 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
22573 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
22576 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
22578 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
22581 * config/avr/avr.c (tree.h): Include it.
22582 (cgraph.h): Include it.
22583 (avr_encode_section_info): Don't warn for uninitialized progmem
22584 variable if it's just an alias.
22586 2017-04-19 Richard Biener <rguenther@suse.de>
22589 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
22590 when needed by AutoPGO.
22592 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
22595 * doc/lto.texi: Remove an extra 'that'.
22597 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
22599 PR rtl-optimization/80429
22600 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
22601 are only used in debug insns.
22603 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22604 Vladimir Makarov <vmakarov@redhat.com>
22606 * config/sparc/predicates.md (input_operand): Add comment. Return
22607 true for any memory operand when LRA is in progress.
22608 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
22610 2017-04-18 Jeff Law <law@redhat.com>
22613 * mips.md ({return,simple_return}_internal): Do not overwrite
22616 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22618 PR tree-optimization/80443
22619 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
22620 instead of adding 1, subtract -1 and similarly instead of subtracting
22623 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
22625 PR rtl-optimization/80357
22626 * haifa-sched.c (tmp_bitmap): New variable.
22627 (model_recompute): Handle duplicate use records.
22628 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
22629 (free_global_sched_pressure_data): Free it.
22631 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
22634 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22635 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22636 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22637 instead of SYSTEM_HEADER_DIR.
22639 2017-04-18 Jeff Law <law@redhat.com>
22641 PR middle-end/80422
22642 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
22643 predecessors after walking up the insn chain.
22645 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22648 * dwarf2out.c (modified_type_die): Try harder not to emit internal
22649 sizetype type into debug info.
22651 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22654 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
22655 unneeded test for TARGET_UPPER_REGS_SF.
22656 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
22658 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22661 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
22662 instead of gsi_after_labels.
22664 2017-04-18 Jeff Law <law@redhat.com>
22666 * regcprop.c (maybe_mode_change): Avoid creating copies of the
22670 2017-04-13 Jeff Law <law@redhat.com>
22671 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22672 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22674 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
22677 * config/avr/avr.c (intl.h): Include it.
22678 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
22680 2017-04-18 Martin Liska <mliska@suse.cz>
22682 PR gcov-profile/78783
22683 * gcov-tool.c (gcov_output_files): Validate that destination
22684 file is either removed by the tool or by a user.
22686 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22687 Guy Benyei <guybe@mellanox.com>
22689 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
22690 block, and do not negate it, the stored id is already negative.
22692 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22694 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
22696 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22699 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
22700 masks of options that should be turned off if the VSX vector
22701 options are turned off.
22702 (OTHER_P8_VECTOR_MASKS): Likewise.
22703 (OTHER_VSX_VECTOR_MASKS): Likewise.
22704 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
22705 rs6000_disable_incompatible_switches to validate no type switches
22707 (rs6000_incompatible_switch): New function to disallow turning on
22708 other vector options if -mno-vsx, -mno-power8-vector, or
22709 -mno-power9-vector are specified.
22711 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22713 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
22715 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22717 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
22718 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
22719 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
22720 (ARG_POINTER_CFA_OFFSET): Likewise.
22722 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22724 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
22725 conditions to take advantage of various optimizations.
22727 2017-04-13 Jeff Law <law@redhat.com>
22729 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22730 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22731 (zero_extendsidi2_dext): Likewise.
22733 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22736 * fold-const.c (fold_ternary_loc): Revert
22737 use op0 instead of fold_convert_loc (loc, type, arg0) part of
22740 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
22742 PR rtl-optimization/80343
22743 * lra-remat.c (update_scratch_ops): Assign original hard reg to
22744 new scratch pseudo.
22746 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
22749 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
22750 to ubsan_encode_value.
22752 2017-04-13 Jeff Law <law@redhat.com>
22754 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
22755 appearing in DEBUG_INSNs.
22757 2017-04-13 Martin Liska <mliska@suse.cz>
22759 PR gcov-profile/80413
22760 * gcov-io.c (gcov_write_string): Copy to buffer just when
22761 allocated size is greater than zero.
22763 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22766 * dwarf2out.c (decls_for_scope): Ignore declarations of
22767 current_function_decl in BLOCK_NONLOCALIZED_VARS.
22769 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
22772 * ipa-visibility.c (non_local_p): Fix typos.
22773 (localize_node): When localizing symbol in same comdat group,
22774 dissolve the group only when we know external symbols are going
22776 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
22778 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22780 PR tree-optimization/79390
22781 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
22782 order does not result in usable sequence, retry with reversed operand
22788 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
22789 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
22790 op0 instead of fold_convert_loc (loc, type, arg0).
22792 2017-04-12 Jeff Law <law@redhat.com>
22794 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
22795 has a delay slot in the generated code.
22797 * config/cris/cris.md (cris_preferred_reload_class): Return
22798 GENNONACR_REGS rather than GENERAL_REGS.
22800 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22803 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
22804 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
22805 signedness of the result type.
22807 2017-04-12 Richard Biener <rguenther@suse.de>
22808 Jeff Law <law@redhat.com>
22810 PR tree-optimization/80359
22811 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
22812 trim stores to TARGET_MEM_REFs.
22814 2017-04-12 Richard Biener <rguenther@suse.de>
22816 PR tree-optimization/79390
22817 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
22818 threading case even more.
22820 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
22823 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
22824 for quad_address_p for TImode, instead of just not indexed_address.
22826 2017-04-12 Richard Biener <rguenther@suse.de>
22827 Bernd Edlinger <bernd.edlinger@hotmail.de>
22829 PR middle-end/79671
22830 * alias.c (component_uses_parent_alias_set_from): Handle
22831 TYPE_TYPELESS_STORAGE.
22832 (get_alias_set): Likewise.
22833 * tree-core.h (tree_type_common): Add typeless_storage flag.
22834 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
22835 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
22836 for types containing members with TYPE_TYPELESS_STORAGE.
22837 (place_field): Likewise.
22838 (layout_type): Likewise for ARRAY_TYPE.
22839 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
22840 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
22841 TYPE_TYPELESS_STORAGE.
22842 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
22844 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22847 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22848 first argument to type.
22850 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22854 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
22855 CONST0_RTX (mode) rather than const0_rtx where appropriate.
22856 (rs6000_expand_binop_builtin): Likewise.
22857 (rs6000_expand_ternop_builtin): Likewise; also add missing
22858 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
22859 vshasigma built-ins.
22860 * doc/extend.texi: Document that vec_xxpermdi's third argument
22861 must be a constant.
22863 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
22865 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
22866 Use shift_const cost parameter when calculating gain of STV shifts.
22868 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
22870 PR rtl-optimization/70478
22871 * lra-constraints.c (process_alt_operands): Check memory for
22872 disfavoring memory insn operand.
22874 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22876 PR middle-end/80100
22877 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
22878 left shift in unsigned HOST_WIDE_INT type.
22880 PR rtl-optimization/80385
22881 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
22882 (not (neg X)) into (plus X -1) for complex or non-integral modes.
22885 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
22886 if they have any depend clauses.
22888 2017-04-11 Martin Liska <mliska@suse.cz>
22891 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
22892 * ipa-split.c (split_function): Create a local comdat symbol
22893 if caller is in a comdat group.
22895 2017-04-11 Martin Liska <mliska@suse.cz>
22898 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
22901 2017-04-11 Martin Sebor <msebor@redhat.com>
22903 PR middle-end/80364
22904 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
22905 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
22907 (directive::set_width, directive::set_precision, format_character):
22909 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
22912 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
22915 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
22916 conflict, set target->arch_name instead of target->cpu_name.
22918 2017-04-11 Richard Biener <rguenther@suse.de>
22920 PR tree-optimization/80374
22921 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
22922 build_zero_cst, remove fold_convertible_p check again.
22924 2017-04-11 Martin Liska <mliska@suse.cz>
22927 * ubsan.c (instrument_object_size): Do not instrument register
22930 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22933 * config/i386/i386-builtin-types.def
22934 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
22935 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
22936 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
22937 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
22938 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
22939 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
22940 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
22941 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
22942 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
22943 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
22944 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
22945 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
22946 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
22947 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
22948 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
22949 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
22950 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
22951 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
22952 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
22953 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
22954 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
22955 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
22956 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
22957 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
22958 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
22959 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
22960 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
22961 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
22962 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
22963 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
22964 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
22965 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
22966 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
22967 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
22968 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
22969 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
22970 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
22971 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
22972 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
22973 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
22974 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
22975 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
22976 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
22977 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
22978 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
22979 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
22981 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
22982 flag to second_arg_count, handle 4 argument function type _COUNT
22983 aliases, handle second_arg_count on second argument rather than last.
22985 2017-04-10 Jeff Law <law@redhat.com>
22987 PR tree-optimization/80374
22988 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
22989 record anything if we can not convert integer_zero_node to the
22992 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22995 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22996 Enhance special handling given to the TARGET_P9_MINMAX option in
22997 relation to certain other options.
22999 2017-04-10 Bin Cheng <bin.cheng@arm.com>
23001 PR tree-optimization/80153
23002 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
23003 remove POINTER_PLUS_EXPR's base part directly, rather than through
23006 2017-04-10 Richard Biener <rguenther@suse.de>
23007 Bin Cheng <bin.cheng@arm.com>
23009 PR tree-optimization/80153
23010 * tree-affine.c (aff_combination_to_tree): Get base pointer from
23011 the first element of pointer type aff_tree. Build result expr in
23013 (add_elt_to_tree): Convert to type unconditionally. Remove other
23014 fold_convert calls.
23015 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
23016 (rewrite_use_nonlinear_expr): Check invariant using iv information.
23018 2017-04-10 Richard Biener <rguenther@suse.de>
23020 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23023 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
23025 PR rtl-optimization/70478
23026 * lra-constraints.c (curr_small_class_check): New.
23027 (update_and_check_small_class_inputs): New.
23028 (process_alt_operands): Update curr_small_class_check. Disfavor
23029 alternative insn memory operands. Check available regs for small
23032 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23035 * config/mips/mips.opt (-mvirt): Update description.
23036 * doc/invoke.texi (-mvirt): Likewise.
23038 2017-04-10 Richard Biener <rguenther@suse.de>
23040 PR middle-end/80362
23041 * fold-const.c (fold_binary_loc): Look at unstripped ops when
23042 looking for NEGATE_EXPR in -A / -B to A / B folding.
23044 2017-04-10 Martin Liska <mliska@suse.cz>
23046 PR gcov-profile/80224
23047 * gcov.c (print_usage): Fix usage string.
23048 (get_gcov_intermediate_filename): Remove.
23049 (output_gcov_file): Use both for normal and intermediate format.
23050 (generate_results): Do not initialize special file for
23051 intermediate format.
23053 2017-04-10 Richard Biener <rguenther@suse.de>
23055 PR tree-optimization/80304
23056 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
23059 2017-04-10 Nathan Sidwell <nathan@acm.org>
23062 * config/rs6000/rs6000.c (rs6000_vector_type): New.
23063 (rs6000_init_builtins): Use it.
23065 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23067 * config/arm/arm.md (<mrc>): Add mode to SET source.
23068 (<mrrc>): Likewise.
23070 2017-04-10 Richard Biener <rguenther@suse.de>
23072 PR middle-end/80344
23073 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
23075 2017-04-10 Jakub Jelinek <jakub@redhat.com>
23078 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
23079 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
23080 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
23081 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
23082 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
23083 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
23084 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
23085 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
23086 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
23087 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
23088 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
23089 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
23090 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
23091 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
23092 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
23093 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
23094 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
23095 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
23096 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
23097 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
23098 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
23099 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
23100 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
23102 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
23104 PR rtl-optimization/70478
23105 * lra-constraints.c: Reverse the last patch.
23107 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
23109 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
23110 Add comment for WCHAR_T.
23112 2017-04-08 Martin Liska <mliska@suse.cz>
23115 2017-04-07 Martin Liska <mliska@suse.cz>
23118 * ipa-split.c (split_function): Add function part to a same comdat
23121 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23124 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
23126 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
23128 * rs6000/rs6000.c (vec_load_pendulum): Rename...
23129 (vec_pairing): ...to this.
23130 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
23131 (rs6000_sched_init): Adjust for name change.
23132 (struct rs6000_sched_context): Likewise.
23133 (rs6000_init_sched_context): Likewise.
23134 (rs6000_set_sched_context): Likewise.
23136 2017-04-07 Jakub Jelinek <jakub@redhat.com>
23142 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
23144 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
23145 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
23146 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
23148 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
23150 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
23152 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23154 PR rtl-optimization/70703
23155 * ira-color.c (update_conflict_hard_regno_costs): Use
23156 int64_t instead of HOST_WIDE_INT.
23158 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23160 PR rtl-optimization/70478
23161 * lra-constraints.c (process_alt_operands): Disfavor alternative
23162 insn memory operands.
23164 2017-04-07 Jeff Law <law@redhat.com>
23166 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
23167 CALL and NOTE_INSN_CALL_ARG_LOCATION.
23169 2017-04-07 Martin Liska <mliska@suse.cz>
23172 * config/aarch64/aarch64.c (aarch64_process_target_attr):
23173 Show error message instead of an ICE.
23175 2017-04-07 Martin Liska <mliska@suse.cz>
23178 * ipa-split.c (split_function): Add function part to a same comdat
23181 2017-04-07 Richard Biener <rguenther@suse.de>
23183 PR middle-end/80341
23184 * tree.c (get_unwidened): Also handle ! for_type case for
23186 * convert.c (do_narrow): Split out from ...
23187 (convert_to_integer_1): ... here. Do not pass final truncation
23188 type to get_unwidened for TRUNC_DIV_EXPR.
23190 2017-04-07 Richard Biener <rguenther@suse.de>
23192 * tree-affine.c (wide_int_ext_for_comb): Take type rather
23194 (aff_combination_const): Adjust.
23195 (aff_combination_scale): Likewise.
23196 (aff_combination_add_elt): Likewise.
23197 (aff_combination_add_cst): Likewise.
23198 (aff_combination_convert): Likewise.
23199 (add_elt_to_tree): Likewise. Remove unused argument.
23200 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
23202 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
23204 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
23206 * config/arm/arm.c (arm_default_short_enums): Use
23207 ARM_DEFAULT_SHORT_ENUMS.
23208 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
23210 2017-04-06 Jakub Jelinek <jakub@redhat.com>
23213 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
23214 members with redundant out-of-class redeclaration.
23216 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23219 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
23220 * config/i386/i386.md (*zero_extendsidi2):
23221 Add (?*x,*x) and (?*v,*v) alternatives.
23223 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23226 * config/i386/i386.c (ix86_expand_builtin)
23227 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
23228 mode from insn data. Convert operands to insn operand mode.
23229 Copy operands that don't satisfy insn predicate to a register.
23231 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
23233 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
23236 2017-04-06 Richard Biener <rguenther@suse.de>
23238 PR tree-optimization/80334
23239 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
23240 preserve alignment of accesses.
23242 2017-04-06 Richard Biener <rguenther@suse.de>
23244 PR tree-optimization/80262
23245 * tree-sra.c (build_ref_for_offset): Preserve address-space
23247 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
23248 Drop useless address-space information on MEM_REF offsets.
23250 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
23252 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
23254 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23256 PR rtl-optimization/70703
23257 * ira-color.c (update_conflict_hard_regno_costs): Use
23258 HOST_WIDE_INT instead of long.
23260 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
23263 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
23264 not defined for x86_64 target. Add -mmmx target option when __SSE2__
23266 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
23267 for x86_64 target. Handle -m3dnowa option.
23269 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23271 PR rtl-optimization/70703
23272 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
23273 (update_conflict_hard_regno_costs): Use long instead of unsigned
23274 arithmetic for cost calculation.
23276 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23277 Bernd Edlinger <bernd.edlinger@hotmail.de>
23280 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
23283 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
23286 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
23287 ptr_mode with Pmode throughout.
23288 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
23289 into probe_stack_range and use DImode.
23291 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
23294 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
23295 call_eh_return is true.
23297 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23299 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
23300 Initialize last_match_fntype_index.
23302 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23305 * tree-nvr.c: Include internal-fn.h.
23306 (pass_return_slot::execute): Ignore internal calls without
23309 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23310 Richard Biener <rguenther@suse.de>
23313 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
23314 captures used multiple times, except for the last use.
23315 * generic-match-head.c: Include gimplify.h.
23317 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23319 PR tree-optimization/79390
23320 * target.h (struct noce_if_info): Declare.
23321 * targhooks.h (default_noce_conversion_profitable_p): Declare.
23322 * target.def (noce_conversion_profitable_p): New target hook.
23323 * ifcvt.h (struct noce_if_info): New type, moved from ...
23324 * ifcvt.c (struct noce_if_info): ... here.
23325 (noce_conversion_profitable_p): Renamed to ...
23326 (default_noce_conversion_profitable_p): ... this. No longer
23328 (noce_try_store_flag_constants, noce_try_addcc,
23329 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
23330 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
23331 instead of noce_conversion_profitable_p.
23332 * config/i386/i386.c: Include ifcvt.h.
23333 (ix86_option_override_internal): Don't override
23334 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
23335 (ix86_noce_conversion_profitable_p): New function.
23336 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
23337 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
23338 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
23339 * doc/tm.texi: Regenerated.
23341 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23343 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
23346 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23349 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
23350 instructions for small multiply cores.
23352 2017-04-04 Jeff Law <law@redhat.com>
23354 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
23356 (mips_expand_vec_perm_const): Initialize elements in orig_perm
23357 that are not set by the loop over the elements.
23359 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23362 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
23363 int mode, convert_modes it to mode as unsigned, otherwise use
23364 lowpart_subreg to mode rather than SImode.
23365 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
23366 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
23367 Use DImode instead of SImode for the shift count operand.
23368 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
23371 2017-04-04 Richard Biener <rguenther@suse.de>
23373 PR middle-end/80281
23374 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
23375 arithmetic done for the negate or the plus. Simplify.
23376 (A - (-B) -> A + B): Likewise.
23377 * fold-const.c (split_tree): Make sure to not negate pointers.
23379 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
23381 PR rtl-optimization/60818
23382 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
23383 a compare of comparisons with the thing compared if this results
23384 in a different machine mode.
23386 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
23388 * alias.c (base_alias_check): Fix typo in comment.
23389 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
23390 * cgraphunit.c (symbol_table::compile): Likewise.
23391 * collect2.c (maybe_run_lto_and_relink): Likewise.
23392 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
23393 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
23394 * config/avr/avr.c (avr_map_op_t): Likewise.
23395 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
23396 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
23397 * config/epiphany/epiphany.md (movcc): Likewise.
23398 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
23399 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
23401 * config/mips/mips.c (mips_save_restore_reg): Likewise.
23402 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
23403 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
23404 * config/sh/sh.c (sh_rtx_costs): Likewise.
23405 * fold-const.c (fold_truth_andor): Likewise.
23406 * genautomata.c (collapse_flag): Likewise.
23407 * gengtype.h (struct type::u::s): Likewise.
23408 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
23409 * input.c (FORMAT_AMOUNT): Likewise.
23410 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
23411 (known_aggs_to_agg_replacement_list): Likewise.
23412 * ipa-inline-analysis.c: Likewise.
23413 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
23414 * ipa-polymorphic-call.c
23415 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
23416 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
23417 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
23419 * modulo-sched.c (apply_reg_moves): Likewise.
23420 * omp-expand.c (build_omp_regions_1): Likewise.
23421 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
23422 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
23423 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
23424 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
23425 * value-prof.c: Likewise.
23426 * var-tracking.c (val_reset): Likewise.
23428 2017-04-03 Richard Biener <rguenther@suse.de>
23430 PR tree-optimization/80275
23431 * fold-const.c (split_address_to_core_and_offset): Handle
23434 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
23436 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
23437 descriptors is at least equal to that of functions.
23439 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23441 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
23443 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23446 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
23447 (mov<IMOD4:mode>): New expander.
23448 (*mov<IMOD4:mode>_internal): New insn and split pattern.
23450 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
23452 PR rtl-optimization/79405
23453 * fwprop.c (propagations_left): New variable.
23454 (forward_propagate_into): Decrement it.
23455 (fwprop_init): Initialize it.
23456 (fw_prop): If the variable has reached zero, stop propagating.
23457 (fwprop_addr): Ditto.
23459 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23462 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
23463 a FUNCTION_DECL, pass it as decl instead of origin to
23466 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
23468 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
23471 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
23474 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
23475 TARGET_VSX_SMALL_INTEGER.
23477 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23479 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23480 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
23482 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23484 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
23485 extraction from odd-numbered MSA register.
23487 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23489 PR middle-end/80173
23490 * expmed.c (store_bit_field_1): Don't attempt to create
23491 a word subreg out of hard registers wider than word if they
23492 have HARD_REGNO_NREGS of 1 for their mode.
23494 PR middle-end/80163
23495 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
23496 conversions to integer types wider than word and pointer.
23499 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
23500 (rtx_equal_for_cselib_p): Pass 0 to it.
23501 * cselib.c (cselib_hasher::equal): Likewise.
23502 (rtx_equal_for_cselib_1): Add depth argument. If depth
23503 is 128, don't look up VALUE locs and punt. Increment
23504 depth in recursive calls when walking VALUE locs.
23506 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
23508 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
23509 (make_gcov_file_name): Use the canonical path name for generating
23511 (read_line): Fix handling of files with ascii null bytes.
23513 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
23515 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
23516 to initialise a vector register instead
23517 of using a const_int.
23519 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23521 PR translation/80189
23522 * gimplify.c (omp_default_clause): Use %qs instead of %s in
23523 diagnostic messages.
23525 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
23528 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
23529 (dfp_diex_<mode>): Update mode of operand 1.
23530 * doc/extend.texi (dxex, dxexq): Document change to return type.
23531 (diex, diexq): Document change to argument type.
23533 2017-03-30 Martin Jambor <mjambor@suse.cz>
23536 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
23537 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
23538 it reflects the signature changes performed at the callee side.
23539 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
23540 to cgraph_build_function_type_skip_args.
23541 (build_function_decl_skip_args): Adjust call to the above function.
23543 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23546 * config/i386/sse.md
23547 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
23548 register as dest whenever it is a MEM not rtx_equal_p to the
23549 corresponding dup operand, and when forcing into reg move the
23550 reg into the memory afterwards.
23551 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
23552 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
23553 for the force_reg mode.
23554 (avx512vl_vextractf128<mode>): Use register as dest either
23555 always when a MEM, or when it is a MEM not rtx_equal_p to the
23556 corresponding dup operand, or even not when it is a CONST_VECTOR
23557 depending on the mode and lo vs. hi.
23558 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
23560 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
23561 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
23562 Likewise. Require that operands[2] is even.
23563 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
23564 Remove extraneous parens. Require that operands[2] is a multiple
23566 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
23567 operands[0] is a MEM if <mask_applied>, the predicates/constraints
23568 disallow memory then.
23570 2017-03-30 Richard Biener <rguenther@suse.de>
23572 PR tree-optimization/77498
23573 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
23574 to non-constants over backedges.
23576 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
23578 PR rtl-optimization/80233
23579 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
23580 as last_combined_insn. Do not test for BARRIER_P separately.
23582 2017-03-29 Andreas Schwab <schwab@suse.de>
23585 * calls.c (prepare_call_address): Convert funexp to Pmode before
23586 copying to temp reg.
23588 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23590 PR tree-optimization/80158
23591 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
23592 Handle possible future case of more than one alternate
23594 (replace_rhs_if_not_dup): Likewise.
23595 (replace_one_candidate): Likewise.
23597 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
23599 PR rtl-optimization/80193
23600 * ira.c (ira): Do not check allocation for LRA.
23602 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
23604 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
23605 (nvptx_output_simt_exit): Declare.
23606 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
23607 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
23608 (init_softstack_frame): Move initialization of crtl->is_leaf to...
23609 (nvptx_declare_function_name): ...here. Emit declaration of local
23610 memory space buffer for omp_simt_enter insn.
23611 (nvptx_output_unisimt_switch): New.
23612 (nvptx_output_softstack_switch): New.
23613 (nvptx_output_simt_enter): New.
23614 (nvptx_output_simt_exit): New.
23615 * config/nvptx/nvptx.h (struct machine_function): New fields
23616 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
23617 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
23618 (UNSPECV_SIMT_EXIT): Ditto.
23619 (omp_simt_enter_insn): New insn.
23620 (omp_simt_enter): New expansion.
23621 (omp_simt_exit): New insn.
23622 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
23624 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
23625 (expand_GOMP_SIMT_ENTER_ALLOC): New.
23626 (expand_GOMP_SIMT_EXIT): New.
23627 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
23628 (GOMP_SIMT_ENTER_ALLOC): Ditto.
23629 (GOMP_SIMT_EXIT): Ditto.
23630 * target-insns.def (omp_simt_enter): New insn.
23631 (omp_simt_exit): Ditto.
23632 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
23634 (lower_rec_simd_input_clauses): Implement SIMT privatization.
23635 (lower_rec_input_clauses): Likewise.
23636 (lower_lastprivate_clauses): Handle SIMT privatization.
23638 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
23639 (ompdevlow_adjust_simt_enter): New.
23640 (find_simtpriv_var_op): New.
23641 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
23642 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
23644 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
23645 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
23646 (copy_decl_for_dup_finish): Ditto.
23648 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
23650 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
23653 * config/i386/i386.c (ix86_option_override_internal): Always
23654 allow -mpreferred-stack-boundary=3 for 64-bit targets.
23656 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23658 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
23660 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23662 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
23663 mark new edge's irreducible flag accordign to it.
23664 (vect_do_peeling): Check loop preheader edge's irreducible flag
23665 and pass it to function slpeel_add_loop_guard.
23667 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
23669 PR tree-optimization/80218
23670 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23671 Update block frequencies and counts.
23673 2017-03-28 Richard Biener <rguenther@suse.de>
23675 PR tree-optimization/78644
23676 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
23677 of a simplification result we may not use it at all.
23679 2017-03-28 Richard Biener <rguenther@suse.de>
23682 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
23683 without arguments, generate default definition of a SSA name.
23685 2017-03-28 Richard Biener <rguenther@suse.de>
23687 PR middle-end/80222
23688 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
23689 TYPE_REF_CAN_ALIAS_ALL references.
23690 * fold-const.c (fold_indirect_ref_1): Likewise.
23692 2017-03-28 Martin Liska <mliska@suse.cz>
23695 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
23696 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
23698 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23699 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
23701 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
23702 (EXTRA_SPECS): Define.
23703 (SUBTARGET_EXTRA_SPECS): Likewise.
23704 (SUBTARGET_CPP_SPEC): Likewise.
23705 * config/arc/elf.h (EXTRA_SPECS): Renamed to
23706 SUBTARGET_EXTRA_SPECS.
23707 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
23709 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23711 * config/arc/simdext.md (vst64_insn): Update pattern.
23712 (vld32wh_insn): Likewise.
23713 (vld32wl_insn): Likewise.
23714 (vld64_insn): Likewise.
23715 (vld32_insn): Likewise.
23717 2017-03-28 Marek Polacek <polacek@redhat.com>
23720 * fold-const.c (fold_comparison): Use protected_set_expr_location
23721 instead of SET_EXPR_LOCATION.
23723 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
23725 * tree.c (add_expr): Avoid name lookup warning.
23727 2017-03-27 Jeff Law <law@redhat.com>
23729 PR tree-optimization/80216
23730 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
23731 function name. Limit recursion depth.
23732 (record_temporary_equivalences): Corresponding changes.
23734 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
23736 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
23739 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23742 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
23744 * cfgcleanup.c (reg_note_cfa_p): New array.
23745 (insns_have_identical_cfa_notes): New function.
23746 (old_insns_match_p): Don't cross-jump in between /f
23747 and non-/f instructions. If both i1 and i2 are frame related,
23748 verify all CFA notes, their order and content.
23750 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
23753 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
23754 HImode and SImode with zero extend to DImode to one insn.
23755 (bswap<mode>2_extenddi): Likewise.
23756 (bswapsi2_extenddi): Likewise.
23757 (bswaphi2_extendsi): Likewise.
23758 (bswaphi2): Combine bswap HImode and SImode into one insn.
23759 Separate memory insns from swapping register.
23760 (bswapsi2): Likewise.
23761 (bswap<mode>2): Likewise.
23762 (bswaphi2_internal): Delete, no longer used.
23763 (bswapsi2_internal): Likewise.
23764 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
23765 store, and gpr<-gpr swap insns.
23766 (bswap<mode>2_store): Likewise.
23767 (bswaphi2_reg): Register only splitter, combine with the splitter.
23768 (bswaphi2 splitter): Likewise.
23769 (bswapsi2_reg): Likewise.
23770 (bswapsi2 splitter): Likewise.
23771 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
23772 the insns into load, store, and register/register insns.
23773 (bswapdi2_ldbrx): Likewise.
23774 (bswapdi2_load): Likewise.
23775 (bswapdi2_store): Likewise.
23776 (bswapdi2_reg): Likewise.
23778 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
23780 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
23781 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
23783 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
23786 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
23788 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23789 special handling for target option conflicts between dform
23790 options (-mpower9-dform, -mpower9-dform-vector,
23791 -mpower9-dform-scalar) and -mno-direct-move.
23793 2017-03-27 Richard Biener <rguenther@suse.de>
23795 PR tree-optimization/80181
23796 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
23798 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23800 * config/arc/predicates.md (move_double_src_operand): Replace the
23801 call to move_double_src_operand with a call to address_operand.
23803 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23805 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
23806 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
23807 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
23809 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23811 * config/arc/predicates.md (long_immediate_loadstore_operand):
23812 Consider scaled addresses cases.
23814 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23816 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
23817 restored when in interrupt.
23818 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
23819 doesn't have delay slot.
23821 2017-03-27 Richard Biener <rguenther@suse.de>
23824 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
23825 inlined thunk clones.
23827 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23830 * asan.c (instrument_derefs): Copy over last operand from
23831 original COMPONENT_REF to the new COMPONENT_REF with
23832 DECL_BIT_FIELD_REPRESENTATIVE.
23833 * ubsan.c (instrument_object_size): Likewise.
23835 2017-03-27 Richard Biener <rguenther@suse.de>
23837 PR tree-optimization/80170
23838 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
23839 sure DR/SCEV didnt fold in constants we do not see when looking
23840 at the reference base alignment.
23842 2017-03-27 Richard Biener <rguenther@suse.de>
23844 PR middle-end/80171
23845 * gimple-fold.c (fold_ctor_reference): Properly guard against
23846 NULL return value from canonicalize_constructor_val.
23848 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
23851 * config/i386/i386.c (ix86_expand_builtin)
23852 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
23853 flags reg setting and flags reg using instructions.
23854 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
23855 clobbering instructions to zero extend op2.
23857 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
23859 * doc/install.texi (Configuration) <--with-aix-soname>:
23860 Update link to AIX ld.
23862 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
23864 PR rtl-optimization/80160
23865 PR rtl-optimization/80159
23866 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
23867 reg_alternate_class into account.
23869 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
23872 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
23873 to consider in curr_insn_transform.
23875 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23877 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
23878 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
23879 and emit_mode_inner.
23881 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23883 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
23884 argument to the overloaded builtin variants. Use the new flag to
23885 deprecate certain builtin variants.
23886 * config/s390/s390-builtin-types.def: Add new builtin types.
23887 * config/s390/s390-builtins.h: Support new flags field for
23888 overloaded builtins.
23889 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
23890 (s390_macro_to_expand): Enable vector float data type.
23891 (s390_cpu_cpp_builtins_internal): Indicate support of the new
23892 builtins by incrementing the __VEC__ version number.
23893 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
23895 (s390_resolve_overloaded_builtin): Emit error messages depending
23896 on the builtin flags.
23897 * config/s390/s390.c (s390_expand_builtin): Support additional
23898 flags argument. Change error message to match the messages
23899 emitted in s390-c.c.
23900 * config/s390/s390.md: New UNSPEC_* constants.
23901 (op_type): Add new instruction types.
23902 * config/s390/vecintrin.h: Add new builtins and test data class
23904 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
23905 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
23906 (VEC_INEXACT, VEC_NOINEXACT): New constants.
23907 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
23908 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
23909 ("vec_mergel<mode>"): V_HW -> VEC_HW.
23911 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
23912 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
23913 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
23914 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
23916 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
23917 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
23918 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
23919 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
23921 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
23922 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
23923 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
23924 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
23925 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
23926 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
23927 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
23929 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
23930 ("vec_scatter_element<V_HW_4:mode>_DI")
23931 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
23932 ("vec_fpint<mode>", "vflls")
23933 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
23934 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
23935 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
23936 ("*vec_cmphe<mode>_cc"): ... these.
23938 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
23939 mode constant instead of magic value.
23941 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23943 * config/s390/s390.c (s390_expand_vec_compare): Support other
23944 vector floating point modes than just V2DF.
23945 (s390_expand_vcond): Likewise.
23946 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
23947 (s390_cannot_change_mode_class): Prevent mode changes between TF
23948 and V1TF in vector registers.
23949 * config/s390/s390.md (DF, SF): New mode attributes.
23950 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
23951 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
23952 SFmode support for VRs.
23953 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
23955 (VFT, VF_HW): New mode iterators.
23956 (vw, sdx): New mode attributes.
23957 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
23958 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
23959 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
23960 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
23961 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
23962 also the new vector floating point modes. Renaming to ...
23964 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
23965 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
23966 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
23967 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
23968 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
23969 ("vec_unordered<mode>"): ... these.
23971 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
23972 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
23973 ("*vec_extendv2df"): New insn definitions.
23975 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23977 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
23978 ("mulditi3_2", "*muldi3_sign"): New patterns.
23979 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
23980 rename the pattern definition.
23982 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23984 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
23986 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
23988 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23990 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
23991 instruction if possible.
23992 * config/s390/vector.md (vec_halfnumelts): New mode
23994 ("*vec_vllezlf<mode>"): New pattern.
23996 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23998 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
23999 ("popcountv4si2", "popcountv2di2"): Rename to ...
24000 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
24001 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
24003 ("popcount<mode>2_vxe"): New pattern.
24005 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24007 * common/config/s390/s390-common.c (processor_flags_table): Add
24009 * config.gcc: Add arch12.
24010 * config/s390/driver-native.c (s390_host_detect_local_cpu):
24011 Default to arch12 for unknown CPU model numbers.
24012 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
24013 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
24014 PROCESSOR_max sanity check.
24015 * config/s390/s390-opts.h (enum processor_type): Add
24017 * config/s390/s390.c (processor_table): Add arch12.
24018 (s390_expand_builtin): Add check for B_VXE flag.
24019 (s390_issue_rate): Add PROCESSOR_ARCH12.
24020 (s390_get_sched_attrmask): Likewise.
24021 (s390_get_unit_mask): Likewise.
24022 (s390_sched_score): Enable z13 scheduling for arch12.
24023 (s390_sched_reorder): Likewise.
24024 (s390_sched_variable_issue): Likewise.
24025 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
24027 (s390_tune_attr): Use z13 scheduling also for arch12.
24028 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
24029 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
24030 (TARGET_VXE_P): New macros.
24031 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
24032 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
24033 * config/s390/s390.opt: Add arch12 as processor_type.
24035 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24037 * config/s390/s390.md
24038 ("fixuns_truncdddi2", "fixuns_trunctddi2")
24039 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
24040 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
24042 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
24043 Rename expanders to ...
24045 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
24046 ("fixuns_truncdddi2_emu"): ... these.
24048 ("fixuns_trunc<mode>si2_emu"): New expander.
24050 ("*fixuns_truncdfdi2_z13"): Rename to ...
24051 ("*fixuns_truncdfdi2_vx"): ... this.
24053 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24055 * config/s390/2964.md: Remove the single element vector compare
24056 instructions which are no longer used.
24057 * config/s390/s390.c (s390_select_ccmode): Remove handling of
24059 (s390_canonicalize_comparison): Remove handling of DFmode
24061 (s390_expand_vec_compare_scalar): Remove function.
24062 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
24063 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
24065 ("*cmp<mode>_ccs"): Add wfcdb instruction.
24067 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24069 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
24071 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
24072 will anyway by matched by mov<mode>_64dfp.
24074 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24076 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
24077 vlef/vstef. Add missing operand to vleif.
24079 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24081 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
24082 pair for all vector types with 64 bit elements.
24083 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
24084 * config/s390/vector.md (V_HW_64): ... here.
24085 (V_128_NOSINGLE): New mode iterator.
24086 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
24087 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
24088 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
24089 ("*vec_load_pairv2di"): Change to ...
24090 ("*vec_load_pair<mode>"): ... this one.
24092 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24094 * config/s390/constraints.md: Add comments.
24095 (jKK): Reject element sizes > 8 bytes.
24096 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
24098 * config/s390/s390.md: Add the s_operand checks formerly in
24099 s390_split_ok_p to various splitters where they are still
24101 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
24102 for 128 bit vectors. Plus two splitters.
24104 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24106 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
24109 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24112 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
24113 error if the boundary argument is not constant.
24115 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24117 PR rtl-optimization/80112
24118 * loop-doloop.c (doloop_condition_get): Don't check condition
24119 if cmp isn't SET with IF_THEN_ELSE src.
24121 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24123 PR tree-optimization/80158
24124 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
24125 replacing a candidate statement, also replace it for the
24126 candidate's alternate interpretation.
24127 (replace_rhs_if_not_dup): Likewise.
24128 (replace_one_candidate): Likewise.
24130 2017-03-24 Richard Biener <rguenther@suse.de>
24132 PR tree-optimization/80167
24133 * graphite-isl-ast-to-gimple.c
24134 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
24136 (translate_isl_ast_to_gimple::get_rename): Likewise.
24138 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24140 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
24141 handling of certain combinations of target options, including the
24142 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
24143 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
24145 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24148 * config/arm/arm.md (*load_multiple): Add reload_completed to
24149 matching condition.
24151 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24152 Richard Biener <rguenth@suse.de>
24154 PR tree-optimization/79908
24155 PR tree-optimization/80136
24156 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24157 been cast away, gimplify_and_add suffices.
24159 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
24161 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
24163 2017-03-23 Richard Biener <rguenther@suse.de>
24165 PR tree-optimization/80032
24166 * gimplify.c (gimple_push_cleanup): Forced unconditional
24167 cleanups still have to go to the conditional_cleanups
24170 2017-03-22 Jakub Jelinek <jakub@redhat.com>
24172 PR tree-optimization/80072
24173 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
24175 (next_operand_entry_id): Change type to unsigned int.
24176 (sort_by_operand_rank): Make sure to return the right return value
24177 even if unsigned fields are bigger than INT_MAX.
24178 (struct oecount): Change cnt and id type to unsigned int.
24179 (oecount_hasher::equal): Formatting fix.
24180 (oecount_cmp): Make sure to return the right return value
24181 even if unsigned fields are bigger than INT_MAX.
24182 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
24185 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
24186 TREE_READONLY on result if writing it more than once.
24189 * doc/invoke.texi (-fsanitize=thread): Document that with
24190 -fnon-call-exceptions atomics are not able to throw
24194 * tsan.c: Include tree-eh.h.
24195 (instrument_builtin_call): Call maybe_clean_eh_stmt or
24196 maybe_clean_or_replace_eh_stmt where needed.
24197 (instrument_memory_accesses): Add cfg_changed argument.
24198 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
24199 if it returned true.
24200 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
24202 PR rtl-optimization/63191
24203 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
24204 wrapper function, moved the whole old content into ...
24205 (ix86_delegitimize_address_1): ... this. New inline function.
24206 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
24207 true as last argument instead of ix86_delegitimize_address.
24209 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24211 * config/aarch64/aarch64.c (generic_branch_cost): Copy
24212 cortexa57_branch_cost.
24214 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24216 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
24218 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24221 * doc/md.texi (Constraints): Document wA constraint.
24222 * config/rs6000/constraints.md (wA): New.
24223 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
24224 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
24225 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
24226 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
24228 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
24231 * gimplify.c (is_oacc_declared): New function.
24232 (oacc_default_clause): Use it to set default flags for acc declared
24233 variables inside parallel regions.
24234 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
24235 declared variables.
24236 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
24237 declare attribute to any decl as necessary.
24239 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24242 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
24243 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
24244 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
24245 (arm_arch_lpae): This.
24246 * config/arm/arm.c (arm_arch7ve): Rename into ...
24247 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
24248 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
24251 2017-03-22 Martin Liska <mliska@suse.cz>
24254 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
24255 error message instead of an ICE.
24257 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24259 * doc/extend.texi (6.11 Additional Floating Types): Revise.
24261 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
24263 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24265 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24268 2017-03-21 Martin Sebor <msebor@redhat.com>
24270 * doc/extend.texi: Use "cannot" instead of "can't."
24271 * doc/hostconfig.texi: Same.
24272 * doc/install.texi: Same.
24273 * doc/invoke.texi: Same.
24274 * doc/loop.texi: Same.
24275 * doc/md.texi: Same.
24276 * doc/objc.texi: Same.
24277 * doc/rtl.texi: Same.
24278 * doc/tm.texi: Same.
24279 * doc/tm.texi.in: Same.
24280 * doc/trouble.texi: Same.
24282 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
24285 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
24286 (collect_checksum_attributes): Set it.
24287 (die_checksum_ordered): Use it.
24289 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24291 PR tree-optimization/79908
24292 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
24293 change: For a VA_ARG whose LHS has been cast away, use
24294 force_gimple_operand to construct the side effects.
24296 2017-03-21 David Malcolm <dmalcolm@redhat.com>
24298 PR translation/80001
24299 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
24300 more amenable to translation.
24301 (oacc_loop_auto_partitions): Likewise.
24303 2017-03-21 Marek Polacek <polacek@redhat.com>
24304 Martin Sebor <msebor@redhat.com>
24306 PR tree-optimization/80109
24307 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
24308 on INTEGRAL_TYPE_P.
24310 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24311 Segher Boessenkool <segher@kernel.crashing.org>
24314 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
24315 check reg_used_between_p between insn and one of succ or succ2
24316 depending on if succ is artificial insn not inserted into insn
24319 2017-03-21 Martin Liska <mliska@suse.cz>
24321 PR gcov-profile/80081
24322 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
24323 * doc/gcc.texi: Include gcov-dump stuff.
24324 * doc/gcov-dump.texi: New file.
24326 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
24328 PR rtl-optimization/79150
24329 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
24330 conditional jump, if the jump is the last insn of the loop.
24332 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24333 Richard Biener <rguenth@suse.de>
24335 PR tree-optimization/79908
24336 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24337 been cast away, use force_gimple_operand to construct the side
24340 2017-03-21 Martin Liska <mliska@suse.cz>
24342 PR libfortran/79956
24343 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
24346 2017-03-21 Brad Spengler <spender@grsecurity.net>
24349 * plugin.c (htab_hash_plugin): New function.
24350 (add_new_plugin): Use it and adjust.
24351 (parse_plugin_arg_opt): Adjust.
24352 (init_one_plugin): Likewise.
24354 2017-03-21 Richard Biener <rguenther@suse.de>
24356 PR tree-optimization/80032
24357 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
24358 if set force the cleanup to happen unconditionally.
24359 (gimplify_target_expr): Push inserted clobbers with force_uncond
24360 to avoid them being removed by control-dependent DCE.
24362 2017-03-21 Richard Biener <rguenther@suse.de>
24364 PR tree-optimization/80122
24365 * tree-inline.c (copy_bb): Do not expans va-arg packs or
24366 va_arg_pack_len when the inlined call stmt requires pack
24368 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
24370 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24373 * tsan.c (instrument_builtin_call): If the memory model argument
24374 is not a constant, assume it is valid.
24377 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
24380 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24382 PR rtl-optimization/79910
24383 * combine.c (can_combine_p): Do not allow combining an I0 or I1
24384 if its dest is used by an insn before I2 (other than the combined
24385 insns themselves, which are properly handled already).
24387 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24390 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24392 * combine.c (record_used_regs): New static function.
24393 (try_combine): Handle situations where there is an additional
24394 instruction between I2 and I3 which needs to have a LOG_LINK
24398 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24400 * combine.c (try_combine): Delete redundant i1 test. Call
24401 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24403 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24406 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
24407 alternatives 13/14.
24409 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24411 PR tree-optimization/80054
24412 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
24413 the optimization if a PHI or any of its arguments is not dominated
24414 by the candidate's basis. Use gphi* rather than gimple* as
24416 (replace_profitable_candidates): Clean up a gimple* variable that
24417 should be a gphi* variable.
24419 2017-03-20 Martin Sebor <msebor@redhat.com>
24422 * doc/extend.texi (attribute constructor): Document present limitation.
24424 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
24427 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
24428 __POWER9_VECTOR__ #ifdef control, change template definition to
24429 use Power9-specific built-in function.
24430 (vec_any_eq): Likewise.
24431 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
24432 to control outcomes from this test.
24433 (vector_ae_<mode>p): For VEC_F modes, likewise.
24435 2017-03-20 Ian Lance Taylor <iant@google.com>
24437 * config/i386/i386.c (ix86_function_regparm): Save an extra
24438 register for -fsplit-stack with DECL_STATIC_CHAIN.
24440 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24443 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
24444 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
24446 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24448 * config/riscv/riscv.c (riscv_print_operand): Use "fence
24450 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
24453 2017-03-20 Marek Polacek <polacek@redhat.com>
24456 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
24458 2017-03-20 Richard Biener <rguenther@suse.de>
24460 PR tree-optimization/80113
24461 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
24462 allocate extra SSA name for PHI def.
24463 (add_close_phis_to_outer_loops): Likewise.
24464 (add_close_phis_to_merge_points): Likewise.
24465 (copy_loop_close_phi_args): Likewise.
24466 (copy_cond_phi_nodes): Likewise.
24468 2017-03-20 Martin Liska <mliska@suse.cz>
24470 PR middle-end/79753
24471 * tree-chkp.c (chkp_build_returned_bound): Do not build
24472 returned bounds for a LHS that's not a BOUNDED_P type.
24474 2017-03-20 Martin Liska <mliska@suse.cz>
24478 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
24479 COMPLEX_CST and VECTOR_CST.
24481 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24484 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
24485 target operand. A new splitter adds the clobber statement in case
24486 the target operand is dead anyway.
24488 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
24490 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
24491 to age-old versions of binutils and glibc.
24493 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
24495 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
24497 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24499 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
24501 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24503 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
24504 requirement for binutils 2.13.
24506 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24508 * combine.c (try_combine): Delete redundant i1 test. Call
24509 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24511 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
24513 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
24514 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
24516 <riscv64-*-elf>: Re-arrange section
24517 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
24518 <riscv32-*-linux>: Likewise.
24519 <riscv64-*-elf>: Likewise
24520 <riscv64-*-linux>: Likewise.
24522 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
24525 * aarch64.opt(verbose-cost-dump): Fix typo.
24527 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
24530 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
24531 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
24533 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24535 * reload.c (find_reloads): When reloading a nonoffsettable address,
24536 use RELOAD_OTHER for it and its address reloads.
24538 PR rtl-optimization/79910
24539 * combine.c (record_used_regs): New static function.
24540 (try_combine): Handle situations where there is an additional
24541 instruction between I2 and I3 which needs to have a LOG_LINK
24544 2017-03-17 Jeff Law <law@redhat.com>
24546 PR tree-optimization/71437
24547 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
24548 conditional in the hash table first.
24549 (vrp_dom_walker::before_dom_children): Extract condition from
24550 ASSERT_EXPR. Record condition, its inverion and any implied
24551 conditions as well.
24553 2017-03-17 Marek Polacek <polacek@redhat.com>
24554 Markus Trippelsdorf <markus@trippelsdorf.de>
24556 PR tree-optimization/80079
24557 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
24560 2017-03-17 Richard Biener <rguenther@suse.de>
24562 PR middle-end/80075
24563 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
24564 Properly verify the LHS before the RHS possibly claims to be
24566 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
24569 2017-03-17 Martin Jambor <mjambor@suse.cz>
24571 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
24572 (List of -O2 options): Likewise.
24573 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
24576 2017-03-17 Tom de Vries <tom@codesourcery.com>
24578 * gcov-dump.c (print_usage): Print bug_report_url.
24580 2017-03-17 Richard Biener <rguenther@suse.de>
24582 PR middle-end/80050
24583 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
24584 (parser::peek): Likewise.
24586 2017-03-17 Richard Biener <rguenther@suse.de>
24588 PR tree-optimization/80048
24589 * sese.c (free_sese_info): Properly release rename_map and
24590 copied_bb_map elements.
24592 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
24594 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
24595 Add linked-list forward and backlinks. Insert on
24596 construction, remove on destruction.
24597 (class pass_store_merging): Add m_stores_head field.
24598 (pass_store_merging::terminate_and_process_all_chains):
24599 Iterate over m_stores_head list.
24600 (pass_store_merging::terminate_all_aliasing_chains):
24602 (pass_store_merging::execute): Check for debug stmts first.
24603 Push new chains onto the m_stores_head stack.
24605 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24608 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
24609 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
24610 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
24612 2017-03-16 Jeff Law <law@redhat.com>
24614 PR tree-optimization/71437
24615 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
24616 member function. Implementation moved into after_dom_children
24617 member function and into the threader's thread_outgoing_edges
24619 (dom_opt_dom_walker::after_dom_children): Simplify by moving
24620 some code into new thread_outgoing_edges.
24621 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
24622 definition. Simplify marker handling (do it here). Assume we always
24623 have the available expression and the const/copies tables.
24624 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
24626 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
24627 * tree-vrp.c (equiv_stack): No longer file scoped.
24628 (vrp_dom_walker): New class.
24629 (vrp_dom_walker::before_dom_children): New member function.
24630 (vrp_dom_walker::after_dom_children): Likewise.
24631 (identify_jump_threads): Setup domwalker. Use it rather than
24632 walking edges in a random order by hand. Simplify setup/finalization.
24633 (finalize_jump_threads): Remove.
24634 (vrp_finalize): Do not call identify_jump_threads here.
24635 (execute_vrp): Do it here instead and call thread_through_all_blocks
24638 PR tree-optimization/71437
24639 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
24641 (simplify_stmt_for_jump_threading): Add basic_block argument. All
24643 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
24644 (dom_opt_dom_walker::thread_across_edge): Remove
24645 handle_dominating_asserts argument. All callers changed.
24646 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
24647 changes. Remove calls to lhs_of_dominating_assert. Other
24648 uses of handle_dominating_asserts turn into unconditional code
24649 (simplify_control_stmt_condition_1): Likewise.
24650 (simplify_control_stmt_condition): Likewise.
24651 (thread_through_normal_block, thread_across_edge): Likewise.
24652 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
24653 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
24654 object if it is not an SSA_NAME.
24655 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
24656 before calling into the VRP specific simplifiers.
24657 (identify_jump_threads): Remove handle_dominating_asserts
24660 2017-03-16 Jakub Jelinek <jakub@redhat.com>
24663 * tree-diagnostic.c (default_tree_printer): No longer static.
24664 * tree-diagnostic.h (default_tree_printer): New prototype.
24666 2017-03-16 Tamar Christina <tamar.christina@arm.com>
24668 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
24669 Change ins into fmov.
24671 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24673 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
24674 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
24675 Use h_con constraint for operand 1.
24676 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
24677 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
24679 2017-03-15 Jeff Law <law@redhat.com>
24681 PR tree-optimization/71437
24682 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
24683 (record_temporary_equivalences): Use it.
24685 PR tree-optimization/71437
24686 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
24687 tree-ssa-scopedtables.
24688 (lookup_avail_expr, build_and_record_new_cond): Likewise.
24689 (record_conditions, record_cond, vuse_eq): Likewise.
24690 (record_edge_info): Adjust to API tweak of record_conditions.
24691 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
24692 (record_temporary_equivalences, optimize_stmt): Likewise.
24693 (eliminate_redundant_computations): Likewise.
24694 (record_equivalences_from_stmt): Likewise.
24695 * tree-ssa-scopedtables.c: Include options.h and params.h.
24696 (vuse_eq): New function, moved from tree-ssa-dom.c
24697 (build_and_record_new_cond): Likewise.
24698 (record_conditions): Likewise. Accept vector of conditions rather
24699 than edge_equivalence structure for first argument.
24700 for the first argument.
24701 (avail_exprs_stack::lookup_avail_expr): New member function, moved
24702 from tree-ssa-dom.c.
24703 (avail_exprs_stack::record_cond): Likewise.
24704 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
24705 from tree-ssa-dom.c.
24706 (avail_exprs_stack): Add new member functions lookup_avail_expr
24708 (record_conditions): Declare.
24710 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
24713 * lra-constraints.c (process_alt_operands): Increase reject for
24714 reloading an input/output operand.
24716 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
24719 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
24720 insns to convert from signed/unsigned char/short to IEEE 128-bit
24722 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
24724 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
24727 * config/i386/i386.c (ix86_vector_duplicate_value): Create
24728 subreg of inner mode for values already in registers.
24730 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
24732 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
24733 iteration reg is used after the loop.
24735 2017-03-14 Martin Sebor <msebor@redhat.com>
24737 PR tree-optimization/79800
24738 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
24739 precision in negative-positive range.
24740 (format_floating): Call non-const overload with adjusted precision.
24742 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24745 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
24748 2017-03-14 Martin Sebor <msebor@redhat.com>
24750 PR middle-end/80020
24751 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
24752 * builtins.def (aligned_alloc): Use it.
24755 * Makefile.in (GTFILES): Add calls.c.
24756 * calls.c: Include "gt-calls.h".
24758 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
24760 PR rtl-optimization/79728
24761 * regs.h (struct target_regs): New field
24762 x_contains_allocatable_regs_of_mode.
24763 (contains_allocatable_regs_of_mode): New macro.
24764 * reginfo.c (init_reg_sets_1): Initialize it, and change
24765 contains_reg_of_mode so it includes global regs as well.
24766 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
24767 rather than contains_regs_of_mode.
24769 2017-03-14 Martin Liska <mliska@suse.cz>
24771 * doc/invoke.texi: Document options that can't be combined with
24772 -fcheck-pointer-bounds.
24774 2017-03-14 Martin Liska <mliska@suse.cz>
24776 PR middle-end/79831
24777 * doc/invoke.texi (-Wchkp): Document the option.
24779 2017-03-14 Martin Liska <mliska@suse.cz>
24781 * Makefile.in: Install gcov-dump.
24783 2017-03-14 Martin Liska <mliska@suse.cz>
24785 * multiple_target.c (expand_target_clones): Bail out for
24786 an invalid attribute.
24788 2017-03-14 Richard Biener <rguenther@suse.de>
24790 * alias.c (struct alias_set_entry): Pack properly.
24791 * cfgloop.h (struct loop): Likewise.
24792 * cse.c (struct set): Likewise.
24793 * ipa-utils.c (struct searchc_env): Likewise.
24794 * loop-invariant.c (struct invariant): Likewise.
24795 * lra-remat.c (struct cand): Likewise.
24796 * recog.c (struct change_t): Likewise.
24797 * rtl.h (struct address_info): Likewise.
24798 * symbol-summary.h (function_summary): Likewise.
24799 * tree-loop-distribution.c (struct partition): Likewise.
24800 * tree-object-size.c (struct object_size_info): Likewise.
24801 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
24802 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
24803 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
24804 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
24805 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
24806 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
24807 (struct _stmt_vec_info): Likewise.
24809 2017-03-14 Martin Liska <mliska@suse.cz>
24812 * multiple_target.c (create_dispatcher_calls): Check that
24813 a target can create a function dispatcher.
24815 2017-03-14 Martin Liska <mliska@suse.cz>
24818 * multiple_target.c (expand_target_clones): Drop local.local
24819 flag for default implementation.
24821 2017-03-14 Richard Biener <rguenther@suse.de>
24823 PR tree-optimization/80030
24824 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
24826 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
24828 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
24829 gcc_fallthrough() instead of __attribute__((fallthrough));
24831 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24833 * doc/gcc.texi: Remove "up" link to (DIR).
24834 * doc/gccint.texi: Ditto.
24836 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24838 * doc/install.texi (Specific) <avr>: Remove reference to
24841 2017-03-13 Jeff Law <law@redhat.com>
24843 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
24844 attribute rather than comments.
24846 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
24847 match_scratch operand is highest.
24849 2017-03-13 Martin Liska <mliska@suse.cz>
24851 PR middle-end/78339
24852 * ipa-pure-const.c (warn_function_noreturn): If the declarations
24853 is a CHKP clone, use original declaration.
24855 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24857 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
24858 (arc_conditional_register_usage): Use a different allocation order
24859 when optimizing for size.
24860 * common/config/arc/arc-common.c (arc_option_optimization_table):
24861 Section anchors default on when optimizing for size.
24863 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24865 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
24867 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24869 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
24870 * config/arc/arc.md (cpu_facility): Add cd variant.
24871 (*movqi_insn): Add code density variant.
24872 (*movhi_insn): Likewise.
24873 (*movqi_insn): Likewise.
24874 (*addsi3_mixed): Likewise.
24875 (subsi3_insn): Likewise.
24877 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24879 * config/arc/arc.md (movsi_cond_exec): Update constraint.
24881 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24883 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
24884 expressions with MINUS and UNARY ops.
24886 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24889 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
24891 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
24892 between vec_select and vector argument.
24893 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
24894 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
24895 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
24896 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
24897 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
24898 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
24900 2017-03-13 Richard Biener <rguenther@suse.de>
24903 * params.def (vect-max-peeling-for-alignment): Fix typo.
24905 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24907 * doc/install.texi (Specific) <mips-*-*>: Remove description of
24908 issue that only occurred with binutils below 2.18.
24910 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24912 * doc/install.texi (Specific) <cris-axis-elf>: No longer
24913 refer to binutils 2.11/2.12 minimum.
24915 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24917 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
24918 ftp.kernel.org and simplify binutils requirement.
24920 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
24922 * doc/invoke.texi (Warning Options): Fix spelling of link-time
24924 (Optimize Options): Ditto. Also remove redundancy.
24926 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24928 PR translation/79848
24929 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
24931 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
24932 to G_ to avoid double translation.
24934 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24936 PR translation/79923
24937 * auto-profile.c (get_combined_location): Convert leading
24938 character of diagnostics to lower case and remove trailing period.
24939 (read_profile): Likewise for various diagnostics.
24940 * config/arm/arm.c (arm_option_override): Remove trailing period
24941 from various diagnostics.
24942 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
24943 (msp430_expand_delay_cycles): Likewise.
24945 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24948 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
24949 full command-line argument, rather than just "str".
24950 (aarch64_validate_march): Likewise.
24951 (aarch64_validate_mtune): Likewise.
24953 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
24955 PR rtl-optimization/78911
24956 * lra-assigns.c (must_not_spill_p): New function.
24957 (spill_for): Use it.
24959 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24961 PR tree-optimization/79981
24962 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
24963 ATOMIC_COMPARE_EXCHANGE ifn result.
24964 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
24965 IFN_ATOMIC_COMPARE_EXCHANGE.
24967 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24970 * opts.c (parse_sanitizer_options): Add missing question mark to
24971 "did you mean" message.
24973 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24975 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
24977 (VMULEUH_UNS): Likewise.
24978 (VMULOUB_UNS): Likewise.
24979 (VMULOUH_UNS): Likewise.
24980 * config/rs6000/rs6000.c (builtin_function_type): Remove
24981 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
24983 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24986 * read-rtl-function.c (function_reader::read_rtx_operand): Update
24987 x with result of extra_parsing_for_operand_code_0.
24988 (function_reader::extra_parsing_for_operand_code_0): Convert
24989 return type from void to rtx, returning x. When reading
24990 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
24991 larger size containing struct block_symbol.
24993 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
24995 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
24996 -mfloat128-hardware without -m64.
24998 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
25001 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
25002 entries to the case statement that marks unsigned arguments to
25003 overloaded functions.
25005 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25007 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
25008 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
25010 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
25013 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
25014 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
25016 2017-03-10 Martin Liska <mliska@suse.cz>
25020 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
25021 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
25024 2017-03-10 Olivier Hainque <hainque@adacore.com>
25026 * tree-switch-conversion (array_value_type): Start by resetting
25027 candidate type to it's main variant.
25029 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25031 PR rtl-optimization/79909
25032 * combine.c (try_combine): Use simplify_replace_rtx on individual
25033 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
25034 of the whole CALL_INSN_FUNCTION_USAGE.
25036 PR tree-optimization/79972
25037 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
25038 get_range_info on SSA_NAMEs. Formatting fixes.
25040 2017-03-10 Richard Biener <rguenther@suse.de>
25041 Jakub Jelinek <jakub@redhat.com>
25043 PR tree-optimization/77975
25044 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
25045 edge to be constant.
25046 (get_val_for): For constant x return it. Formatting fix.
25047 (loop_niter_by_eval): Avoid pointless looping if the next iteration
25048 would use the same bases as the current one.
25050 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25052 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
25053 instead of vec_select for V1TImode.
25054 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
25056 (VSX_LE_128): Add V1TI to this mode iterator.
25057 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
25058 (*vsx_le_perm_store_<mode>): Likewise.
25059 (pre-reload splitter for VSX stores): Likewise.
25060 (post-reload splitter for VSX stores): Likewise.
25061 (*vsx_xxpermdi2_le_<mode>): Likewise.
25062 (*vsx_lxvd2x2_le_<mode>): Likewise.
25063 (*vsx_stxvd2x2_le_<mode>): Likewise.
25065 2017-03-09 Michael Eager <eager@eagercon.com>
25067 Correct failures with --enable-checking=yes,rtl.
25069 * config/microblaze/microblaze.c (microblaze_expand_shift):
25070 Replace GET_CODE test with CONST_INT_P and INTVAL test with
25071 test for const0_rtx.
25072 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
25073 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
25075 2017-03-09 Richard Biener <rguenther@suse.de>
25077 PR tree-optimization/79977
25078 * graphite-scop-detection.c (scop_detection::merge_sese):
25079 Handle the case of extra exits to blocks dominating the entry.
25081 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
25083 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25086 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
25088 PR rtl-optimization/79949
25089 * lra-constraints.c (process_alt_operands): Check memory when
25090 trying to predict a cycle. Print about the overall increase.
25092 2017-03-09 Richard Biener <rguenther@suse.de>
25094 PR middle-end/79971
25095 * gimple-expr.c (useless_type_conversion_p): Preserve
25096 TYPE_SATURATING for fixed-point types.
25098 2017-03-09 Richard Biener <rguenther@suse.de>
25101 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
25102 alignment of BLKmode params.
25104 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25107 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
25108 (VALL_NO_V2Q): Likewise.
25110 * config/aarch64/aarch64-simd.md
25111 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
25113 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
25114 VALL_NO_V2Q mode iterator.
25115 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
25117 2017-03-09 Martin Liska <mliska@suse.cz>
25119 PR tree-optimization/79631
25120 * tree-chkp-opt.c (chkp_is_constant_addr): Call
25121 tree_int_cst_sign_bit just for INTEGER constants.
25123 2017-03-09 Martin Liska <mliska@suse.cz>
25127 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
25130 2017-03-09 Marek Polacek <polacek@redhat.com>
25133 * tree.c (inchash::add_expr): Handle TREE_VEC.
25135 2017-03-09 Martin Liska <mliska@suse.cz>
25138 (chkp_narrow_size_and_offset): New function.
25139 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
25140 (void chkp_parse_bit_field_ref): New function.
25141 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
25142 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
25144 2017-03-09 Martin Liska <mliska@suse.cz>
25147 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
25148 (chkp_find_bounds_1): Remove gcc_unreachable.
25150 2017-03-09 Jakub Jelinek <jakub@redhat.com>
25153 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
25154 BUILT_IN_SYNC*, determine the access type from the size suffix and
25155 always build a MEM_REF with that type. Handle forgotten
25156 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
25159 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
25160 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
25161 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
25162 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
25163 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
25164 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
25165 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
25166 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
25167 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
25168 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
25169 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
25170 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
25171 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
25172 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
25173 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
25174 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
25175 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
25176 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
25177 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
25178 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
25179 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
25180 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
25181 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
25182 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
25183 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
25184 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
25185 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
25186 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
25187 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
25188 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
25189 definitions outside of __OPTIMIZE__ guarded section.
25192 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
25193 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
25194 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
25195 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
25198 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25200 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
25201 ("vfenez<mode>"): Add missing constraints.
25203 2017-03-08 Martin Sebor <msebor@redhat.com>
25206 * config/nds32/nds32.c (nds32_option_override):
25207 Fix misspelled diagnostic.
25209 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25212 * gimplify.c (gimplify_omp_for): Replace index var in outer
25213 taskloop statement with an artificial variable and add
25214 OMP_CLAUSE_PRIVATE clause for it.
25216 2017-03-08 Richard Biener <rguenther@suse.de>
25218 PR tree-optimization/79955
25219 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
25220 for accesses that are completely outside of the variable.
25222 2017-03-08 Andrew Haley <aph@redhat.com>
25224 PR tree-optimization/79943
25225 * tree-ssa-loop-split.c (compute_new_first_bound): When
25226 calculating the new upper bound, (END-BEG) should be added, not
25229 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25231 * config/avr/avr.md (setmemhi): Make sure match_dup
25232 operand number comes before match_scratch.
25234 2017-03-08 Richard Biener <rguenther@suse.de>
25236 PR tree-optimization/79920
25237 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
25238 with ncopies == 1 to ...
25239 (vect_transform_slp_perm_load): ... here. Properly compute
25240 all element loads by iterating VF times over the group. Do
25241 not handle ncopies (computed in a broken way) in
25242 vect_create_mask_and_perm.
25244 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25247 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
25248 is a uniform vector, use uniform_vector_p return value instead of
25249 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
25251 2017-03-07 Marek Polacek <polacek@redhat.com>
25253 PR middle-end/79809
25254 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
25255 (alloca_call_type): Likewise.
25257 2017-03-07 Martin Liska <mliska@suse.cz>
25259 * gcov.c (process_args): Put comment to correct location.
25261 2017-03-07 Martin Liska <mliska@suse.cz>
25263 PR middle-end/68270
25264 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
25265 Use array_at_struct_end_p instead of DECL_CHAIN (field).
25266 (chkp_narrow_bounds_for_field): Likewise.
25267 (chkp_parse_array_and_component_ref): Pass one more argument to
25270 2017-03-07 Richard Biener <rguenther@suse.de>
25272 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
25275 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25277 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
25278 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
25280 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25283 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
25284 to end of description.
25285 (PARAM_MAX_STORES_TO_MERGE): Likewise.
25287 2017-03-07 Jakub Jelinek <jakub@redhat.com>
25289 PR rtl-optimization/79901
25290 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
25292 (*avx512f_<code><mode>3<mask_name>): ... this.
25293 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
25294 iterator instead of VI8_AVX2_AVX512BW.
25296 PR rtl-optimization/79901
25297 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
25298 min/max expander, expand it using expand_vec_cond_expr.
25301 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
25304 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25307 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
25308 to void * for PCH reasons.
25309 * dwarf2out.c (output_loc_operands, output_die): Cast
25310 v.val_vec.array to unsigned char *.
25312 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
25315 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
25318 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
25320 PR rtl-optimization/79571
25321 * lra-constraints.c (process_alt_operands): Calculate static
25322 reject and subtract it from overall when only addresses will be
25325 2017-03-06 Julia Koval <julia.koval@intel.com>
25328 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
25329 incoming stack boundary to 128 for 64-bit targets.
25331 2017-03-06 Richard Biener <rguenther@suse.de>
25333 PR tree-optimization/79894
25334 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
25335 to NULL after folding it.
25337 2017-03-06 Richard Biener <rguenther@suse.de>
25339 PR tree-optimization/79824
25340 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
25341 check disabling peeling for gaps.
25343 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
25345 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25346 attributes): Document gettimeofday.
25348 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25350 * config/s390/s390.c (s390_option_override_internal): Set
25351 PARAM_MIN_VECT_LOOP_BOUND
25353 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25355 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
25356 * config/s390/s390.md: Likewise.
25358 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25361 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
25362 (<avx2_avx512>_perm<mode>): Rename to ...
25363 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
25365 (<avx512>_perm<mode>_mask): Rename to ...
25366 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
25368 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
25369 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
25370 instead of VI8F_256_512.
25371 (avx512f_perm<mode>): New define_expand.
25372 (avx512f_perm<mode>_mask): Likewise.
25373 (avx512f_perm<mode>_1<mask_name>): New define_insn.
25374 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
25376 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25378 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
25379 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
25381 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
25383 2017-03-06 Martin Liska <mliska@suse.cz>
25386 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
25387 when having a SSA NAME w/o VAR_DECL assigned to it.
25389 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25391 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
25392 msa_dpsub_<su>_d): Fix MODE for vec_select.
25394 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25396 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
25398 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
25400 2017-03-06 Richard Biener <rguenther@suse.de>
25402 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
25403 * plugin.c (register_plugin_info): Likewise.
25404 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
25406 2017-03-05 Jakub Jelinek <jakub@redhat.com>
25408 * config/i386/sse.md (sse_storehps, sse_storelps,
25409 avx_<castmode><avxsizesuffix>_<castmode>,
25410 avx512f_<castmode><avxsizesuffix>_<castmode>,
25411 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
25412 in condition that at least one operand is not a MEM.
25414 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25416 PR middle-end/79805
25417 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
25418 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
25420 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
25421 gimple_call_nothrow_p flag based on whether original builtin can throw.
25422 If it can, emit following stmts on the fallthrough edge.
25423 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
25424 don't create new bb if inserting just debug stmts on the edge, try to
25425 insert them on the fallthru bb or just reset debug stmts.
25427 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
25430 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
25431 restore recog_data (including the operand rtxes inside it) around
25432 the call to get_insn_template.
25434 2017-03-03 Martin Sebor <msebor@redhat.com>
25436 PR tree-optimization/79699
25437 * context.c (context::~context): Free MPFR caches to avoid
25438 a memory leak on program exit.
25440 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25442 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
25443 Use wide_int::ulow () instead of .elt (0).
25445 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25447 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
25448 (*pushxf): Limit oF constraint to 32bit targets and add oC
25449 constraint for 64bit targets.
25450 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
25451 (*pushdf): Change rmF constraint to rmC.
25453 2017-03-03 Martin Liska <mliska@suse.cz>
25455 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
25456 Remove unused variable.
25458 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25461 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
25462 is a memory operand, increase num_memory.
25463 (ix86_expand_args_builtin): Likewise.
25465 2017-03-03 Jan Hubicka <jh@suse.cz>
25468 * ipa-devirt.c (maybe_record_node): Properly handle
25469 __cxa_pure_virtual visibility.
25471 2017-03-03 Martin Liska <mliska@suse.cz>
25473 PR tree-optimization/79803
25474 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
25476 (pass_loop_prefetch::execute): Disabled optimization if an
25477 assumption about L1 cache size is not met.
25479 2017-03-03 Martin Liska <mliska@suse.cz>
25481 PR rtl-optimization/79574
25482 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
25483 (hash_scan_set): Likewise.
25484 (dump_hash_table): Likewise.
25485 (hoist_code): Likewise.
25487 2017-03-03 Richard Biener <rguenther@suse.de>
25489 * fixed-value.c (fixed_from_string): Restore use of elt (1)
25490 in place of uhigh ().
25491 (fixed_convert_from_real): Likewise.
25493 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25496 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
25498 2017-03-03 Richard Biener <rguenther@suse.de>
25500 PR middle-end/79818
25501 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
25502 TYPE_OVERFLOW_UNDEFINED check.
25504 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25506 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
25508 (vector_ae_<mode>_p): Likewise.
25509 (vector_nez_<mode>_p): Likewise.
25510 (vector_ne_v2di_p): Likewise.
25511 (vector_ae_v2di_p): Likewise.
25512 (vector_ne_<mode>_p): Likewise.
25513 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
25515 (vsx_tsqrt<mode>2_fe): Likewise.
25517 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
25520 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
25522 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25524 PR rtl-optimization/79780
25525 * cprop.c (one_cprop_pass): When second and further conditional trap
25526 in a single basic block is turned into an unconditional trap, turn it
25527 into a deleted note to avoid RTL verification failures.
25529 2017-03-02 Richard Biener <rguenther@suse.de>
25531 * fold-const.c (const_binop): Use ulow () instead of elt (0).
25533 2017-03-02 Richard Biener <rguenther@suse.de>
25535 PR tree-optimization/79345
25537 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
25538 param and abort the walk, returning -1 if it is hit.
25539 (walk_aliased_vdefs): Take a limit param and pass it on.
25540 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
25541 defaulting to 0 and return a signed int.
25542 * tree-ssa-uninit.c (struct check_defs_data): New struct.
25543 (check_defs): New helper.
25544 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
25545 about uninitialized memory.
25546 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
25547 bogus uninitialized warning.
25548 (fixed_convert_from_real): Likewise.
25550 2017-03-02 Bin Cheng <bin.cheng@arm.com>
25552 PR tree-optimization/66768
25553 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
25554 iv_use if base object can't be determined.
25556 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25558 PR tree-optimization/79345
25559 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
25560 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
25561 (get_pattern_stats): Initialize it.
25562 * genemit.c (gen_expand): Verify match_scratch numbers come after
25563 match_operand/match_dup numbers.
25564 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
25565 match_scratch numbers.
25566 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
25568 * config/s390/s390.md (trunctdsd2): Likewise.
25570 2017-03-02 Richard Biener <rguenther@suse.de>
25572 * wide-int.h (wide_int_storage::operator=): Implement in terms
25575 2017-03-02 Richard Biener <rguenther@suse.de>
25577 PR tree-optimization/79777
25578 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
25579 the to insert expression to sth existing.
25581 2017-03-01 Martin Sebor <msebor@redhat.com>
25583 PR middle-end/79692
25584 * gimple-ssa-sprintf.c
25585 (directive::known_width_and_precision): New function.
25586 (format_integer): Use it.
25587 (get_mpfr_format_length): Consider the full range of precision
25588 when computing %g output with the # flag. Set the likely byte
25589 count to 3 rather than 1 when precision is indeterminate.
25590 (format_floating): Correct the lower bound of precision.
25592 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25594 * doc/invoke.texi: Document default code model for 64-bit Linux.
25596 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25599 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
25600 udiv rather than div since input pattern is unsigned.
25602 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
25604 * config/i386/i386.c (print_reg): Warn for values of
25605 unsupported size in integer register.
25607 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
25610 * config/rs6000/predicates.md (current_file_function_operand): Do
25611 not allow self calls to be local if the function is replaceable.
25613 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
25616 * config/rs6000/altivec.h (vec_ctz and others): Change the
25617 preprocessor macro that controls conditional compilation from
25618 _ARCH_PWR9 to __POWER9_VECTOR__.
25619 (vec_all_ne): Change parameterization of __altivec_scalar_pred
25620 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
25621 control (instead of _ARCH_PWR9 control) so that template
25622 definition uses power9-specific function.
25623 (vec_any_eq): Likewise.
25624 (vec_all_ne): Change macro definition to use a power9-specific
25625 expansion under #ifdef __POWER9_VECTOR__ control (instead of
25626 _ARCH_PWR9 control).
25627 (vec_any_eq) Likewise.
25628 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
25629 expansion for CMPNEF to remove support for xvcmpnesp instruction.
25630 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
25631 support for xvcmpnedp instruction.
25632 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
25633 macro expansion so that Power9 implementation of vec_all_ne does
25634 not use the AltiVec predicate framework.
25635 (VCMPNEH_P): Likewise.
25636 (VCMPNEW_P): Likewise.
25637 (VCMPNED_P): Likewise.
25638 (VCMPNEFP_P): Likewise.
25639 (VCMPNEDP_P): Likewise.
25640 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
25641 implementation of vec_any_eq to not use AltiVec predicate
25643 (VCMPAEH_P): Likewise.
25644 (VCMPAEW_P): Likewise.
25645 (VCMPAED_P): Likewise.
25646 (VCMPAEFP_P): Likewise.
25647 (VCMPAEDP_P): Likewise.
25648 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
25649 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
25650 not use the AltiVec predicate framework.
25651 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
25652 of vec_any_eq to not use AltiVec predicate framework.
25653 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25654 support for predefined __POWER9_VECTOR__ macro to indicate that
25655 Power9 instruction selection is enabled.
25656 (altivec_overloaded_builtins): Remove extraneous
25657 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
25658 function argument types RS6000_BTI_bool_V16QI and
25659 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
25660 entry for overloaded function argument types RS6000_BTI_bool_V4SI
25661 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
25662 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
25663 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
25664 Power9 for implementations of vec_cmpne. Change the signature for
25665 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
25666 (representing vec_all_ne) to remove the previously described first
25667 argument of type RS6000_BTI_INTSI, as this was an artifact of
25668 reliance on the AltiVec predicate framework, which is no longer
25669 used in the implementation of these functions. Add
25670 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
25671 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
25672 since, unlike the AltiVec predicate framework implementation, we
25673 do not share function descriptors between vec_alle and vec_anyeq.
25674 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
25675 set of modes that receive special treatment even when
25676 TARGET_P9_VECTOR is true. The special treatment emits code that
25677 does not depend on Power9 instructions.
25678 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
25679 define_expand to not rely on AltiVec predicate framework.
25680 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25682 (vector_ne_v2di_p): Change this define_expand to not rely on
25683 AltiVec predicate framework.
25684 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
25686 (vector_ne_<mode>_p): Change this define_expand to not rely on
25687 AltiVec predicate framework.
25688 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25690 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
25691 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
25692 define_insn pattern.
25693 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
25694 define_insn pattern because the xvcmpne<VSs>. instruction is not
25696 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
25697 instruction is not supported.
25699 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25701 * config/nvptx/nvptx.c: Include intl.h.
25703 2017-03-01 Martin Jambor <mjambor@suse.cz>
25706 * ipa-prop.h (ipa_bits): Removed field known.
25707 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
25708 to pointers. Adjusted their comments to warn about their sharing.
25709 (ipcp_transformation_summary): Change bits to a vector of pointers.
25710 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
25711 (ipa_get_ipa_bits_for_value): Declare.
25712 * tree-vrp.h (value_range): Mark as GTY((for_user)).
25713 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
25714 (ipa_bits_hash_table): Likewise.
25715 (ipa_vr_ggc_hash_traits): Likewise.
25716 (ipa_vr_hash_table): Likewise.
25717 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
25718 being pointers and vr_known being removed.
25719 (ipa_set_jf_unknown): Likewise.
25720 (ipa_get_ipa_bits_for_value): New function.
25721 (ipa_set_jfunc_bits): Likewise.
25722 (ipa_get_value_range): New overloaded functions.
25723 (ipa_set_jfunc_vr): Likewise.
25724 (ipa_compute_jump_functions_for_edge): Use the above functions to
25725 construct bits and vr parts of jump functions.
25726 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
25727 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25729 (ipcp_grow_transformations_if_necessary): Also allocate
25730 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25732 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
25733 them. Fix too long lines.
25734 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
25735 vr_known being removed.
25736 (ipa_read_jump_function): Use new setter functions to construct bits
25737 and vr parts of jump functions or set them to NULL.
25738 (write_ipcp_transformation_info): Adjust for bits being pointers.
25739 (read_ipcp_transformation_info): Likewise.
25740 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
25742 Include gt-ipa-prop.h.
25743 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
25745 (ipcp_store_bits_results): Likewise.
25746 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
25747 Do not write to existing jump functions but use a temporary instead.
25749 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25752 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
25753 attempt to use its first operand as BIT_FIELD_REF base.
25755 2017-03-01 Richard Biener <rguenther@suse.de>
25757 PR middle-end/79721
25758 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
25759 interpolating formula in wrapping arithmetic.
25760 (chrec_apply): Convert chrec_evaluate return value to wanted type.
25762 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25764 PR tree-optimization/79734
25765 * tree-vect-generic.c (expand_vector_condition): Optimize
25766 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
25767 Handle VEC_COND_EXPR where comparison has different inner width from
25768 type's inner width.
25770 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
25772 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
25773 markup, and similar issues. Remove @opindex entries for things
25774 that aren't options. Add missing -mmpy-option entries.
25776 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25778 PR tree-optimization/79737
25779 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
25780 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
25781 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
25782 instead of byte_size. Formatting fix.
25783 (shift_bytes_in_array_right): Formatting fix.
25785 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
25788 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
25789 condition on optimize for the leaf function test.
25791 2017-02-28 Martin Liska <mliska@suse.cz>
25794 * read-rtl-function.c (function_reader::handle_unknown_directive):
25795 Bail out when one uses -flto.
25797 2017-02-28 Martin Liska <mliska@suse.cz>
25799 * common.opt: Replace space with tabular for options of <number>
25801 * config/i386/i386.opt: Show <number> value for
25802 -mlarge-data-threshold.
25803 * opts.c (print_filtered_help): Do not display number in hexadecimal
25806 2017-02-28 Martin Liska <mliska@suse.cz>
25808 * common.opt: Fix --help=option -Q for options which are of
25811 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
25813 * config/i386/i386.c (print_reg): Error out for values
25814 of 8-bit size in invalid integer register.
25816 2017-02-28 Martin Sebor <msebor@redhat.com>
25818 PR tree-optimization/79691
25819 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
25821 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25824 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
25825 gcc_unreachable with output_operand_lossage.
25827 2017-02-28 Richard Biener <rguenther@suse.de>
25829 PR tree-optimization/79740
25830 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
25832 (visit_nary_op): Insert the nary into the hashtable if we
25833 pattern-matched sth.
25834 * tree-ssa-pre.c (eliminate_insert): Robustify.
25836 2017-02-28 Richard Biener <rguenther@suse.de>
25838 PR middle-end/79731
25839 * fold-const.c (decode_field_reference): Reject out-of-bound
25842 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25844 * config/i386/i386.c: Include intl.h.
25845 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
25846 instead of just cond ? "..." : "...".
25847 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
25848 * coverage.c (read_counts_file): Likewise.
25849 * omp-offload.c: Include intl.h.
25850 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
25851 of just cond ? "..." : "...".
25852 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
25853 of just cond ? "..." : "...".
25855 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
25858 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
25860 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
25861 'tune for' CPU name.
25862 * config/arm/arm-cpu-data.h: Regenerated.
25864 2017-02-28 Richard Biener <rguenther@suse.de>
25866 PR tree-optimization/79732
25867 * tree-inline.c (expand_call_inline): Do not shadow var.
25869 2017-02-28 Richard Biener <rguenther@suse.de>
25871 PR tree-optimization/79723
25872 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
25873 address-space properly.
25875 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
25877 * doc/optinfo.texi (Optimization groups): Fix option used for
25879 * doc/invoke.texi (-fopt-info): Document "omp".
25880 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
25881 (OPTGROUP_ALL): Add OPTGROUP_OMP.
25882 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
25883 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
25884 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
25886 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
25888 * dumpfile.c (optgroup_options): Instead of "openmp", associate
25889 OPTGROUP_OMP with "omp".
25891 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
25894 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
25895 for arithmetic shift of unsigned V2DI.
25897 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
25899 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
25900 arc/linux.h headers.
25901 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
25902 (LINK_SPEC): Likewise.
25903 (ARC_TLS_EXTRA_START_SPEC): Likewise.
25904 (EXTRA_SPECS): Likewise.
25905 (STARTFILE_SPEC): Likewise.
25906 (ENDFILE_SPEC): Likewise.
25907 (LIB_SPEC): Likewise.
25908 (TARGET_SDATA_DEFAULT): Likewise.
25909 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
25910 (MULTILIB_DEFAULTS): Likewise.
25911 (DWARF2_UNWIND_INFO): Likewise.
25912 * config/arc/big.h: New file.
25913 * config/arc/elf.h: Likewise.
25914 * config/arc/linux.h: Likewise.
25915 * config/arc/t-uClibc: Remove.
25917 2017-02-27 Bin Cheng <bin.cheng@arm.com>
25919 PR tree-optimization/77536
25920 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
25921 (tree_transform_and_unroll_loop): Use above function to compute the
25922 estimated niter of unrolled loop and use it when scaling profile.
25923 Also use count info rather than frequency if it's non-zero.
25924 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
25925 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
25926 (vect_transform_loop): Call above function.
25928 2017-02-27 Richard Biener <rguenther@suse.de>
25930 PR tree-optimization/45397
25931 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
25932 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
25933 (visit_nary_op): Add pattern matching for CSEing sign-changed
25934 or truncated operations with wider ones.
25936 2017-02-27 Richard Biener <rguenther@suse.de>
25938 PR tree-optimization/79690
25939 * tree-vect-stmts.c (vectorizable_store): Use vector type
25940 built from the DR with address-space.
25942 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
25944 * doc/invoke.texi (Optimize Options): Refine the description
25945 of asan-use-after-return.
25947 2017-02-25 Alan Modra <amodra@gmail.com>
25949 PR rtl-optimization/79584
25950 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
25951 base, not ad->base_term, the reg within base. Remove assertion
25952 that ad->base == ad->base_term. Replace gen_int_mode using
25953 bogus mode with const0_rtx.
25955 2017-02-25 Jakub Jelinek <jakub@redhat.com>
25957 PR middle-end/79396
25958 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
25959 FMA_EXPR like tcc_binary or tcc_unary.
25961 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
25964 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
25966 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
25967 (output_loc_operands): Handle DW_OP_call_ref and
25968 DW_OP_GNU_variable_value.
25969 (struct variable_value_struct): New type.
25970 (struct variable_value_hasher): Likewise.
25971 (variable_value_hash): New variable.
25972 (string_types): Remove.
25973 (copy_loc_descr): New function.
25974 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
25975 (prepend_loc_descr_to_each): New function.
25976 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
25977 instead of add_loc_descr_to_each if the first argument is single
25978 location list and the second has multiple.
25979 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
25980 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
25981 when looking for variable value which doesn't have other location info.
25982 (loc_list_from_tree): Formatting fix.
25983 (gen_array_type_die): Simplify DW_AT_string_length handling.
25984 (adjust_string_types): Remove.
25985 (gen_subprogram_die): Don't call adjust_string_types nor test/set
25986 string_types. Call resolve_variable_values.
25987 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
25988 (resolve_addr_in_expr): Likewise. Add A argument.
25989 (copy_deref_exprloc): Remove deref argument. Adjust for the
25990 original expression being DW_OP_GNU_variable_value with optionally
25991 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
25992 optionally after it.
25993 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
25994 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
25995 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
25996 (variable_value_hasher::hash, variable_value_hasher::equal): New
25998 (resolve_variable_value_in_expr, resolve_variable_value,
25999 resolve_variable_values, note_variable_value_in_expr,
26000 note_variable_value): New functions.
26001 (dwarf2out_early_finish): Call note_variable_value on all toplevel
26004 2017-02-24 Jakub Jelinek <jakub@redhat.com>
26007 * opts.h (handle_generated_option): Add GENERATED_P argument.
26008 * opts-common.c (handle_option): Adjust function comment.
26009 (handle_generated_option): Add GENERATED_P argument, pass it to
26011 (control_warning_option): Pass false to handle_generated_option
26013 * opts.c (maybe_default_option): Pass true to handle_generated_option
26015 * optc-gen.awk: Likewise.
26017 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26019 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
26020 a REG, look at the REG it is a SUBREG of.
26021 (splitter for cmpeqsi_t): Ditto.
26023 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26025 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
26026 the special USEs with the pattern of the insn, not the insn itself.
26028 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
26031 * doc/invoke.texi: Document -mload-store-pairs.
26033 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26034 Sandra Loosemore <sandra@codesourcery.com>
26036 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
26037 argument isn't a CONST_INT.
26038 (nios2_alternate_compare_const): Assert op is a CONST_INT.
26039 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
26040 (nios2_validate_compare): Bypass alternate compare logic if *op2
26041 is not a CONST_INT.
26042 (ldstwm_operation_p): Return false if first_base is not a REG or
26043 if first_offset is not a CONST_INT.
26045 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26047 * config/cris/cris.md: Use correct operand in a define_peephole2.
26049 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26051 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
26053 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26055 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
26056 this_insn if it is an INSN or JUMP_INSN.
26057 (force_offsettable): Look at base, not at addr.
26058 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
26059 on things that aren't necessarily CONST_INTs.
26061 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
26063 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
26064 -mfpmath=sse is the default also for x86-32 targets with SSE2
26065 instruction set when @option{-ffast-math} is enabled
26067 2017-02-24 Jeff Law <law@redhat.com>
26069 PR rtl-optimizatoin/79286
26070 * ira.c (update_equiv_regs): Drop may_trap_p exception to
26073 2017-02-24 Richard Biener <rguenther@suse.de>
26075 PR tree-optimization/79389
26076 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
26079 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
26081 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
26082 function comment to reflect reality.
26083 (loop_exits_before_overflow): Fix typo in function description.
26085 2017-02-24 Richard Biener <rguenther@suse.de>
26087 PR tree-optimization/79389
26088 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
26089 properly that a threading opportunity exists. Detect conditional
26090 copy/constant propagation opportunities.
26092 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
26094 * config/visium/visium.md (type): Add trap.
26095 (b): New mode attribute.
26096 (*btst): Rename into...
26097 (*btst<mode>): ...this and adjust.
26098 (*cbranchsi4_btst_insn): Rename into...
26099 (*cbranch<mode>4_btst_insn): ...this and adjust.
26100 (trap): New define_insn.
26102 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26104 PR tree-optimization/79389
26105 * ifcvt.c (struct noce_if_info): Add rev_cond field.
26106 (noce_reversed_cond_code): New function.
26107 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
26108 reversed_comparison_code. Formatting fix.
26109 (noce_try_store_flag): Test rev_cond != NULL in addition to
26110 reversed_comparison_code.
26111 (noce_try_store_flag_constants): Likewise.
26112 (noce_try_store_flag_mask): Likewise.
26113 (noce_try_addcc): Use rev_cond if non-NULL instead of
26114 reversed_comparison_code.
26115 (noce_try_cmove_arith): Likewise. Formatting fixes.
26116 (noce_try_minmax, noce_try_abs): Clear rev_cond.
26117 (noce_find_if_block): Initialize rev_cond.
26118 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
26119 instead of false as last argument never attempt to reverse it
26122 2017-02-23 Bin Cheng <bin.cheng@arm.com>
26124 PR tree-optimization/79663
26125 * tree-predcom.c (combine_chains): Process refs in reverse order
26126 only for ZERO length chains, and add explaining comment.
26128 2017-02-23 Jeff Law <law@redhat.com>
26130 PR tree-optimization/79578
26131 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
26132 in call to operand_equal_p.
26134 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
26137 * config/i386/cpuid.h: Fix another undefined behavior.
26139 2017-02-23 Richard Biener <rguenther@suse.de>
26141 PR tree-optimization/79683
26142 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
26143 vector types for data-refs.
26145 2017-02-23 Martin Liska <mliska@suse.cz>
26147 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
26149 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26151 PR middle-end/79665
26152 * internal-fn.c (get_range_pos_neg): Moved to ...
26153 * tree.c (get_range_pos_neg): ... here. No longer static.
26154 * tree.h (get_range_pos_neg): New prototype.
26155 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
26156 are known to be in between 0 and signed maximum inclusive, try to
26157 expand both unsigned and signed divmod and use the cheaper one from
26160 2017-02-22 Jeff Law <law@redhat.com>
26162 PR tree-optimization/79578
26163 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
26164 to compare base operands.
26166 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
26169 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
26170 gpc_reg_operand instead of fpr_reg_operand.
26172 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
26174 * config/mips/mips.c (mips_return_in_memory): Force FP
26175 vector types to be returned in memory for o32 ABI.
26177 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26179 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
26180 instead of DW_TAG_member for static data member declarations and don't
26181 set no_linkage_name for static inline data members.
26182 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
26185 2017-02-22 Martin Liska <mliska@suse.cz>
26187 * doc/invoke.texi: Replace inequality signs with square brackets
26190 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26192 PR tree-optimization/68644
26193 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
26195 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
26198 * lra-constraints.c (simplify_operand_subreg): Handle
26199 WORD_REGISTER_OPERATIONS targets.
26201 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26204 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
26205 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
26206 elimination by swapping fld*.
26208 2017-02-22 Richard Biener <rguenther@suse.de>
26210 PR tree-optimization/79673
26211 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
26212 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
26213 irrelevant address-space qualifiers and avoiding a
26214 ADDR_SPACE_CONVERT_EXPR from fold_convert.
26216 2017-02-22 Richard Biener <rguenther@suse.de>
26218 PR tree-optimization/79666
26219 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
26220 to not symbolically negate if that may introduce undefined
26223 2017-02-22 Martin Liska <mliska@suse.cz>
26226 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
26227 * data-streamer-out.c (streamer_write_gcov_count_stream):
26229 * value-prof.c (stream_out_histogram_value): Make assert more
26230 precise based on type of counter.
26232 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
26235 * config/i386/i386.md (standard_x87sse_constant_load splitter):
26236 Use nonimmediate_operand instead of memory_operand for operand 1.
26237 (float-extend standard_x87sse_constant_load splitter): Ditto.
26239 2017-02-21 Jeff Law <law@redhat.com>
26241 PR tree-optimization/79621
26242 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
26243 blocks with edges to themselves.
26245 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26248 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
26249 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
26250 Use gimple_call_builtin_p.
26253 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
26254 on temporarily removed DEBUG_INSNs.
26256 PR tree-optimization/79649
26257 * tree-loop-distribution.c (classify_partition): Give up on
26258 non-generic address space loads/stores.
26260 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
26262 * doc/loop.texi (Loop manipulation): Remove nonexistent
26263 tree_ssa_loop_version from the documentation.
26264 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
26266 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26269 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
26270 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
26271 * config/rs6000/rs6000.c: Include except.h.
26272 (rs6000_expand_split_stack_prologue): Call
26273 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
26275 2017-02-21 Martin Jambor <mjambor@suse.cz>
26278 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
26279 have been analyzed.
26281 2017-02-21 Martin Jambor <mjambor@suse.cz>
26283 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
26284 for backward compatibility only.
26285 * doc/invoke.texi (Option Summary): Remove all references to
26286 -fipa-cp-alignment.
26288 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
26292 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26294 * lra-constraints.c (curr_insn_transform): Handle
26295 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26297 2017-02-21 Martin Liska <mliska@suse.cz>
26299 * config/i386/i386.opt: Replace -masm-dialect with -masm.
26301 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
26303 PR translation/79638
26304 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
26306 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
26309 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
26310 (arm_function_ok_for_sibcall): Return false for an indirect call by
26311 descriptor if all the argument registers are used.
26312 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
26313 alignment of the function.
26315 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26317 PR tree-optimization/61441
26318 * simplify-rtx.c (simplify_const_unary_operation): For
26319 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
26320 the sNaN unmodified.
26322 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26324 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26325 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26326 instead of SYSTEM_HEADER_DIR.
26328 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
26329 Martin Liška <mliska@suse.cz>
26331 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
26332 Fix typos and grammar, use active voice, and clarify.
26334 2017-02-20 Marek Polacek <polacek@redhat.com>
26336 PR middle-end/79537
26337 * gimplify.c (gimplify_expr): Handle unused *&&L;.
26340 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
26342 2017-02-20 Jakub Jelinek <jakub@redhat.com>
26345 * config/i386/i386.c (ix86_expand_builtin): Handle
26346 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
26347 ix86_builtins_isa[fcode].isa as a requirement of those
26348 flags and any other flag in the bitmask.
26349 (ix86_init_mmx_sse_builtins): Use 0 instead of
26350 ~OPTION_MASK_ISA_64BIT as mask.
26351 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
26352 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
26353 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
26354 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
26356 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26359 * lra-constraints.c (split_reg): Check requested split mode
26360 is supported by the register.
26362 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26364 * lra-constraints.c (simplify_operand_subreg): Remove early
26367 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26370 * lra-constraints.c (curr_insn_transform): Tighten condition
26371 for converting SUBREG reloads from OP_OUT to OP_INOUT.
26373 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26376 * lra-constraints.c (curr_insn_transform): Handle
26377 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26379 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
26382 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
26384 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
26386 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
26389 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
26392 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26394 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
26397 2017-02-18 Jakub Jelinek <jakub@redhat.com>
26399 * final.c (last_columnnum, override_columnnum): New variables.
26400 (final_start_function): Set last_columnnum, pass it to begin_prologue
26401 hook and pass 0 to dwarf2out_begin_prologue.
26402 (final_scan_insn): Update override_columnnum. Pass last_columnnum
26403 to source_line debug hook.
26404 (notice_source_line): Compute last_columnnum and for debug_column_info
26405 return true on column changes.
26406 * debug.h (struct gcc_debug_hooks): Add column argument to
26407 source_line and begin_prologue hooks.
26408 (debug_nothing_int_charstar_int_bool): Remove prototype.
26409 (debug_nothing_int_int_charstar,
26410 debug_nothing_int_int_charstar_int_bool): New prototypes.
26411 (dwarf2out_begin_prologue): Add column argument.
26412 * debug.c (do_nothing_debug_hooks): Adjust source_line and
26413 begin_prologue hooks.
26414 (debug_nothing_int_charstar_int_bool): Remove.
26415 (debug_nothing_int_int_charstar,
26416 debug_nothing_int_int_charstar_int_bool): New functions.
26417 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
26418 through to dwarf2out_source_line.
26419 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
26420 (dwarf2out_source_line): Add column argument, emit it if requested.
26421 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
26423 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26424 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26425 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
26426 through to dwarf2out_begin_prologue.
26427 (vmsdbgout_source_line): Add column argument, pass it through to
26428 dwarf2out_source_line.
26429 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
26430 dbxout_source_line caller.
26431 (dbxout_source_line): Add column argument.
26433 * common.opt (gno-column-info, gcolumn-info): New options.
26434 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
26435 (check_die): Also test for multiple DW_AT_decl_column attributes.
26436 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
26437 DW_AT_decl_column if requested.
26438 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
26440 (gen_variable_die): Likewise.
26441 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
26442 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
26445 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
26446 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
26447 (ix86_handle_option): Handle OPT_m3dnowa.
26448 * doc/invoke.texi (-m3dnowa): Document.
26449 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
26450 -m3dnowa instead of -m3dnow -march=athlon.
26453 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
26454 instead of gcc_assert for K, r and R code checks. Formatting fixes.
26456 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26459 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
26460 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
26461 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
26462 generator for vsx_xxpermdi_<mode>_be.
26463 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
26464 force big-endian semantics.
26465 (vsx_xxpermdi_<mode>_be): New define_expand with same
26466 implementation as previous version of vsx_xxpermdi_<mode>.
26468 2017-02-17 Jakub Jelinek <jakub@redhat.com>
26470 PR tree-optimization/79327
26471 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
26472 variable, its initialization and use.
26474 2017-02-17 Julia Koval <julia.koval@intel.com>
26476 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
26477 (OPTION_MASK_ISA_PKU_UNSET): New.
26478 (ix86_handle_option): Handle -mrdpid.
26479 * config/i386/cpuid.h (bit_RDPID): New.
26480 * config/i386/driver-i386.c (host_detect_local_cpu):
26481 Detect RDPID feature.
26482 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
26483 * config/i386/i386-c.c (ix86_target_macros_internal):
26485 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
26486 (ix86_valid_target_attribute_inner_p): Add "rdpid".
26487 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26488 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
26489 * config/i386/i386.md (define_insn "rdpid"): New.
26490 * config/i386/i386.opt Add -mrdpid.
26491 * config/i386/immintrin.h (_rdpid_u32): New.
26493 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
26495 PR rtl-optimization/79541
26496 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
26497 instead of transforming it into USE.
26499 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
26501 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
26502 If HONOR_SNANS (SFmode) force the input to a register.
26503 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
26504 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
26505 an frsp or similar insn.
26507 2017-02-17 Martin Liska <mliska@suse.cz>
26509 PR rtl-optimization/79577
26510 * params.def (selsched-max-sched-times): Increase minimum to 1.
26512 2017-02-17 Martin Liska <mliska@suse.cz>
26514 PR rtl-optimization/79574
26515 * gcse.c (want_to_gcse_p): Prevent integer overflow.
26517 2017-02-17 Martin Liska <mliska@suse.cz>
26519 PR tree-optimization/79529
26520 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
26521 ssa_defined_default_def_p to handle cases which are implicitly
26523 * tree-ssa.c (ssa_defined_default_def_p): New function.
26524 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
26525 which are implicitly defined.
26526 * tree-ssa.h (ssa_defined_default_def_p): Declare.
26528 2017-02-17 Richard Biener <rguenther@suse.de>
26530 PR middle-end/79576
26531 * params.def (max-ssa-name-query-depth): Limit to 10.
26533 2017-02-17 Richard Biener <rguenther@suse.de>
26535 PR tree-optimization/79552
26536 * tree-ssa-structalias.c (visit_loadstore): Properly verify
26539 2017-02-17 Richard Biener <rguenther@suse.de>
26542 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
26544 2017-02-17 Marek Polacek <polacek@redhat.com>
26546 PR middle-end/79536
26547 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
26548 (fold_negate_expr): New wrapper.
26550 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
26552 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
26553 Correct terminology and de-emphasize pre-standard behavior.
26555 2017-02-16 Alan Modra <amodra@gmail.com>
26557 PR rtl-optimization/79286
26558 * ira.c (def_dominates_uses): New function.
26559 (update_equiv_regs): Don't create an equivalence for insns that
26560 may trap where the register def does not dominate the use.
26562 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
26564 PR rtl-optimization/78127
26565 * lra.c (lra): Call lra_eliminate before finish the loop after
26568 2017-02-16 Richard Biener <rguenther@suse.de>
26570 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
26572 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
26573 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
26574 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
26575 (isl_val_int_from_wi): New function.
26576 (extract_affine_gmp): Rename to ...
26577 (extract_affine_wi): ... this, take a widest_int.
26578 (extract_affine_int): Just wrap extract_affine_wi.
26579 (add_param_constraints): Use isl_val_int_from_wi.
26580 (add_loop_constraints): Likewise, and extract_affine_wi.
26582 2017-02-15 Jeff Law <law@redhat.com>
26584 PR middle-end/79521
26585 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
26586 ira_init_register_move_cost_if_necessary.
26588 2017-02-15 Martin Sebor <msebor@redhat.com>
26590 PR middle-end/32003
26591 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
26592 removed in a prior commit.
26594 2017-02-15 Bin Cheng <bin.cheng@arm.com>
26596 PR tree-optimization/79347
26597 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
26598 counters during peeling.
26600 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
26602 * Makefile.in (site.exp): Remove "set ISLVER".
26604 2017-02-15 Jakub Jelinek <jakub@redhat.com>
26607 * real.c (real_from_integer): Call real_convert even for decimal.
26609 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
26612 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
26614 2017-02-14 Andrew Pinski <apinski@cavium.com>
26616 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
26617 cores and change the partno/implementer to be correct.
26618 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
26619 the 'B" as the implementer.
26620 * config/aarch64/aarch64-tune.md: Regenerate.
26622 2017-02-14 Carl Love <cel@us.ibm.com>
26624 * config/rs6000/rs6000.c: Add case statement entry to make the
26625 xvcvuxdsp built-in argument unsigned.
26626 * config/rs6000/vsx.md: Fix the source and return operand types so they
26627 match the instruction definitions from the ISA document. Fix typo
26628 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
26631 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
26634 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
26635 member early_clobber_alts.
26636 * lra-lives.c (reg_early_clobber_p): New.
26637 (process_bb_lives): Use it.
26638 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
26639 (debug_operand_data): Initialize early_clobber_alts.
26640 (setup_operand_alternative): Set up early_clobber_alts.
26641 (collect_non_operand_hard_regs): Ditto. Pass early clobber
26642 alternatives to new_insn_reg.
26643 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
26645 (lra_update_insn_regno_info): Pass the new arg.
26647 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26649 PR middle-end/79505
26650 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
26651 (new_oacc_loop_raw): Don't clear already cleared fields.
26654 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
26655 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
26656 _mm512_prefetch_i64gather_ps): New inline functions and macros.
26658 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
26661 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
26663 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
26666 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
26667 the extra instruction to the right place to store 128-bit constant
26670 2017-02-14 Martin Sebor <msebor@redhat.com>
26672 PR middle-end/79448
26673 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
26674 warning for strings of unknown length.
26676 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26678 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
26680 2017-02-14 Jeff Law <law@redhat.com>
26683 * ira-costs.c (scan_one_insn): Initialize register move costs
26684 for pseudos seen in USE/CLOBBER insns.
26686 PR tree-optimization/79095
26687 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
26688 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
26689 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
26690 if the operands are known to be not equal, then the resulting range
26692 (intersect_ranges): If the new range is ~[0,0] and the old range is
26693 wide, then prefer ~[0,0].
26694 * tree-vrp.c (overflow_comparison_p_1): New function.
26695 (overflow_comparison_p): New function.
26696 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
26697 if NAME is used in an overflow test.
26698 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
26699 overflow check that can be expressed as an equality test, then adjust
26700 ops to be that equality test.
26702 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26704 * config/s390/s390-builtin-types.def: Remove flags argument.
26705 * config/s390/s390.c (s390_init_builtins): Likewise.
26707 2017-02-14 Martin Liska <mliska@suse.cz>
26709 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
26710 vector. Fix trailing white spaces.
26712 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
26714 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
26717 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26719 PR rtl-optimization/68664
26720 * config/arm/arm.c (arm_sched_can_speculate_insn):
26721 New function. Declare prototype.
26722 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26724 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26726 PR rtl-optimization/68664
26727 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
26729 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26731 2017-02-14 Amit Pawar <amit.pawar@amd.com>
26733 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
26734 max skip bytes for function, loop and jump.
26736 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26738 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
26739 ABS_EXPR for gimple dump.
26741 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26744 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
26746 PR tree-optimization/79408
26747 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
26748 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
26749 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
26750 also if rhs1 is INTEGER_CST.
26752 2017-02-14 Richard Biener <rguenther@suse.de>
26754 PR middle-end/79432
26755 * tree-into-ssa.c (insert_phi_nodes): When the function can
26756 have abnormal edges rewrite SSA names with broken use-def
26757 dominance out of SSA and register them for PHI insertion.
26759 2017-02-13 Martin Sebor <msebor@redhat.com>
26761 PR middle-end/79496
26762 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
26763 clearing info.nowrite flag when snprintf size argument is a range.
26765 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26767 * cprop.c (cprop_jump): Add missing space in string literal.
26768 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
26769 (get_constraint_for_component_ref): Likewise.
26770 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
26771 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
26772 * lra-constraints.c (process_alt_operands): Likewise.
26773 * ipa-inline.c (inline_small_functions): Likewise.
26774 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
26775 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26776 * trans-mem.c (diagnose_tm_1_op): Likewise.
26777 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
26778 (grid_parallel_clauses_gridifiable): Likewise.
26780 * config/nvptx/mkoffload.c (process): Add space in between
26783 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
26784 "MOD4_SSE_REGS" and "ALL_REGS".
26786 * spellcheck.c (test_data): Add , in between "foo" and "food".
26788 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26791 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
26792 boundary crossing check and subsequent code generation agree.
26794 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26796 * config/aarch64/aarch64.c (has_memory_op): Delete.
26797 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
26800 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26802 PR rtl-optimization/79388
26803 PR rtl-optimization/79450
26804 * combine.c (distribute_notes): When removing TEM_INSN for which
26805 corresponding dest has last value recorded, invalidate that last
26808 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26810 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
26811 of explicit '@'. Add missing assembly comment marker on branch costs
26814 2017-02-13 Nathan Sidwell <nathan@acm.org>
26816 * gengtype-lex.l (<in_struct>): Add '/'.
26818 2017-02-13 Martin Liska <mliska@suse.cz>
26821 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
26823 2017-02-13 Richard Biener <rguenther@suse.de>
26825 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
26827 * configure: Re-generate.
26828 * config.in: Likewise.
26829 * graphite-dependences.c: Simplify as if
26830 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
26831 * graphite-isl-ast-to-gimple.c: Likewise.
26832 * graphite-optimize-isl.c: Likewise.
26833 * graphite-poly.c: Likewise.
26834 * graphite-sese-to-poly.c: Likewise.
26835 * graphite.h: Likewise.
26836 * toplev.c: Include isl/version.h and use isl_version () for
26837 printing the ISL version.
26838 * doc/install.texi: Update ISL requirement.
26840 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
26842 * doc/standards.texi (Standards): Update reference to
26845 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
26847 * doc/extend.texi (Named Address Spaces): sourceware.org now
26849 * doc/install.texi (Binaries): Ditto.
26852 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26854 * doc/cpp.texi: Replace "stringify"/"stringification" with C
26855 standard terminology "stringize"/"stringizing" throughout.
26856 * doc/cppinternals.texi: Likewise.
26858 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26860 * doc/extend.texi: Fix some spelling mistakes and typos.
26861 * doc/invoke.texi: Likewise.
26863 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26866 * params.def (inline-min-speedup) Change from 10 to 8.
26868 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26870 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
26873 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26876 * ipa-inline-analysis.c (get_minimal_bb): New function.
26877 (record_modified): Use it.
26878 (remap_edge_change_prob): Handle also ancestor functions.
26880 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
26882 * doc/contrib.texi (Contributors): Remove broken link into
26883 the Mauve CVS repository.
26885 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26887 PR middle-end/79454
26888 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
26889 result computation whenever lhs doesn't have vector mode, not
26890 just when it has BLKmode.
26892 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26894 * doc/makefile.texi (profiledbootstrap): Refer to the
26895 installation instructions only in textual form.
26897 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26900 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
26902 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26904 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
26905 (Specific): Update mingw-w64 reference.
26907 (Specific): Remove broken link to Renesas RX processor.
26909 2017-02-10 Richard Biener <rguenther@suse.de>
26911 * toplev.c (process_options): Do not mention obsolete graphite
26912 options when printing sorry message about missing graphite support.
26913 Mention -floop-nest-optimize.
26915 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
26917 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
26918 (vtst_p16): Likewise.
26919 (vtstq_p8): Likewise.
26920 (vtstq_p16): Likewise.
26922 (vtstq_p64): Likewise.
26923 * config/arm/arm_neon.h (vgetq_lane_p64): New.
26924 (vset_lane_p64): New.
26925 (vsetq_lane_p64): New.
26927 2017-02-10 Jakub Jelinek <jakub@redhat.com>
26929 PR tree-optimization/79411
26930 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
26931 stmt operands are SSA_NAMEs used in abnormal phis.
26932 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
26935 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26938 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
26941 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26943 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
26945 2017-02-09 Jakub Jelinek <jakub@redhat.com>
26947 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
26951 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
26952 not arbitrary TREE_CONSTANT.
26955 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
26956 "omp declare target link" attribute unless is_global_var.
26957 * omp-offload.c (find_link_var_op): Likewise.
26959 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
26960 Chung-Lin Tang <cltang@codesourcery.com>
26962 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
26964 (gimplify_adjust_omp_clauses): Don't delete TILE.
26965 (gimplify_omp_for): Deal with TILE.
26966 * internal-fn.c (expand_GOACC_TILE): New function.
26967 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
26969 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
26970 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
26972 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
26973 avoid DIV for outermost collapse var.
26974 (expand_oacc_for): Insert tile element loop as needed. Adjust.
26975 Remove out of date comments, fix whitespace.
26976 * omp-general.c (omp_extract_for_data): Deal with tiling.
26977 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
26978 adjust OLF_DIM_BASE value.
26979 (struct omp_for_data): Add tiling field.
26980 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
26981 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
26982 for auto loops. Remove default auto determining, moved to
26983 oacc_loop_fixed_partitions.
26984 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
26985 stmts, add e_mask field.
26986 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
26987 (oacc_thread_numbers): Use oacc_dim_call.
26988 (oacc_xform_tile): New.
26989 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
26990 (finish_oacc_loop): Adjust for ifns vector.
26991 (oacc_loop_discover_walk): Append loop abstraction sites to list,
26992 add case for GOACC_TILE fns.
26993 (oacc_loop_xform_loop): Delete.
26994 (oacc_loop_process): Iterate over call list directly, and add
26995 handling for GOACC_TILE fns.
26996 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
26998 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
26999 vector partitioning to outer loops. Assign 2 partitions to loops
27000 when available. Add TILE handling.
27001 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
27002 (execite_oacc_device_lower): Process GOACC_TILE fns,
27003 ignore unknown specs.
27004 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
27005 * tree.c (omp_clause_num_ops): Adjust TILE ops.
27006 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
27008 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
27010 * configure.ac (ACX_BUGURL): Update.
27011 * configure: Regenerate.
27013 2017-02-09 Richard Biener <rguenther@suse.de>
27015 PR tree-optimization/69823
27016 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27017 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
27019 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27021 * config/arc/arc-c.def: Add __NPS400__ definition.
27022 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
27023 (TARGET_NPS400): Define.
27025 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27027 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
27029 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
27030 pointer, arch_info.
27031 (arc_cpu_types): Fill the arch_info field with a pointer into the
27032 arc_arch_types table.
27033 (arc_selected_cpu): Declare.
27034 * config/arc/arc.c (arc_selected_cpu): Make global.
27035 (arc_selected_arch): Delete.
27036 (arc_base_cpu): Delete.
27037 (arc_override_options): Remove references to deleted variables,
27038 update access to arch information.
27039 (ARC_OPT): Update access to arch information.
27040 (ARC_OPTX): Likewise.
27041 * config/arc/arc.h (arc_base_cpu): Remove declaration.
27042 (TARGET_ARC600): Update access to arch information.
27043 (TARGET_ARC601): Likewise.
27044 (TARGET_ARC700): Likewise.
27045 (TARGET_EM): Likewise.
27046 (TARGET_HS): Likewise.
27047 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
27050 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
27053 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
27054 condition/operands for integer GE/LE/GEU/LEU operations.
27056 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
27058 PR translation/79397
27059 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
27062 2017-02-08 Martin Jambor <mjambor@suse.cz>
27065 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
27066 whether allocation happened.
27067 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
27068 nothing was allocated.
27070 2017-02-08 Jakub Jelinek <jakub@redhat.com>
27072 PR tree-optimization/79408
27073 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
27074 constant, but SSA_NAME with a known integer range, use the minimum
27075 of that range instead of op1 to determine if modulo can be replaced
27076 with its first operand.
27078 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27080 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
27082 2017-02-08 Richard Biener <rguenther@suse.de>
27084 PR tree-optimization/71824
27085 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27086 Check all loops contained in the merged region.
27088 2017-02-07 Andrew Pinski <apinski@cavium.com>
27090 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
27092 2017-02-07 Andrew Pinski <apinski@cavium.com>
27094 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
27095 (thunderxt88): Likewise.
27096 (thunderxt81): Disable LSE and change v8.1 to v8.
27097 (thunderxt83): Likewise.
27099 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27100 Richard Biener <rguenther@suse.de>
27102 PR middle-end/79399
27103 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
27104 type from int to size_t.
27105 * ira-costs.c (struct_costs_size): Change type from int to size_t.
27107 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27109 PR rtl-optimization/79386
27110 * cprop.c (bypass_conditional_jumps): Initialize
27111 bypass_last_basic_block already before splitting bbs after
27112 unconditional traps...
27113 (bypass_conditional_jumps): ... rather than here.
27116 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
27117 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
27118 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
27119 fix -masm=intel patterns.
27121 2017-02-07 Richard Biener <rguenther@suse.de>
27123 PR tree-optimization/79256
27124 PR middle-end/79278
27125 * builtins.c (get_object_alignment_2): Use min_align_of_type
27126 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
27127 and ADJUST_FIELD_ALIGN.
27129 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
27131 * doc/tm.texi: Regenerate.
27132 * stor-layout.c (layout_decl): Adjust.
27133 (update_alignment_for_field): Likewise.
27134 (place_field): Likewise.
27135 (min_align_of_type): Likewise.
27136 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
27137 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
27138 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
27139 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
27140 * config/frv/frv.c (frv_adjust_field_align): Likewise.
27141 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
27142 * config/i386/i386.c (x86_field_alignment): Likewise.
27143 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
27144 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
27145 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
27146 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
27147 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
27148 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
27152 2017-01-30 Richard Biener <rguenther@suse.de>
27154 PR tree-optimization/79256
27155 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27156 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27159 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
27161 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
27162 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
27163 builtins to SImode and emit a zero-extend, if necessary.
27165 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27167 * docs/invoke.texi (RISC-V Options): Alphabetize.
27169 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27171 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
27174 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27176 * config/riscv/riscv.c: New file.
27177 * common/config/riscv/riscv-common.c: Likewise.
27178 * config.gcc: Likewise.
27179 * config/riscv/constraints.md: Likewise.
27180 * config/riscv/elf.h: Likewise.
27181 * config/riscv/generic.md: Likewise.
27182 * config/riscv/linux.h: Likewise.
27183 * config/riscv/multilib-generator: Likewise.
27184 * config/riscv/peephole.md: Likewise.
27185 * config/riscv/pic.md: Likewise.
27186 * config/riscv/predicates.md: Likewise.
27187 * config/riscv/riscv-builtins.c: Likewise.
27188 * config/riscv/riscv-c.c: Likewise.
27189 * config/riscv/riscv-ftypes.def: Likewise.
27190 * config/riscv/riscv-modes.def: Likewise.
27191 * config/riscv/riscv-opts.h: Likewise.
27192 * config/riscv/riscv-protos.h: Likewise.
27193 * config/riscv/riscv.h: Likewise.
27194 * config/riscv/riscv.md: Likewise.
27195 * config/riscv/riscv.opt: Likewise.
27196 * config/riscv/sync.md: Likewise.
27197 * config/riscv/t-elf-multilib: Likewise.
27198 * config/riscv/t-linux: Likewise.
27199 * config/riscv/t-linux-multilib: Likewise.
27200 * config/riscv/t-riscv: Likewise.
27201 * configure.ac: Likewise.
27202 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
27203 Waterman as RISC-V maintainers.
27204 * doc/install.texi: Add RISC-V entries.
27205 * doc/invoke.texi: Add RISC-V options section.
27206 * doc/md.texi: Add RISC-V constraints section.
27207 * configure: Regenerated.
27209 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
27212 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
27213 false values to be constant vectors with all 0 or all 1 bits set.
27214 (vcondu<mode><mode>): Likewise.
27215 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
27217 (fpmask_comparison_operator): Update comment.
27218 (vecint_comparison_operator): New predicate.
27219 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
27220 vector conditionals when the true and false values are constant
27221 vectors with all 0 bits or all 1 bits set.
27223 2017-02-06 Martin Sebor <msebor@redhat.com>
27225 PR tree-optimization/79376
27226 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
27228 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
27230 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
27231 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
27232 to simplify split condition.
27234 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27236 * omp-expand.c (oxpand_omp_atomic_fetch_op,
27237 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
27240 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
27242 PR rtl-optimization/68664
27243 * target.def (can_speculate_insn): New hook.
27244 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
27245 * doc/tm.texi: Regenerate.
27246 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
27247 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
27248 (rs6000_sched_can_speculate_insn): New function.
27250 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27252 PR tree-optimization/79284
27253 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
27254 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
27255 vectorizable_mask_load_store, vectorizable_operation,
27256 vect_is_simple_cond, get_same_sized_vectype): Use it instead
27257 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
27258 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
27259 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
27260 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
27261 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27262 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
27263 is_gimple_assign (stmt). Replace another such test with
27264 is_gimple_assign (stmt).
27266 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27269 * config/avr/avr.c (rtl-iter.h): Include it.
27270 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
27271 (avr_legitimate_combined_insn): ...and implementation.
27273 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
27275 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
27276 * config/s390/s390.c (s390_const_operand_ok)
27277 (s390_canonicalize_comparison, s390_extract_part)
27278 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
27279 (s390_contiguous_bitmask_p, s390_rtx_costs)
27280 (legitimize_pic_address): Likewise.
27281 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
27282 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
27283 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
27284 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
27285 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
27287 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27289 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
27290 REGNO($0) == REGNO($1).
27292 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27294 * config/s390/linux.h(SIZE_TYPE): Add comment.
27296 2017-02-06 Julian Brown <julian@codesourcery.com>
27297 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27298 Virendra Pathak <virendra.pathak@broadcom.com>
27300 * config/aarch64/aarch64-cores.def: Change the scheduler
27302 * config/aarch64/aarch64.md: Include thunderx2t99.md.
27303 * config/aarch64/thunderx2t99.md: New file.
27305 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27307 * doc/standards.texi (Go Language): Update link to language
27310 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27312 * tree-eh.c (lower_resx): Sanitize profile.
27313 (cleanup_empty_eh_move_lp): Likewise.
27315 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27318 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
27320 * cfgloopmanip.h (loop_version): Update prototype.
27321 * modulo-sched.c (sms_schedule): Update call of loop_version.
27322 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
27323 * tree-parloops.c (gen_parallel_loop): Likewise.
27324 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
27325 * tree-ssa-loop-split.c (split_loop): Likewise.
27326 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
27327 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
27329 2017-02-05 Martin Liska <mliska@suse.cz>
27332 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
27334 (print_operand_address): Initialize a struct to zero.
27336 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27338 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
27339 garbage collector only in textual form.
27341 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27343 * doc/extend.texi (x86 specific memory model extensions for
27344 transactional memory): Simplify a phrase.
27346 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
27349 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
27350 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
27351 (atomic_storedi_1): Likewise.
27353 2017-02-04 Jakub Jelinek <jakub@redhat.com>
27355 PR tree-optimization/79338
27356 * tree-parloops.c (gather_scalar_reductions): Don't call
27357 vect_analyze_loop_form for loop->inner before destroying loop's
27360 2017-02-03 Martin Sebor <msebor@redhat.com>
27362 PR tree-optimization/79327
27363 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
27364 when precision has resulted in leading zeros.
27365 (format_integer): Adjust the likely counter to assume an unknown
27366 argument that may be zero is non-zero.
27368 2017-02-03 Jason Merrill <jason@redhat.com>
27371 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
27372 avoid copying non-taken branch.
27374 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27376 PR tree-optimization/79340
27377 * tree-vect-loop.c (vectorizable_reduction): Release
27378 vec_defs elements after safe_splicing them into other vectors.
27381 PR tree-optimization/79327
27382 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
27383 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
27385 (format_integer): Use wide_int_to_tree instead of build_int_cst
27386 + to_?hwi. If argmin is NULL, just set argmin and argmax to
27387 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
27388 of shortest and longest sequence.
27390 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
27392 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
27393 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
27395 2017-02-03 Walter Lee <walt@tilera.com>
27398 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
27399 after initial stackframe link reg save.
27400 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
27402 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27405 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
27406 wu for stxssp alternative.
27408 2017-02-03 Martin Sebor <msebor@redhat.com>
27410 PR tree-optimization/79352
27411 * gimple-fold.c (get_range_strlen): Add argument.
27412 (get_range_strlen): Change return type to bool.
27413 (get_maxval_strlen): Pass in a dummy argument.
27414 * gimple-fold.h (get_range_strlen): Change return type to bool.
27415 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
27416 * tree.h (array_at_struct_end_p): Add argument.
27417 * tree.c (array_at_struct_end_p): Handle it.
27419 2017-02-03 Martin Liska <mliska@suse.cz>
27422 * multiple_target.c (create_dispatcher_calls): Redirect edge
27423 from a caller of a dispatcher.
27424 (expand_target_clones): Make the clones local.
27425 (ipa_target_clone): Do both target clones and resolvers.
27426 (ipa_dispatcher_calls): Remove the pass.
27427 (pass_dispatcher_calls::gate): Likewise.
27428 (make_pass_dispatcher_calls): Likewise.
27429 * passes.def (pass_target_clone): Put as very first IPA early
27432 2017-02-03 Martin Liska <mliska@suse.cz>
27434 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
27435 in case of a function with ifunc attribute.
27437 2017-02-03 Martin Liska <mliska@suse.cz>
27439 * cgraph.c (cgraph_node::dump): Dump function version info.
27440 * symtab.c (symtab_node::dump_base): Add missing new line.
27442 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27444 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
27445 (ifcombine_ifandif): Use it.
27447 2017-02-03 Martin Liska <mliska@suse.cz>
27449 * doc/invoke.texi: Document default value for
27450 use-after-scope-direct-emission-threshold.
27452 2017-02-03 Martin Liska <mliska@suse.cz>
27454 PR tree-optimization/79339
27455 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
27456 (format_floating): Likewise.
27458 2017-02-03 Martin Liska <mliska@suse.cz>
27461 * ipa-prop.c (ipa_node_params_t::insert): Remove current
27463 (ipa_node_params_t::remove): Likewise.
27464 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
27465 initialization from removed ipa_node_params_t::insert.
27466 (ipa_node_params::~ipa_node_params): Move from removed
27467 ipa_node_params_t::release.
27468 * symbol-summary.h (symbol_summary::m_released): New member.
27469 Do not release a summary twice. Do not allow to call finalizer
27470 for types of a summary that live in GGC memory.
27472 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27474 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
27477 2017-02-02 Martin Sebor <msebor@redhat.com>
27479 PR middle-end/79275
27480 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
27481 (format_string): Tighten up the range of output for non-constant
27482 strings and correct the expected range for wide non-constant strings.
27484 2017-02-02 Martin Sebor <msebor@redhat.com>
27486 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
27488 PR middle-end/32003
27489 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
27491 (-fdump-tree-@var): Add to index and document how to come up
27492 with pass-specific option and dump file names.
27493 (-fdump-passes): Clarify where to look for output.
27495 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27497 PR middle-end/77445
27498 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
27499 statistics of the analyzed path; allow threading for speed when
27500 any of BBs along the path are optimized for speed.
27502 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
27504 PR middle-end/78468
27505 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
27506 settings of the virtual registers.
27509 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
27511 * explow.c (get_dynamic_stack_size): Take known alignment of stack
27512 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
27515 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27517 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
27518 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
27520 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27522 * config/s390/s390.md: Add missing comments with the expanded
27524 * config/s390/vector.md: Likewise.
27525 * config/s390/vx-builtins.md: Likewise.
27527 2017-02-02 Jakub Jelinek <jakub@redhat.com>
27530 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
27531 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
27532 conditions on a single line.
27534 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27536 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27537 __S390_VX__ to __VX__.
27539 2017-02-01 Andrew Pinski <apinski@cavium.com>
27541 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
27542 stmt_info to record_stmt_cost.
27543 (vect_get_known_peeling_cost): Pass stmt_info if known to
27545 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
27546 cpu_vector_cost field into
27547 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
27548 field into vec_int_stmt_cost and vec_fp_stmt_cost.
27549 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
27550 splitting of scalar_stmt_cost and vec_stmt_cost.
27551 (thunderx_vector_cost): Likewise.
27552 (cortexa57_vector_cost): LIkewise.
27553 (exynosm1_vector_cost): Likewise.
27554 (xgene1_vector_cost): Likewise.
27555 (thunderx2t99_vector_cost): Improve after the splitting of the two
27557 (aarch64_builtin_vectorization_cost): Update for the splitting of
27558 scalar_stmt_cost and vec_stmt_cost.
27560 2017-02-01 Torvald Riegel <triegel@redhat.com>
27561 Richard Henderson <rth@redhat.com>
27563 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
27564 conditional on existance of a fast atomic load.
27565 * optabs-query.c (can_atomic_load_p): New function.
27566 * optabs-query.h (can_atomic_load_p): Declare it.
27567 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
27568 no fast atomic load is available for the particular size of access.
27569 (expand_atomic_compare_and_swap): Likewise.
27570 (expand_atomic_load): Likewise.
27571 (expand_atomic_store): Likewise.
27572 (expand_atomic_fetch_op): Likewise.
27573 * testsuite/lib/target-supports.exp
27574 (check_effective_target_sync_int_128): Remove x86 because it provides
27575 no fast atomic load.
27576 (check_effective_target_sync_int_128_runtime): Likewise.
27578 2017-02-01 Richard Biener <rguenther@suse.de>
27580 * graphite.c: Include tree-vectorizer.h for find_loop_location.
27581 (graphite_transform_loops): Provide opt-info for optimized nests.
27582 * tree-parloop.c (parallelize_loops): Provide opt-info for
27583 parallelized loops.
27585 2017-02-01 Richard Biener <rguenther@suse.de>
27587 PR middle-end/79315
27588 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
27589 was not set before.
27591 2017-02-01 Richard Biener <rguenther@suse.de>
27593 PR tree-optimization/71824
27594 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27595 Verify the loops are valid in the merged SESE region.
27596 (scop_detection::can_represent_loop_1): Check analyzing the
27597 evolution of the number of iterations in the region succeeds.
27599 2017-01-31 Ian Lance Taylor <iant@golang.org>
27601 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
27602 REG_ARGS_SIZE note to 32-bit push insns and call insn.
27604 2017-01-31 David Malcolm <dmalcolm@redhat.com>
27606 PR preprocessor/79210
27607 * input.c (get_substring_ranges_for_loc): Replace line_width
27608 assertion with error-handling.
27610 2017-01-31 Richard Biener <rguenther@suse.de>
27612 PR tree-optimization/77318
27613 * graphite-sese-to-poly.c (extract_affine): Fix assert.
27614 (create_pw_aff_from_tree): Take loop parameter.
27615 (add_condition_to_pbb): Pass loop of the condition to
27616 create_pw_aff_from_tree.
27618 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27620 * config/s390/s390.c (s390_asan_shadow_offset): New function.
27621 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
27623 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
27627 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
27629 (convert_int_to_float128): Likewise.
27630 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
27631 (convert_int_to_float128): Likewise.
27632 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
27633 (UNSPEC_IEEE128_CONVERT): Likewise.
27634 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
27635 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
27636 Use local variables for IBM extended format.
27637 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
27638 (fix_trunc<mode>si2_fprs): Likewise.
27639 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
27640 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
27641 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
27642 to know that we can now have integers of all sizes in vector
27644 (fix<uns>_<mode>di2_hw): Likewise.
27645 (float<uns>_<mode>si2_hw): Likewise.
27646 (fix_<mode>si2_hw): Likewise.
27647 (fixuns_<mode>si2_hw): Likewise.
27648 (float<uns>_<mode>di2_hw): Likewise.
27649 (float_<mode>di2_hw): Likewise.
27650 (float_<mode>si2_hw): Likewise.
27651 (floatuns_<mode>di2_hw): Likewise.
27652 (floatuns_<mode>si2_hw): Likewise.
27653 (xscvqp<su>wz_<mode>): Delete, no longer used.
27654 (xscvqp<su>dz_<mode>): Likewise.
27655 (xscv<su>dqp_<mode>): Likewise.
27656 (ieee128_mfvsrd_64bit): Likewise.
27657 (ieee128_mfvsrd_32bit): Likewise.
27658 (ieee128_mfvsrwz): Likewise.
27659 (ieee128_mtvsrw): Likewise.
27660 (ieee128_mtvsrd_64bit): Likewise.
27661 (ieee128_mtvsrd_32bit): Likewise.
27663 2017-01-31 Martin Liska <mliska@suse.cz>
27666 * ipa-prop.c (ipa_free_all_node_params): Call release method
27667 instead of ~sumbol_summary to not to trigger double times
27670 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
27672 PR tree-optimization/71691
27673 * bitmap.h (class auto_bitmap): New.
27674 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
27675 is_maybe_undefined instead of ssa_undefined_value_p.
27677 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27679 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27680 __S390_ARCH_LEVEL__ to __ARCH__.
27682 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27684 PR tree-optimization/79267
27685 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
27686 if should_remove_lhs_p is true.
27688 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
27691 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
27692 (add_alignment_attribute): New.
27693 (base_type_die): Add alignment attribute.
27694 (subrange_type_die): Likewise.
27695 (modified_type_die): Likewise.
27696 (gen_array_type_die): Likewise.
27697 (gen_descr_array_type_die: Likewise.
27698 (gen_enumeration_type_die): Likewise.
27699 (gen_subprogram_die): Likewise.
27700 (gen_variable_die): Likewise.
27701 (gen_field_die): Likewise.
27702 (gen_ptr_to_mbr_type_die): Likewise.
27703 (gen_struct_or_union_type_die): Likewise.
27704 (gen_subroutine_type_die): Likewise.
27705 (gen_typedef_die): Likewise.
27706 (base_type_cmp): Compare alignment attribute.
27708 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27711 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
27712 (setb_unsigned) New pattern for setb with CCUNS.
27713 * config/rs6000/rs6000.c (expand_block_compare): Use a different
27714 subfc./subfe sequence to avoid overflow problems. Generate a
27715 shorter sequence with cmpld/setb for power9.
27716 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
27717 for generating subfc. instruction.
27718 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
27719 now uses this instruction.
27721 2017-01-30 Ian Lance Taylor <iant@google.com>
27724 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
27725 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
27727 2017-01-30 Martin Sebor <msebor@redhat.com>
27729 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
27730 Move constant to the right of a relational operator.
27731 (get_mpfr_format_length, format_character, format_string): Ditto.
27732 (should_warn_p, maybe_warn): Same.
27734 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
27736 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
27739 * asan.c (get_translation_unit_decl): Remove function.
27740 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
27742 2017-01-30 Martin Liska <mliska@suse.cz>
27744 PR gcov-profile/79259
27745 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
27746 -fprofile-generate.
27748 2017-01-30 Martin Liska <mliska@suse.cz>
27751 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
27752 Initialize variables with NULL value.
27754 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
27757 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
27759 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
27761 2017-01-30 Richard Biener <rguenther@suse.de>
27763 PR tree-optimization/79276
27764 * tree-vrp.c (process_assert_insertions): Properly adjust common
27765 when removing a duplicate.
27767 * gcc.dg/torture/pr79276.c: New testcase.
27769 2017-01-30 Richard Biener <rguenther@suse.de>
27771 PR tree-optimization/79256
27772 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27773 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27775 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
27777 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
27780 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
27781 ("*r<noxa>sbg_<mode>_sll_bitmask")
27782 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
27783 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
27784 Use contiguous_bitmask_nowrap_operand.
27786 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27789 * config/rs6000/altivec.h (vec_xl): Revise #define.
27790 (vec_xst): Likewise.
27792 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
27794 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
27796 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
27798 PR rtl-optimization/79194
27799 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
27800 traps before call to bypass_conditional_jumps.
27802 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
27804 PR tree-optimization/71374
27805 * lra-constraints.c (check_conflict_input_operands): New.
27806 (match_reload): Use it.
27808 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
27811 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
27812 account to calculate conflict_set.
27814 2017-01-27 Bin Cheng <bin.cheng@arm.com>
27816 PR rtl-optimization/78559
27817 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
27818 other_insn in combine.
27820 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27822 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
27823 uint16_type_node for BT_UINT16.
27825 2017-01-27 David Malcolm <dmalcolm@redhat.com>
27827 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
27828 "RTL Tests" to menu.
27829 (GIMPLE Tests): New node.
27830 (RTL Tests): New node.
27832 2017-01-27 Richard Biener <rguenther@suse.de>
27834 PR tree-optimization/79245
27835 * tree-loop-distribution.c (distribute_loop): Apply cost
27836 modeling also to detected patterns.
27838 2017-01-27 Richard Biener <rguenther@suse.de>
27840 PR tree-optimization/71433
27841 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
27842 (compare_assert_loc): New function.
27843 (process_assert_insertions): Sort and optimize assert locations
27844 to remove duplicates and push down identical assertions on
27845 edges to their destination block.
27847 2017-01-27 Richard Biener <rguenther@suse.de>
27849 PR tree-optimization/79244
27850 * tree-vrp.c (remove_range_assertions): Forcefully propagate
27851 out SSA names even if abnormal.
27853 2017-01-27 Jakub Jelinek <jakub@redhat.com>
27855 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
27856 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
27857 instead of MPFR_RNDN.
27859 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
27862 * arm.c (arm_option_override): Don't call build_target_option_node
27863 until after doing all option overrides.
27864 (arm_valid_target_attribute_tree): Likewise.
27866 2017-01-27 Martin Liska <mliska@suse.cz>
27868 * doc/invoke.texi (-fprofile-arcs): Document profiling support
27869 for {cd}tors and C++ {cd}tors.
27871 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
27873 * config/s390/s390.md ("*setmem_long_and")
27874 ("*setmem_long_and_31z"): Use zero_extend instead of and.
27876 2017-01-26 Martin Sebor <msebor@redhat.com>
27878 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
27881 2017-01-26 Martin Sebor <msebor@redhat.com>
27883 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
27884 HAVE_DFmode before using XFmode or DFmode.
27885 (parse_directive): Avoid using the z length modifier to avoid
27886 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
27888 PR middle-end/78703
27889 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
27890 to accept adjustment as an array.
27891 (get_int_range): New function.
27892 (struct directive): Make width and prec arrays.
27893 (directive::set_width, directive::set_precision): Call get_int_range.
27894 (format_integer, format_floating): Handle width and precision ranges.
27895 (format_string, parse_directive): Same.
27897 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27900 * dwarf2out.c (generate_skeleton_bottom_up): For children with
27901 comdat_type_p set, just clone them, but keep the children in the
27905 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
27906 which have direct callers with -fvar-tracking-assignments enabled
27908 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
27909 inside of type units.
27911 2017-01-26 Martin Sebor <msebor@redhat.com>
27913 PR middle-end/78703
27914 * gimple-ssa-sprintf.c (struct result_range): Add likely and
27916 (struct format_result): Replace number_chars, number_chars_min,
27917 and number_chars_max with a single member of struct result_range.
27919 (format_result::operator+=): Adjust.
27920 (struct fmtresult): Remove bounded. Handle likely and unlikely
27922 (fmtresult::adjust_for_width_or_precision): New function.
27923 (fmtresult:type_max_digits): New function.
27924 (bytes_remaining): Handle likely and unlikely counters.
27925 (min_bytes_remaining): Remove.
27926 (format_percent): Simplify.
27927 (format_integer, format_floating): Set likely and unlikely counters.
27928 (get_string_length, format_character, format_string): Same.
27929 (format_plain, should_warn_p): New function.
27930 (maybe_warn): Call should_warn_p. Update diagnostic messages
27931 and handle those for all directives, including plain strings.
27932 (format_directive): Handle likely and unlikely counters.
27933 Remove unnecessary quoting from diagnostics. Add an informational
27935 (add_bytes): Remove.
27936 (pass_sprintf_length::compute_format_length): Simplify.
27937 (try_substitute_return_value): Handle likely and unlikely counters.
27939 2017-01-26 Carl Love <cel@us.ibm.com>
27941 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
27942 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
27944 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
27947 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
27948 endianess for subregs into account.
27949 * lra-constraints.c (lra_constraints): Do risky transformations
27950 always on the first iteration.
27951 * lra-lives.c (check_pseudos_live_through_calls): Add arg
27952 last_call_used_reg_set.
27953 (process_bb_lives): Define and use last_call_used_reg_set.
27954 * lra.c (lra): Always continue after lra_constraints on the first
27957 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
27959 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
27961 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
27963 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27965 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
27966 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
27967 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
27968 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
27969 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
27970 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
27971 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
27972 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
27973 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
27975 2017-01-26 Marek Polacek <polacek@redhat.com>
27978 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
27979 for the third operand.
27981 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27983 PR middle-end/79236
27984 * omp-low.c (struct omp_context): Add simt_stmt field.
27985 (scan_omp_for): Return omp_context *.
27986 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
27987 context to the _simt_ SIMD stmt.
27988 (lower_omp_for): For combined SIMD with sibling _simt_
27989 SIMD, make sure to use the same decls in _looptemp_
27990 clauses as in the sibling.
27992 2017-01-26 David Sherwood <david.sherwood@arm.com>
27994 PR middle-end/79212
27995 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
27998 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28001 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
28002 emit fld b; fld a; if possible.
28004 * brig-builtins.def: Update copyright years.
28005 * config/arm/arm_acle_builtins.def: Update copyright years.
28007 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
28010 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
28011 constraint instead of o for the stxsd instruction.
28013 2017-01-25 Carl Love <cel@us.ibm.com>
28015 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
28016 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
28018 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
28020 * doc/invoke.texi (C++ Dialect Options): Fix typo.
28022 2017-01-25 Richard Biener <rguenther@suse.de>
28024 PR tree-optimization/69264
28025 * target.def (vector_alignment_reachable): Improve documentation.
28026 * doc/tm.texi: Regenerate.
28027 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
28029 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
28030 earlier changes with respect to TYPE_USER_ALIGN.
28031 (vector_alignment_reachable_p): Likewise. Improve dumping.
28033 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28036 * config/arm/arm.md (xordi3): Force constant operand into a register
28039 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28041 * doc/invoke.texi (-fstore-merging): Correct default optimization
28042 levels at which it is enabled.
28043 (-O): Move -fstore-merging from list to...
28046 2017-01-25 Richard Biener <rguenther@suse.de>
28049 * omp-expand.c: Include debug.h.
28050 (expand_omp_taskreg): Make sure to generate early debug before
28051 outlining anything from a function.
28052 (expand_omp_target): Likewise.
28053 (grid_expand_target_grid_body): Likewise.
28055 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
28058 * asan.c (get_translation_unit_decl): New function.
28059 (asan_add_global): Extract modules file name from globals
28060 TRANSLATION_UNIT_DECL name.
28062 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
28065 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
28066 for long calls with APCS frame and VFP.
28068 2017-01-24 David Malcolm <dmalcolm@redhat.com>
28070 * cfg.c (original_copy_tables_initialized_p): New function.
28071 * cfg.h (original_copy_tables_initialized_p): New decl.
28072 * cfgrtl.c (relink_block_chain): Guard the call to
28073 free_original_copy_tables with a call to
28074 original_copy_tables_initialized_p.
28075 * cgraph.h (symtab_node::native_rtl_p): New decl.
28076 * cgraphunit.c (symtab_node::native_rtl_p): New function.
28077 (symtab_node::needed_p): Don't assert for early assembly output
28078 for __RTL functions.
28079 (cgraph_node::finalize_function): Set "force_output" for __RTL
28081 (cgraph_node::analyze): Bail out early for __RTL functions.
28082 (analyze_functions): Update assertion to support __RTL functions.
28083 (cgraph_node::expand): Bail out early for __RTL functions.
28084 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
28086 * function.h (struct function): Update comment for field
28088 * gimple-expr.c: Include "tree-pass.h".
28089 (gimple_has_body_p): Return false for __RTL functions.
28090 * Makefile.in (OBJS): Add run-rtl-passes.o.
28091 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
28093 (gcc::pass_manager::get_clean_slate): New accessor.
28094 * passes.c: Include "insn-addr.h".
28095 (should_skip_pass_p): Add logging. Update logic for running
28096 "expand" to be compatible with both __GIMPLE and __RTL. Guard
28097 property-provider override so it is only done for gimple passes.
28099 (skip_pass): New function.
28100 (execute_one_pass): Call skip_pass when skipping passes.
28101 * read-md.c (md_reader::read_char): Support filtering
28102 the input to a subset of line numbers.
28103 (md_reader::md_reader): Initialize fields
28104 m_first_line and m_last_line.
28105 (md_reader::read_file_fragment): New function.
28106 * read-md.h (md_reader::read_file_fragment): New decl.
28107 (md_reader::m_first_line): New field.
28108 (md_reader::m_last_line): New field.
28109 * read-rtl-function.c (function_reader::create_function): Only
28110 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
28111 curr_properties. Set DECL_INITIAL to a dummy block.
28112 (read_rtl_function_body_from_file_range): New function.
28113 * read-rtl-function.h (read_rtl_function_body_from_file_range):
28115 * run-rtl-passes.c: New file.
28116 * run-rtl-passes.h: New file.
28118 2017-01-24 Jeff Law <law@redhat.com>
28120 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
28123 2017-01-24 Bin Cheng <bin.cheng@arm.com>
28125 PR tree-optimization/79159
28126 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
28127 (record_nonwrapping_iv): Improve boundary using above function if no
28128 value range information.
28130 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
28131 Martin Jambor <mjambor@suse.cz>
28133 * brig-builtins.def: New file.
28134 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
28135 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
28136 (DEF_HSAIL_SAT_BUILTIN): Likewise.
28137 (DEF_HSAIL_INTR_BUILTIN): Likewise.
28138 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
28139 * builtin-types.def (BT_INT8): New.
28140 (BT_INT16): Likewise.
28141 (BT_UINT8): Likewise.
28142 (BT_UINT16): Likewise.
28143 (BT_FN_ULONG): Likewise.
28144 (BT_FN_UINT_INT): Likewise.
28145 (BT_FN_UINT_ULONG): Likewise.
28146 (BT_FN_UINT_LONG): Likewise.
28147 (BT_FN_UINT_PTR): Likewise.
28148 (BT_FN_ULONG_PTR): Likewise.
28149 (BT_FN_INT8_FLOAT): Likewise.
28150 (BT_FN_INT16_FLOAT): Likewise.
28151 (BT_FN_UINT32_FLOAT): Likewise.
28152 (BT_FN_UINT16_FLOAT): Likewise.
28153 (BT_FN_UINT8_FLOAT): Likewise.
28154 (BT_FN_UINT64_FLOAT): Likewise.
28155 (BT_FN_UINT16_UINT32): Likewise.
28156 (BT_FN_UINT32_UINT16): Likewise.
28157 (BT_FN_UINT16_UINT16_UINT16): Likewise.
28158 (BT_FN_INT_PTR_INT): Likewise.
28159 (BT_FN_UINT_PTR_UINT): Likewise.
28160 (BT_FN_LONG_PTR_LONG): Likewise.
28161 (BT_FN_ULONG_PTR_ULONG): Likewise.
28162 (BT_FN_VOID_UINT64_UINT64): Likewise.
28163 (BT_FN_UINT8_UINT8_UINT8): Likewise.
28164 (BT_FN_INT8_INT8_INT8): Likewise.
28165 (BT_FN_INT16_INT16_INT16): Likewise.
28166 (BT_FN_INT_INT_INT): Likewise.
28167 (BT_FN_UINT_FLOAT_UINT): Likewise.
28168 (BT_FN_FLOAT_UINT_UINT): Likewise.
28169 (BT_FN_ULONG_UINT_UINT): Likewise.
28170 (BT_FN_ULONG_UINT_PTR): Likewise.
28171 (BT_FN_ULONG_ULONG_ULONG): Likewise.
28172 (BT_FN_UINT_UINT_UINT): Likewise.
28173 (BT_FN_VOID_UINT_PTR): Likewise.
28174 (BT_FN_UINT_UINT_PTR: Likewise.
28175 (BT_FN_UINT32_UINT64_PTR): Likewise.
28176 (BT_FN_INT_INT_UINT_UINT): Likewise.
28177 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
28178 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
28179 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
28180 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
28181 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
28182 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
28183 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
28184 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
28185 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
28186 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
28187 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
28188 * doc/frontends.texi: List BRIG FE.
28189 * doc/install.texi (Testing): Add BRIG tesring requirements.
28190 * doc/invoke.texi (Overall Options): Mention BRIG.
28191 * doc/standards.texi (Standards): Doucment BRIG HSA version.
28193 2017-01-24 Richard Biener <rguenther@suse.de>
28195 PR translation/79208
28196 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
28198 2017-01-24 Martin Jambor <mjambor@suse.cz>
28201 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
28202 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
28203 and known_contexts.
28205 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
28207 PR middle-end/79123
28208 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
28209 casts from signed to unsigned really don't have a range.
28211 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
28213 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
28214 GMP_RNDx for compatiblity.
28216 2017-01-24 Martin Liska <mliska@suse.cz>
28219 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
28220 that would prevent us to call alloca with -1 as argument.
28222 2017-01-24 Jakub Jelinek <jakub@redhat.com>
28224 * dwarf2out.c (output_compilation_unit_header, output_file_names):
28225 Avoid -Wformat-security warning.
28227 2017-01-23 Andrew Pinski <apinski@cavium.com>
28229 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
28232 2017-01-23 Martin Sebor <msebor@redhat.com>
28234 PR middle-end/78703
28235 * gimple-ssa-sprintf.c (warn_level): New global.
28236 (format_integer): Use it here and throughout the rest of the file.
28237 Use the same switch to compute sign as base.
28238 (maybe_warn): New function.
28239 (format_directive): Factor out warnings into maybe_warn.
28240 Add debugging output. Use warn_level.
28241 (add_bytes): Use warn_level.
28242 (pass_sprintf_length::compute_format_length): Add debugging output.
28243 (try_substitute_return_value): Same.
28244 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
28246 PR middle-end/78703
28247 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
28248 (struct fmtresult, format_integer, format_floating): Adjust.
28249 (fmtresult::fmtresult): Set max correctly in two argument ctor.
28250 (get_string_length, format_string,format_directive): Same.
28251 (pass_sprintf_length::compute_format_length): Same.
28252 (try_substitute_return_value): Simplify slightly.
28254 PR middle-end/78703
28255 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
28256 (fmtresult::operator+=): Outlined.
28257 (struct fmtresult): Add ctors.
28258 (struct conversion_spec): Rename...
28259 (struct directive): ...to this. Add and remove data members.
28260 (directive::set_width, directive::set_precision): New functions.
28261 (format_percent): Use fmtresult ctor.
28262 (get_width_and_precision): Remove.
28263 (format_integer): Make naming changes. Avoid computing width and
28265 (format_floating): Same. Adjust indentation.
28266 (format_character, format_none): New functions.
28267 (format_string): Moved character handling to format_character.
28268 (format_directive): Remove arguments, change return type.
28269 (parse_directive): New function.
28270 (pass_sprintf_length::compute_format_length): Move directive
28271 parsing to parse_directive.
28273 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28275 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
28276 (assign_assembler_name_if_needed): ... this.
28277 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
28278 (assign_assembler_name_if_needed): ... this.
28279 (free_lang_data_in_cgraph): Adjust callers.
28280 * cgraphunit.c (cgraph_node::analyze): Likewise.
28281 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
28284 2017-01-23 Richard Biener <rguenther@suse.de>
28286 PR tree-optimization/79088
28287 PR tree-optimization/79188
28288 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
28289 resetting loop bounds after last path deletion. Reset loop
28290 bounds of the target loop, make code match the comments.
28291 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
28292 Make sure loops need no fixups.
28294 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28296 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
28297 exponent support with double type for first argument.
28298 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
28299 type returned by __builtin_vec_extract_sig,
28300 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
28301 functions from "vector int" to "vector unsigned int" or from
28302 "vector long long int" to "vector unsigned long long int".
28303 Changed type returned by __builtin_vec_extract_exp,
28304 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
28305 functions from "vector int" to "vector unsigned int" or from
28306 "vector long long int" to "vector unsigned long long int".
28307 Changed return type of __builtin_vec_test_data_class,
28308 __builtin_vec_test_data_class_sp, and
28309 __builtin_vec_test_data_class_dp from "vector int" to
28310 "vector bool int" or from "vector long long int" to "vector bool
28311 long long int" and changed second argument type from "unsigned
28312 int" to "int". Added new overloaded function forms "vector float
28313 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
28314 "vector float __builtin_vec_insert_exp_sp (vector float, vector
28315 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
28316 double, vector unsigned long long int)" and "vector double
28317 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
28318 long int)". Changed return type of
28319 __builtin_scalar_test_data_class and
28320 __builtin_scalar_test_data_class_sp and
28321 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
28322 int" and changed second argument from "unsigned int" to "int".
28323 Changed type returned by __builtin_scalar_test_neg,
28324 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
28325 from "int" to "bool int". Added new overloaded function form
28326 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
28327 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
28328 exponent double-precision with floating point first argument.
28329 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
28330 documentation of scalar_test_data_class, scalar_test_neg,
28331 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
28332 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
28333 vec_test_data_class built-in functions to reflect refinements in
28334 their type signatures.
28336 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
28338 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
28340 (aarch64_elf_asm_destructor): Likewise.
28342 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
28344 PR rtl-optimization/78634
28345 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
28346 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
28347 * ifcvt.c (noce_try_cmove): Add missing cost check.
28349 PR rtl-optimization/71724
28350 * combine.c (if_then_else_cond): Look for situations where it is
28351 beneficial to undo the work of one of the recursive calls.
28353 2017-01-23 Bin Cheng <bin.cheng@arm.com>
28355 PR tree-optimization/70754
28356 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
28357 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
28358 combined stmt before it if not NULL.
28359 (combine_chains): Process refs reversely and compute dominance point
28362 2017-01-23 Martin Liska <mliska@suse.cz>
28364 PR tree-optimization/79196
28365 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
28366 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
28368 (strlen_optimize_stmt): Call the renamed function.
28370 2017-01-23 Michael Matz <matz@suse.de>
28372 PR tree-optimization/78384
28373 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
28375 2017-01-23 Richard Biener <rguenther@suse.de>
28377 PR tree-optimization/79186
28378 * tree-vrp.c (register_new_assert_for): Make sure we've seen
28379 both incoming edges before moving an assert.
28381 2017-01-23 Martin Jambor <mjambor@suse.cz>
28383 * ipa-prop.c (load_from_param_1): Removed.
28384 (load_from_unmodified_param): Bits from load_from_param_1 put back
28386 (load_from_param): Removed.
28387 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
28388 with stmt. Reverted back to use of load_from_unmodified_param.
28390 2017-01-23 Martin Jambor <mjambor@suse.cz>
28393 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
28394 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
28395 field a pointer to garbage collected vector, mark lattices and
28396 ipcp_orig_node with GTY((skip)).
28397 (ipa_get_param_count): Adjust to descriptors being a pointer.
28398 (ipa_get_param): Likewise.
28399 (ipa_get_type): Likewise.
28400 (ipa_get_param_move_cost): Likewise.
28401 (ipa_set_param_used): Likewise.
28402 (ipa_get_controlled_uses): Likewise.
28403 (ipa_set_controlled_uses): Likewise.
28404 (ipa_is_param_used): Likewise.
28405 (ipa_node_params_t): Move into garbage collector. New methods insert
28407 (ipa_node_params_sum): Annotate wth GTY(()).
28408 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
28410 (ipa_load_from_parm_agg): Adjust declaration.
28411 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
28412 * ipa-profile.c (ipa_profile): Likewise.
28413 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
28414 (ipa_populate_param_decls): Make descriptors parameter garbage
28416 (ipa_dump_param): Adjust to descriptors being a pointer.
28417 (ipa_alloc_node_params): Likewise.
28418 (ipa_initialize_node_params): Likewise.
28419 (load_from_param_1): Make descriptors parameter garbage collected.
28420 (load_from_unmodified_param): Likewise.
28421 (load_from_param): Likewise.
28422 (ipa_load_from_parm_agg): Likewise.
28423 (ipa_node_params::~ipa_node_params): Removed.
28424 (ipa_free_all_node_params): Remove call to delete operator.
28425 (ipa_node_params_t::insert): New.
28426 (ipa_node_params_t::remove): Likewise.
28427 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
28428 copy known_csts and known_contexts vectors.
28429 (ipa_read_node_info): Adjust to descriptors being a pointer.
28430 (ipcp_modif_dom_walker): Make m_descriptors field garbage
28432 (ipcp_transform_function): Make descriptors variable garbage
28435 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
28437 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
28438 * config/i386/avx512dqintrin.h: Ditto.
28439 * config/i386/avx512fintrin.h: Ditto.
28440 * config/i386/i386.c: Handle new builtins.
28441 * config/i386/i386-builtin.def: Add new builtins.
28442 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
28443 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
28445 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28446 Martin Liska <mliska@suse.cz>
28448 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
28449 * asan.c (asan_expand_poison_ifn): Support stores and use
28450 appropriate ASAN report function.
28451 * internal-fn.c (expand_ASAN_POISON_USE): New function.
28452 * internal-fn.def (ASAN_POISON_USE): Declare.
28453 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
28454 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
28455 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
28456 ASAN_POISON calls w/o LHS.
28457 * tree-ssa.c (execute_update_addresses_taken): Create clobber
28458 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
28459 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
28460 * gimplify.c (asan_poison_variables): Add attribute
28461 use_after_scope_memory to variables that really needs to live
28463 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
28464 having the attribute.
28466 2017-01-23 Martin Liska <mliska@suse.cz>
28468 * asan.c (create_asan_shadow_var): New function.
28469 (asan_expand_poison_ifn): Likewise.
28470 * asan.h (asan_expand_poison_ifn): New declaration.
28471 * internal-fn.c (expand_ASAN_POISON): Likewise.
28472 * internal-fn.def (ASAN_POISON): New builtin.
28473 * sanopt.c (pass_sanopt::execute): Expand
28474 asan_expand_poison_ifn.
28475 * tree-inline.c (copy_decl_for_dup_finish): Make function
28477 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
28478 * tree-ssa.c (is_asan_mark_p): New function.
28479 (execute_update_addresses_taken): Rewrite local variables
28480 (identified just by use-after-scope as addressable) into SSA.
28482 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28484 * doc/install.texi (Specific): opensource.apple.com uses https
28485 now. Remove trailing slash.
28487 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28489 * README.Portability: Remove note on an Irix compatibility issue.
28491 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
28493 * gcov.c (INCLUDE_ALGORITHM): Define.
28494 (INCLUDE_VECTOR): Define.
28495 No longer include <vector> and <algorithm> directly.
28497 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
28499 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
28501 * doc/invoke.texi (Code Gen Options): Ditto.
28503 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
28506 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
28508 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
28510 rtl-optimization/79125
28511 * cprop.c (local_cprop_pass): Handle cases where we make an
28512 unconditional trap.
28514 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
28518 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
28519 read from, for big endian.
28521 2017-01-20 Jiong Wang <jiong.wang@arm.com>
28523 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
28524 register pauth builtins for LP64 only.
28526 2017-01-20 Marek Polacek <polacek@redhat.com>
28529 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
28532 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28534 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
28536 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
28537 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
28538 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
28540 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28543 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
28544 in big-endian mode when they are not a single duplicated value.
28546 2017-01-20 Richard Biener <rguenther@suse.de>
28548 * BASE-VER: Bump to 7.0.1.
28550 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28552 * omp-low.c (omplow_simd_context): New struct. Use it...
28553 (lower_rec_simd_input_clauses): ...here and...
28554 (lower_rec_input_clauses): ...here to hold common data. Adjust all
28555 references to idx, lane, max_vf, is_simt.
28557 2017-01-20 Graham Markall <graham.markall@embecosm.com>
28559 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
28562 2017-01-20 Martin Jambor <mjambor@suse.cz>
28564 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
28565 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
28567 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
28568 (GTFILES): Rename hsa.c to hsa-common.c.
28569 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
28570 * hsa-dump.c: Likewise.
28571 * hsa-gen.c: Likewise.
28572 * hsa-regalloc.c: Likewise.
28573 * ipa-hsa.c: Likewise.
28574 * omp-expand.c: Likewise.
28575 * omp-low.c: Likewise.
28576 * toplev.c: Likewise.
28578 2017-01-20 Marek Polacek <polacek@redhat.com>
28581 * doc/invoke.texi: Document -Wduplicated-branches.
28582 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
28583 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
28584 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
28585 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
28586 return 0 only when not OEP_LEXICOGRAPHIC.
28587 (fold_build_cleanup_point_expr): Use the expression
28588 location when building CLEANUP_POINT_EXPR.
28589 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
28590 * tree.c (add_expr): Handle error_mark_node.
28592 2017-01-20 Martin Liska <mliska@suse.cz>
28595 * tree-profile.c (init_ic_make_global_vars): Do not call
28597 (gimple_init_gcov_profiler): Likewise.
28599 2017-01-20 Martin Liska <mliska@suse.cz>
28602 * cgraph.h (maybe_create_reference): Remove argument and
28604 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
28606 * ipa-cp.c (create_specialized_node): Likewise.
28607 * symtab.c (symtab_node::maybe_create_reference): Handle
28608 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
28610 2017-01-20 Martin Liska <mliska@suse.cz>
28612 * read-rtl-function.c (function_reader::create_function): Use
28613 build_decl instread of build_decl_stat.
28615 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
28617 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
28618 * config/i386/avx512dqintrin.h: Ditto.
28619 * config/i386/avx512fintrin.h: Ditto.
28620 * config/i386/i386-builtin-types.def: Add new types.
28621 * config/i386/i386.c: Handle new types.
28622 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
28623 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
28624 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
28625 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
28626 (__builtin_ia32_kshiftridi): New.
28627 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
28629 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
28633 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
28634 define to rs6000_init_stack_protect_guard.
28635 (rs6000_init_stack_protect_guard): New function.
28637 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28638 Yunqiang Su <yunqiang.su@imgtec.com>
28640 * config.gcc (supported_defaults): Add madd4.
28641 (with_madd4): Add validation.
28642 (all_defaults): Add madd4.
28643 * config/mips/mips.opt (mmadd4): New option.
28644 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28646 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28648 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
28649 (ISA_HAS_FUSED_MADD4): Likewise.
28650 * doc/invoke.texi (-mmadd4): Document the new option.
28651 * doc/install.texi (--with-madd4): Document the new option.
28653 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28655 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
28656 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
28657 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
28658 (aarch64_init_pauth_hint_builtins): New.
28659 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
28660 (aarch64_expand_builtin): Expand new builtins.
28662 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28664 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
28665 * combine-stack-adj.c (no_unhandled_cfa): Handle
28666 REG_CFA_TOGGLE_RA_MANGLE.
28667 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
28668 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
28669 info for return address signing.
28670 (aarch64_expand_epilogue): Likewise.
28672 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28674 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
28675 * config/aarch64/aarch64-protos.h
28676 (aarch64_return_address_signing_enabled): New declaration.
28677 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
28679 (aarch64_expand_prologue): Sign return address before it's pushed onto
28681 (aarch64_expand_epilogue): Authenticate return address fetched from
28683 (aarch64_override_options): Sanity check for ILP32 and ISA level.
28684 (aarch64_attributes): New function attributes for "sign-return-address".
28685 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
28686 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
28687 ("*do_return"): Generate combined instructions according to key index.
28688 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
28689 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
28691 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
28692 * config/aarch64/aarch64.opt (msign-return-address=): New.
28693 * doc/extend.texi (AArch64 Function Attributes): Documents
28694 "sign-return-address=".
28695 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
28697 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28699 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
28700 overall option summary.
28702 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28704 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
28705 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
28706 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
28707 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
28709 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
28711 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
28712 -mpower9-minmax by default for -mcpu=power9.
28713 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
28714 128-bit floating point.
28716 2017-01-20 Alan Modra <amodra@gmail.com>
28718 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
28719 optimizing for size.
28721 2017-01-20 Alan Modra <amodra@gmail.com>
28724 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
28725 for strcmp and strncmp from corresponding builtin decl.
28727 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28729 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
28730 instead of i386/rtems-64.h.
28731 * config/i386/rtems-64.h: Remove.
28733 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28737 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
28739 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
28741 2017-01-19 Tamar Christina <tamar.christina@arm.com>
28743 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
28744 Change int to HOST_WIDE_INT.
28745 * config/aarch64/aarch64-protos.h
28746 (aarch64_simd_gen_const_vector_dup): Likewise.
28747 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
28749 2017-01-19 David Malcolm <dmalcolm@redhat.com>
28751 * langhooks-def.h (lhd_type_for_size): New decl.
28752 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
28753 * langhooks.c (lhd_type_for_size): New function, taken from
28756 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
28758 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
28759 define_bypass for CR latency.
28760 (power9-cracked-alu): Update bypass latency and remove power9-branch.
28761 (power9-alu2): Add define_bypass for CR latency.
28763 (power9-mul): Update insn latency.
28764 (power9-mul-compare): Update insn latency, bypass latency and remove
28767 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28769 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
28771 * config/aarch64/aarch64.md
28772 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
28773 aarch64_nopcrelative_literal_loads.
28774 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
28776 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
28778 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
28779 TARGET_LOONGSON_3A.
28780 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
28782 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
28785 * config.gcc (supported_defaults): Add lxc1-sxc1.
28786 (with_lxc1_sxc1): Add validation.
28787 (all_defaults): Add lxc1-sxc1.
28788 * config/mips/mips.opt (mlxc1-sxc1): New option.
28789 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28791 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28792 __mips_no_lxc1_sxc1.
28793 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
28794 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
28795 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
28797 2017-01-19 Richard Biener <rguenther@suse.de>
28799 PR tree-optimization/72488
28800 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
28801 sure to restore SSA info.
28802 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
28804 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
28806 PR rtl-optimization/79121
28807 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
28808 of the inner type when shifting an extended value.
28810 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28813 * symtab.c (symtab_node::equal_address_to): Fix comparing of
28814 interposable aliases.
28816 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
28819 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
28820 Use the evmergelohi instruction.
28821 (mov_si<mode>_e500_subreg4_2_le): Likewise.
28822 (mov_sitf_e500_subreg8_2_be): Likewise.
28823 (mov_sitf_e500_subreg12_2_le): Likewise.
28824 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
28825 (mov_si<mode>_e500_subreg4_2_be): Likewise.
28826 (mov_sitf_e500_subreg8_2_le): Likewise.
28827 (mov_sitf_e500_subreg12_2_be): Likewise.
28829 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28831 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
28832 attribute from vecsimple to vecperm.
28833 (altivec_vbpermq2): Likewise.
28835 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28838 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
28840 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28841 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
28842 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
28843 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
28844 case where N arg is SIZE_MAX.
28845 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
28846 (cmpstrsi): Add pattern.
28848 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28850 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28851 __builtin_vec_revb builtins.
28852 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
28853 built-in functions to support generation of the ISA 3.0 XXBR<x>
28854 vector byte reverse instructions.
28855 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
28856 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
28857 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
28858 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
28859 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
28860 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
28861 (P9V_BUILTIN_VEC_REVB): Likewise.
28862 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
28863 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
28864 (p9_xxbrq_v16qi): Likewise.
28865 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
28866 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
28867 (p9_xxbrh_v8hi): Likewise.
28868 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
28869 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
28870 vec_revb built-in functions.
28872 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
28874 PR rtl-optimization/78952
28875 * config/i386/i386.md (any_extract): New code iterator.
28876 (*insvqi_2): Use any_extract for source operand.
28877 (*insvqi_3): Use any_shiftrt for source operand.
28879 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
28881 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
28883 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
28885 2017-01-18 Matthias Klose <doko@ubuntu.com>
28887 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
28889 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28891 * config/rs6000/altivec.h (vec_bperm): Change #define.
28892 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
28893 (altivec_vbpermq2): New define_insn.
28894 (altivec_vbpermd): Likewise.
28895 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
28896 function interface.
28897 (VBPERMD): Likewise.
28898 (VBPERM): New polymorphic function interface.
28899 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
28900 Add entries for P9V_BUILTIN_VEC_VBPERM.
28901 * doc/extend.texi: Add interfaces for vec_bperm.
28903 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28905 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
28906 first letter of error messages.
28907 (s390_resolve_overloaded_builtin): Likewise.
28908 * config/s390/s390.c (s390_expand_builtin): Likewise.
28909 (s390_invalid_arg_for_unprototyped_fn): Likewise.
28910 (s390_valid_target_attribute_inner_p): Likewise.
28911 * config/s390/s390.md ("tabort"): Likewise.
28913 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
28915 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
28916 (ISA_AVOID_DIV_HILO): New macro.
28917 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
28918 (ISA_HAS_DDIV): Likewise.
28920 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28922 * doc/invoke.texi (fabi-version): Correct number of occurrences.
28924 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28926 * doc/invoke.texi (fabi-version): Spelling fix.
28928 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28931 * doc/invoke.texi (fabi-version): Mention mangling fix for
28934 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28937 * doc/invoke.texi (fabi-version): Document discriminator mangling.
28939 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
28942 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
28943 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
28945 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
28947 (stack_protect_test): Ditto.
28948 * config/rs6000/rs6000.opt (mstack-protector-guard=,
28949 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
28951 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
28952 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
28953 -mstack-protector-guard-offset=.
28954 (RS/6000 and PowerPC Options): Ditto.
28956 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28958 * config/i386/i386.h (MASK_CLASS_P): New define.
28959 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
28960 there are no registers from different register sets also when
28961 mask registers are used. Update function comment.
28962 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
28963 to (*k/*r) and (*k/*km) alternatives.
28965 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
28967 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
28968 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
28969 (EH_RETURN_HANDLER_RTX): New define.
28970 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
28971 Force frame pointer in EH return functions.
28972 (aarch64_expand_epilogue): Add barrier for eh_return.
28973 (aarch64_final_eh_return_addr): Remove.
28974 (aarch64_eh_return_handler_rtx): New function.
28975 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
28977 (aarch64_eh_return_handler_rtx): New prototype.
28979 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28981 * config/rs6000/altivec.h (vec_rlmi): New #define.
28982 (vec_vrlnm): Likewise.
28983 (vec_rlnm): Likewise.
28984 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
28985 (UNSPEC_VRLNM): Likewise.
28986 (VIlong): New mode iterator.
28987 (altivec_vrl<VI_char>mi): New define_insn.
28988 (altivec_vrl<VI_char>nm): Likewise.
28989 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
28991 (VRLDNM): Likewise.
28992 (RLNM): New polymorphic function entry.
28993 (VRLWMI): New monomorphic function entry.
28994 (VRLDMI): Likewise.
28995 (RLMI): New polymorphic function entry.
28996 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
28997 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
28998 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
29001 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29004 * dwarf2out.c (field_byte_offset): Restore the
29005 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
29006 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
29007 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
29010 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29013 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
29015 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29018 * dwarf2out.c (add_data_member_location_attribute): For constant
29019 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
29020 instead of DW_AT_data_member_location, DW_AT_bit_offset and
29021 DW_AT_byte_size attributes.
29023 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29025 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
29026 after forcing to constant memory when the code model is medium.
29028 2017-01-17 Julia Koval <julia.koval@intel.com>
29031 * config/i386/avx512fintrin.h
29032 (_mm512_i32gather_ps): Change __addr type to void const*.
29033 (_mm512_mask_i32gather_ps): Ditto.
29034 (_mm512_i32gather_pd): Ditto.
29035 (_mm512_mask_i32gather_pd): Ditto.
29036 (_mm512_i64gather_ps): Ditto.
29037 (_mm512_mask_i64gather_ps): Ditto.
29038 (_mm512_i64gather_pd): Ditto.
29039 (_mm512_mask_i64gather_pd): Ditto.
29040 (_mm512_i32gather_epi32): Ditto.
29041 (_mm512_mask_i32gather_epi32): Ditto.
29042 (_mm512_i32gather_epi64): Ditto.
29043 (_mm512_mask_i32gather_epi64): Ditto.
29044 (_mm512_i64gather_epi32): Ditto.
29045 (_mm512_mask_i64gather_epi32): Ditto.
29046 (_mm512_i64gather_epi64): Ditto.
29047 (_mm512_mask_i64gather_epi64): Ditto.
29048 (_mm512_i32scatter_ps): Change __addr type to void*.
29049 (_mm512_mask_i32scatter_ps): Ditto.
29050 (_mm512_i32scatter_pd): Ditto.
29051 (_mm512_mask_i32scatter_pd): Ditto.
29052 (_mm512_i64scatter_ps): Ditto.
29053 (_mm512_mask_i64scatter_ps): Ditto.
29054 (_mm512_i64scatter_pd): Ditto.
29055 (_mm512_mask_i64scatter_pd): Ditto.
29056 (_mm512_i32scatter_epi32): Ditto.
29057 (_mm512_mask_i32scatter_epi32): Ditto.
29058 (_mm512_i32scatter_epi64): Ditto.
29059 (_mm512_mask_i32scatter_epi64): Ditto.
29060 (_mm512_i64scatter_epi32): Ditto.
29061 (_mm512_mask_i64scatter_epi32): Ditto.
29062 (_mm512_i64scatter_epi64): Ditto.
29063 (_mm512_mask_i64scatter_epi64): Ditto.
29064 * config/i386/avx512pfintrin.h
29065 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
29066 (_mm512_mask_prefetch_i32gather_ps): Ditto.
29067 (_mm512_mask_prefetch_i64gather_pd): Ditto.
29068 (_mm512_mask_prefetch_i64gather_ps): Ditto.
29069 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
29070 (_mm512_prefetch_i32scatter_ps): Ditto.
29071 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
29072 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
29073 (_mm512_prefetch_i64scatter_pd): Ditto.
29074 (_mm512_prefetch_i64scatter_ps): Ditto.
29075 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
29076 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
29077 * config/i386/avx512vlintrin.h
29078 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
29079 (_mm_mmask_i32gather_ps): Ditto.
29080 (_mm256_mmask_i32gather_pd): Ditto.
29081 (_mm_mmask_i32gather_pd): Ditto.
29082 (_mm256_mmask_i64gather_ps): Ditto.
29083 (_mm_mmask_i64gather_ps): Ditto.
29084 (_mm256_mmask_i64gather_pd): Ditto.
29085 (_mm_mmask_i64gather_pd): Ditto.
29086 (_mm256_mmask_i32gather_epi32): Ditto.
29087 (_mm_mmask_i32gather_epi32): Ditto.
29088 (_mm256_mmask_i32gather_epi64): Ditto.
29089 (_mm_mmask_i32gather_epi64): Ditto.
29090 (_mm256_mmask_i64gather_epi32): Ditto.
29091 (_mm_mmask_i64gather_epi32): Ditto.
29092 (_mm256_mmask_i64gather_epi64): Ditto.
29093 (_mm_mmask_i64gather_epi64): Ditto.
29094 (_mm256_i32scatter_ps): Change __addr type to void*.
29095 (_mm256_mask_i32scatter_ps): Ditto.
29096 (_mm_i32scatter_ps): Ditto.
29097 (_mm_mask_i32scatter_ps): Ditto.
29098 (_mm256_i32scatter_pd): Ditto.
29099 (_mm256_mask_i32scatter_pd): Ditto.
29100 (_mm_i32scatter_pd): Ditto.
29101 (_mm_mask_i32scatter_pd): Ditto.
29102 (_mm256_i64scatter_ps): Ditto.
29103 (_mm256_mask_i64scatter_ps): Ditto.
29104 (_mm_i64scatter_ps): Ditto.
29105 (_mm_mask_i64scatter_ps): Ditto.
29106 (_mm256_i64scatter_pd): Ditto.
29107 (_mm256_mask_i64scatter_pd): Ditto.
29108 (_mm_i64scatter_pd): Ditto.
29109 (_mm_mask_i64scatter_pd): Ditto.
29110 (_mm256_i32scatter_epi32): Ditto.
29111 (_mm256_mask_i32scatter_epi32): Ditto.
29112 (_mm_i32scatter_epi32): Ditto.
29113 (_mm_mask_i32scatter_epi32): Ditto.
29114 (_mm256_i32scatter_epi64): Ditto.
29115 (_mm256_mask_i32scatter_epi64): Ditto.
29116 (_mm_i32scatter_epi64): Ditto.
29117 (_mm_mask_i32scatter_epi64): Ditto.
29118 (_mm256_i64scatter_epi32): Ditto.
29119 (_mm256_mask_i64scatter_epi32): Ditto.
29120 (_mm_i64scatter_epi32): Ditto.
29121 (_mm_mask_i64scatter_epi32): Ditto.
29122 (_mm256_i64scatter_epi64): Ditto.
29123 (_mm256_mask_i64scatter_epi64): Ditto.
29124 (_mm_i64scatter_epi64): Ditto.
29125 (_mm_mask_i64scatter_epi64): Ditto.
29126 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
29127 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
29128 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
29129 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
29130 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
29131 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
29132 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
29133 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
29134 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
29135 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
29136 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
29137 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
29138 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
29139 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
29140 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
29141 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
29142 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
29143 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
29144 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
29145 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
29146 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
29147 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
29148 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
29149 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
29150 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
29151 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
29152 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
29153 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
29154 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
29155 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
29156 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
29157 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
29158 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
29159 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
29160 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
29161 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
29162 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
29163 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
29164 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
29165 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
29166 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
29167 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
29168 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
29169 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
29170 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
29171 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
29172 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
29173 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
29174 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
29175 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
29176 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
29177 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
29178 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
29179 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
29180 definitions accordingly.
29182 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
29183 Kuan-Lin Chen <kuanlinchentw@gmail.com>
29186 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
29189 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
29192 * ira-conflicts.c (ira_build_conflicts): Update total conflict
29193 hard regs for inner regno.
29195 2017-01-17 Martin Liska <mliska@suse.cz>
29198 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
29199 assumption and add comment.
29201 2017-01-17 Nathan Sidwell <nathan@acm.org>
29203 * ipa-visibility.c (localize_node): New function, broken out of ...
29204 (function_and_variable_visibility): ... here. Call it.
29206 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29208 PR middle-end/77445
29209 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
29210 correctly set frequency of oudgoing edge.
29211 (duplicate_thread_path): Fix profile updating.
29213 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29216 * configure.ac: Add GCC_BASE_VER.
29217 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
29218 version from BASE-VER file.
29219 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
29220 (gcc.o): Depend on $(BASEVER).
29221 * common.opt (dumpfullversion): New option.
29222 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
29223 * doc/invoke.texi: Document -dumpfullversion.
29224 * doc/install.texi: Document --with-gcc-major-version-only.
29225 * configure: Regenerated.
29227 2017-01-17 Richard Biener <rguenther@suse.de>
29229 PR tree-optimization/71433
29230 * tree-vrp.c (register_new_assert_for): Merge same asserts
29231 on all incoming edges.
29232 (process_assert_insertions_for): Handle insertions at the
29235 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
29237 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
29238 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
29240 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
29243 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
29246 2017-01-17 Alan Modra <amodra@gmail.com>
29249 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
29250 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
29251 symbolic stack limit when pic.
29253 2017-01-16 Martin Sebor <msebor@redhat.com>
29255 PR tree-optimization/78608
29256 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
29258 2017-01-16 Jeff Law <law@redhat.com>
29261 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
29262 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
29263 for several include directories that may be relative to sysroot.
29264 * config/i386/x-mingw32 (gplus_includedir): Define.
29265 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
29266 (native_system_includedir): Likewise.
29267 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
29268 override if TARGET_SYSTEM_ROOT is defined.
29269 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
29271 PR tree-optimization/79090
29272 PR tree-optimization/33562
29273 PR tree-optimization/61912
29274 PR tree-optimization/77485
29275 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
29276 and computed trims into the dump file.
29278 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29280 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
29282 2017-01-16 Jakub Jelinek <jakub@redhat.com>
29285 * gimplify.c (gimplify_init_constructor): If want_value and
29286 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
29290 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
29291 sequence. Formatting fixes.
29292 (doloop_optimize): Formatting fixes.
29295 * gcc.c (debug_level_greater_than_spec_func): New function.
29296 (static_spec_functions): Add debug-level-gt spec function.
29297 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
29299 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29300 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29301 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
29302 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
29303 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
29304 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
29306 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
29308 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
29309 QImode fixups to general and mask registers only.
29311 2017-01-16 Carl Love <cel@us.ibm.com>
29313 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
29314 for built-in functions
29315 vector signed char vec_nabs (vector signed char)
29316 vector signed short vec_nabs (vector signed short)
29317 vector signed int vec_nabs (vector signed int)
29318 vector signed long long vec_nabs (vector signed long long)
29319 vector float vec_nabs (vector float)
29320 vector double vec_nabs (vector double)
29321 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
29323 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
29324 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
29325 * doc/extend.texi: Update the documentation file for the new built-in
29328 2017-01-16 Martin Sebor <msebor@redhat.com>
29330 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
29333 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29335 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
29336 UNSPEC_VSX__XXSPLTD to require special splat handling.
29338 2017-01-16 David Malcolm <dmalcolm@redhat.com>
29341 * system.h: Poison strndup.
29343 2017-01-16 Alan Modra <amodra@gmail.com>
29346 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
29349 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
29351 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
29353 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
29355 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
29356 call recog here. Assert that INSN_CODE (insn) is non-negative.
29358 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29361 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
29363 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
29364 in the currently scheduled RTL fragment.
29366 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29368 PR rtl-optimization/78751
29369 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
29372 2017-01-14 Jeff Law <law@redhat.com>
29374 PR tree-optimization/79090
29375 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
29376 variable length stores.
29377 (compute_trims): Delete dead assignment to *trim_tail.
29378 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
29381 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
29383 PR rtl-optimization/78626
29384 PR rtl-optimization/78727
29385 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
29386 of a block, and split such blocks after everything else is finished.
29388 2017-01-14 Alan Modra <amodra@gmail.com>
29391 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
29392 target legitimate_combined_insn.
29393 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
29394 (rs6000_legitimate_combined_insn): New function.
29395 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
29397 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
29398 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
29399 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
29401 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
29403 * doc/frontends.texi (G++ and GCC): Remove references to Java.
29405 2017-01-13 Jeff Law <law@redhat.com>
29407 PR tree-optimization/33562
29408 PR tree-optimization/61912
29409 PR tree-optimization/77485
29410 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
29412 (delete_dead_assignment): Likewise.
29413 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
29414 statement to delete_dead_call and delete_dead_assignment.
29416 2017-01-13 David Malcolm <dmalcolm@redhat.com>
29419 * substring-locations.c (format_warning_va): Strengthen case 1 so
29420 that both endpoints of the substring must be within the format
29421 range for just the substring to be printed.
29423 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
29425 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
29426 * config/i386/i386.c (ix86_target_string): Add missing options
29427 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
29428 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
29429 flags_other and ix86_target_other to flags2_other. Display unknown
29431 (ix86_valid_target_attribute_inner_p): Add missing options and
29432 reorder options by implied ISAs, as in ix86_target_string.
29434 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29436 * hash-table.h (hash_table::too_empty_p): New function.
29437 (hash_table::expand): Use it.
29438 (hash_table::traverse): Likewise.
29439 (hash_table::empty_slot): Use sizeof (value_type) instead of
29440 sizeof (PTR) to convert bytes to elements. Shrink the table
29441 if the current size is excessive for the current number of
29444 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29446 * ira-costs.c (record_reg_classes): Break from the inner loop
29447 early once alt_fail is known to be true. Update outer loop
29448 handling accordingly.
29450 2017-01-13 Jeff Law <law@redhat.com>
29452 * tree-ssa-dse.c (decrement_count): New function.
29453 (increment_start_addr, maybe_trim_memstar_call): Likewise.
29454 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
29455 when we know the partially dead statement is a mem* function.
29457 PR tree-optimization/61912
29458 PR tree-optimization/77485
29459 * tree-ssa-dse.c: Include expr.h.
29460 (maybe_trim_constructor_store): New function.
29461 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
29463 PR tree-optimization/33562
29464 PR tree-optimization/61912
29465 PR tree-optimization/77485
29466 * doc/invoke.texi: Document new dse-max-object-size param.
29467 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
29468 * tree-ssa-dse.c: Include params.h.
29469 (dse_store_status): New enum.
29470 (initialize_ao_ref_for_dse): New, partially extracted from
29472 (valid_ao_ref_for_dse, normalize_ref): New.
29473 (setup_live_bytes_from_ref, compute_trims): Likewise.
29474 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
29475 (maybe_trim_partially_dead_store): Likewise.
29476 (maybe_trim_complex_store): Likewise.
29477 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
29478 Track what bytes live from the original store. Return tri-state
29479 for dead, partially dead or live.
29480 (dse_dom_walker): Add constructor, destructor and new private members.
29481 (delete_dead_call, delete_dead_assignment): New extracted from
29483 (dse_optimize_stmt): Make a member of dse_dom_walker.
29484 Use initialize_ao_ref_for_dse.
29486 PR tree-optimization/33562
29487 PR tree-optimization/61912
29488 PR tree-optimization/77485
29489 * sbitmap.h (bitmap_count_bits): Prototype.
29490 (bitmap_clear_range, bitmap_set_range): Likewise.
29491 * sbitmap.c (bitmap_clear_range): New function.
29492 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
29494 2017-01-13 Martin Liska <mliska@suse.cz>
29497 * function.c (set_cfun): Add new argument force.
29498 * function.h (set_cfun): Likewise.
29499 * ipa-inline-transform.c (inline_call): Use the function when
29500 strict alising from is dropped for function we inline to.
29502 2017-01-13 Richard Biener <rguenther@suse.de>
29504 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
29505 for dumping GIMPLE INTEGER_CSTs.
29507 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29509 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
29510 to 201112L since C++17.
29512 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
29515 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
29516 if -fsanitize=kernel-address is present.
29518 2017-01-13 Richard Biener <rguenther@suse.de>
29520 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
29521 as _Literal ( type ) number in case usual suffixes do not
29522 preserve all information.
29524 2017-01-13 Richard Biener <rguenther@suse.de>
29526 PR tree-optimization/77283
29527 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
29528 and ssa-iterators.h.
29529 (is_feasible_trace): Implement a cost model based on joiner
29532 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
29535 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
29536 char or short to __float128/_Float128 directly.
29538 2017-01-12 Martin Sebor <msebor@redhat.com>
29540 to -Wformat-overflow.
29541 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
29542 (min_bytes_remaining): Same.
29543 (get_string_length): Same.
29544 (format_string): Same.
29545 (format_directive): Same.
29547 (pass_sprintf_length::handle_gimple_call): Same.
29549 2017-01-12 Jakub Jelinek <jakub@redhat.com>
29551 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
29552 info.nowrite calls with no lhs that can't throw. Return bool
29553 whether gsi_remove has been called or not.
29554 (pass_sprintf_length::handle_gimple_call): Return bool whether
29555 try_substitute_return_value called gsi_remove. Formatting fix.
29556 (pass_sprintf_length::execute): Don't use gsi_remove if
29557 handle_gimple_call returned true.
29560 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
29561 be removed due to side-effects, don't remove following barrier nor
29562 turn the successor edge into fallthru edge.
29564 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29567 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
29568 element-reversing loads and stores as not swappable.
29570 2017-01-12 Nathan Sidwell <nathan@acm.org>
29571 Nicolai Stange <nicstange@gmail.com>
29573 * combine.c (try_combine): Don't ignore result of overlap checking
29574 loop. Combine overlap & asm check into single loop.
29576 2017-01-12 Richard Biener <rguenther@suse.de>
29578 * tree-pretty-print.c (dump_generic_node): Provide -gimple
29579 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
29581 2017-01-12 Richard Biener <rguenther@suse.de>
29583 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
29584 and TS_TARGET_OPTION directly derive from TS_BASE.
29585 * tree-core.h (tree_optimization_option): Derive from tree_base.
29586 (tree_target_option): Likewise.
29588 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
29590 * config/i386/i386.c (memory_address_length): Increase len
29591 only when rip_relative_addr_p returns false.
29593 2017-01-11 Julia Koval <julia.koval@intel.com>
29595 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
29596 (OPTION_MASK_ISA_SGX_SET): New.
29597 (ix86_handle_option): Handle OPT_msgx.
29598 * config.gcc: Added sgxintrin.h.
29599 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
29600 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
29601 * config/i386/i386.c (ix86_target_string): Add -msgx.
29603 (ix86_option_override_internal): Handle new options.
29604 (ix86_valid_target_attribute_inner_p): Add sgx.
29605 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
29606 * config/i386/i386.opt: Add msgx.
29607 * config/i386/sgxintrin.h: New file.
29608 * config/i386/x86intrin.h: Add sgxintrin.h.
29610 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29613 * fold-const.c (maybe_nonzero_address): Return 1 for function
29615 (tree_single_nonzero_warnv_p): Don't handle function local objects
29619 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
29622 2017-01-11 David Malcolm <dmalcolm@redhat.com>
29625 * opts.c: Include "spellcheck.h"
29626 (struct string_fragment): New struct.
29627 (struct edit_distance_traits<const string_fragment &>): New
29629 (get_closest_sanitizer_option): New function.
29630 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
29632 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29634 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
29636 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
29637 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
29638 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
29639 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
29640 for initial die_offset if dwarf_split_debug_info.
29641 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
29642 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
29644 (output_skeleton_debug_sections): Formatting fix. Use
29645 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
29646 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
29648 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
29650 * config/arm/cortex-a53.md: Add bypasses for
29651 cortex_a53_r2f_cvt.
29652 (cortex_a53_r2f): Only use for transfers.
29653 (cortex_a53_f2r): Likewise.
29654 (cortex_a53_r2f_cvt): Add reservation for conversions.
29655 (cortex_a53_f2r_cvt): Likewise.
29657 2017-01-11 Tamar Christina <tamar.christina@arm.com>
29659 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
29660 to all inlined functions, change static to extern.
29662 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
29665 * config/arm/arm.c (legitimize_pic_address): Handle reference to
29667 (arm_assemble_integer): Likewise.
29669 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
29671 * config.gcc: Use new awk script to check CPU, FPU and architecture
29672 parameters for --with-... options.
29673 * config/arm/parsecpu.awk: New file
29674 * config/arm/arm-cpus.in: New file.
29675 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
29677 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
29679 * config/arm/t-arm: Update dependency rules.
29680 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
29681 of processing .def files.
29682 * config/arm/genopt.sh: Deleted.
29683 * config/arm/gentune.sh: Deleted.
29684 * config/arm/arm-cores.def: Deleted.
29685 * config/arm/arm-arches.def: Deleted.
29686 * config/arm/arm-fpus.def: Deleted.
29687 * config/arm/arm-tune.md: Regenerated.
29688 * config/arm/arm-tables.opt: Regenerated.
29689 * config/arm/arm-cpu.h: New generated file.
29690 * config/arm/arm-cpu-data.h: New generated file.
29691 * config/arm/arm-cpu-cdata.h: New generated file.
29693 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
29696 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
29698 (input_varpool_node): Unpack dynamically_initialized bit.
29700 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
29702 PR rtl-optimization/79032
29703 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
29704 the alignment of the adjusted memory reference against that of MODE,
29705 instead of the alignment of the original memory reference.
29707 2017-01-11 Martin Jambor <mjambor@suse.cz>
29709 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
29711 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
29712 decorated functions.
29714 2017-01-11 Richard Biener <rguenther@suse.de>
29716 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
29717 set range/nonnull info for PHI results. Do not set it on
29718 stmts marked for removal.
29720 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
29722 * expr.c (store_field): In the bitfield case, fetch the return value
29723 from the registers before applying a single big-endian adjustment.
29724 Always do a final load for a BLKmode value not larger than a word.
29726 2017-01-10 David Malcolm <dmalcolm@redhat.com>
29729 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29730 that we correctly handle column numbers greater than
29731 LINE_MAP_MAX_COLUMN_NUMBER.
29733 2017-01-10 Martin Sebor <msebor@redhat.com>
29735 PR middle-end/78245
29736 * gimple-ssa-sprintf.c (get_destination_size): Call
29737 {init,fini}object_sizes.
29738 * tree-object-size.c (addr_object_size): Adjust.
29739 (pass_through_call): Adjust.
29740 (pass_object_sizes::execute): Adjust.
29741 * tree-object-size.h (fini_object_sizes): Declare.
29743 2017-01-10 Martin Sebor <msebor@redhat.com>
29745 PR tree-optimization/78775
29746 * builtins.c (get_size_range): Move...
29747 * calls.c: ...to here.
29748 (alloc_max_size): Accept zero argument.
29749 (operand_signed_p): Remove.
29750 (maybe_warn_alloc_args_overflow): Call get_size_range.
29751 * calls.h (get_size_range): Declare.
29753 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
29755 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
29756 from TI's devices.csv file as of September 2016.
29757 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
29759 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
29761 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
29762 * doc/invoke.texi: Likewise.
29763 * doc/md.texi: Likewise.
29764 * doc/objc.texi: Likewise.
29766 2017-01-10 Joshua Conner <joshconner@google.com>
29768 * config/arm/fuchsia-elf.h: New file.
29769 * config/fuchsia.h: New file.
29770 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
29771 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
29773 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
29775 2016-01-10 Richard Biener <rguenther@suse.de>
29777 PR tree-optimization/79034
29778 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29779 Propagate out degenerate PHIs in the joiner.
29781 2017-01-10 Martin Liska <mliska@suse.cz>
29783 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
29784 (sort_congruence_classes_by_decl_uid): Likewise.
29785 (sort_congruence_class_groups_by_decl_uid): Likewise.
29786 (sem_item_optimizer::merge_classes): Sort class, groups in these
29787 classes and members in the groups by DECL_UID of declarations.
29788 This would make merge operations stable.
29790 2017-01-10 Martin Liska <mliska@suse.cz>
29792 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
29793 usage of m_classes_vec.
29794 (sem_item_optimizer::~sem_item_optimizer): Likewise.
29795 (sem_item_optimizer::get_group_by_hash): Likewise.
29796 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
29797 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
29798 (sem_item_optimizer::verify_classes): Likewise.
29799 (sem_item_optimizer::process_cong_reduction): Likewise.
29800 (sem_item_optimizer::dump_cong_classes): Likewise.
29801 (sem_item_optimizer::merge_classes): Likewise.
29802 * ipa-icf.h (congruence_class_hash): Rename from
29803 congruence_class_group_hash. Remove declaration of m_classes_vec.
29805 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
29807 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
29808 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
29809 * config.gcc: Add avx512vpopcntdqintrin.h.
29810 * config/i386/avx512vpopcntdqintrin.h: New.
29811 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
29812 * config/i386/i386-builtin-types.def: Add new types.
29813 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
29814 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
29815 __builtin_ia32_vpopcountq_v8di_mask): New.
29816 * config/i386/i386-c.c (ix86_target_macros_internal): Define
29817 __AVX512VPOPCNTDQ__.
29818 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
29819 (PTA_AVX512VPOPCNTDQ): Define.
29820 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
29821 TARGET_AVX512VPOPCNTDQ_P): Define.
29822 * config/i386/i386.opt: Add mavx512vpopcntdq.
29823 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
29824 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
29826 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29828 PR middle-end/77484
29829 * predict.def (PRED_CALL): Set to 67.
29831 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
29833 * expr.c (store_field): In the bitfield case, if the value comes from
29834 a function call and is of an aggregate type returned in registers, do
29835 not modify the field mode; extract the value in all cases if the mode
29836 is BLKmode and the size is not larger than a word.
29838 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
29841 * config/i386/cpuid.h: Fix undefined behavior.
29843 2017-01-04 Jeff Law <law@redhat.com>
29845 PR tree-optimization/79007
29846 PR tree-optimization/67955
29847 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
29848 conservative for pt.null when flag_non_call_exceptions is on.
29850 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29852 PR translation/79019
29853 PR translation/79020
29854 * params.def (PARAM_INLINE_MIN_SPEEDUP,
29855 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
29856 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
29858 * config/avr/avr.opt (maccumulate-args): Likewise.
29859 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
29860 * common.opt (freport-bug): Likewise.
29861 * cif-code.def (CIF_FINAL_ERROR): Likewise.
29862 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
29863 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
29864 translatable string.
29865 * config/i386/i386.c (function_value_32): Likewise.
29866 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
29867 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
29869 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
29870 * common/config/msp430/msp430-common.c (msp430_handle_option):
29872 * symtab.c (symtab_node::verify_base): Likewise.
29873 * opts.c (set_debug_level): Likewise.
29874 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
29875 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
29876 missing whitespace to translatable strings.
29877 * config/avr/avr.md (bswapsi2): Fix typo in comment.
29878 * config/sh/superh.h: Likewise.
29879 * config/i386/xopintrin.h: Likewise.
29880 * config/i386/znver1.md: Likewise.
29881 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
29882 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
29883 * double-int.h (struct double_int): Likewise.
29884 * double-int.c (div_and_round_double): Likewise.
29885 * wide-int.cc: Likewise.
29886 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
29887 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
29888 * cfgcleanup.c (crossjumps_occured): Renamed to ...
29889 (crossjumps_occurred): ... this.
29890 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
29893 PR tree-optimization/78899
29894 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
29895 returning bool return struct loop *, NULL for failure and the new
29897 (versionable_outer_loop_p): Don't version outer loop if it has
29898 dont_vectorized bit set.
29899 (tree_if_conversion): When versioning outer loop, ensure
29900 tree_if_conversion is performed also on the inner loop of the
29901 non-vectorizable outer loop copy.
29902 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
29903 LOOP_VECTORIZED in inner loop of the scalar outer loop and
29904 prevent vectorization of it.
29905 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
29906 the outer loop vectorization of the non-scalar version is attempted
29907 before vectorization of the inner loop in scalar version. If
29908 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
29909 vectorization of its inner loop.
29910 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
29911 has 2 inner loops, rename also on edges from bb whose single pred
29912 is outer_loop->header. Fix typo in function comment.
29914 2017-01-09 Martin Sebor <msebor@redhat.com>
29917 * asan.c (asan_emit_stack_protection): Increase local buffer size
29918 to avoid snprintf truncation warning.
29920 2017-01-09 Andrew Pinski <apinski@cavium.com>
29922 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
29923 to reference thunderx2t99 for the tuning structure
29924 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
29926 (thunderx2t99_extra_costs): This.
29927 * config/aarch64/aarch64-tune.md: Regenerate.
29928 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
29929 (vulcan_addrcost_table): This.
29930 (vulcan_regmove_cost): Rename to ...
29931 (thunderx2t99_regmove_cost): This.
29932 (vulcan_vector_cost): Rename to ...
29933 (thunderx2t99_vector_cost): this.
29934 (vulcan_branch_cost): Rename to ...
29935 (thunderx2t99_branch_cost): This.
29936 (vulcan_tunings): Rename to ...
29937 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
29938 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
29940 2017-01-09 Martin Jambor <mjambor@suse.cz>
29944 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
29945 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
29946 (propagate_vr_accross_jump_function): Use the above function for all
29947 value range computations for pass-through jump functions and type
29948 converasion from explicit value range values.
29949 (ipcp_propagate_stage): Do not attempt to deduce types of formal
29950 parameters from TYPE_ARG_TYPES.
29951 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
29952 (ipa_write_node_info): Stream type of the actual argument.
29953 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
29955 2017-01-09 Martin Liska <mliska@suse.cz>
29958 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
29959 (lookup_compiler): Do not show error message with have_E.
29961 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29963 PR tree-optimization/78938
29964 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
29965 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
29966 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
29967 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
29970 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29972 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
29975 2017-01-09 Richard Biener <rguenther@suse.de>
29977 PR tree-optimization/78997
29978 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
29979 name condition properly.
29981 2017-01-09 Richard Biener <rguenther@suse.de>
29984 * dwarf2out.c (is_cxx): New overload with context.
29985 (is_naming_typedef_decl): Use it.
29987 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29989 * invoke.texi (Option Summary): Correct spacing in option lists
29990 and add line breaks to fix over-long lines.
29992 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29994 PR middle-end/17660
29996 * extend.texi (Common Variable Attributes): Add xref to GCC
29997 Internals manual to explain mode attribute keywords.
29999 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30002 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
30003 and Preprocessor Options.
30004 (Options for Linking): Document -pthread here....
30005 (RS/6000 and PowerPC Options): ...not here.
30006 (Solaris 2 Options): ...or here.
30007 * doc/cppopts.texi: Document -pthread.
30009 2017-01-08 Martin Sebor <msebor@redhat.com>
30011 PR middle-end/77708
30012 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
30013 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
30014 New member functions.
30015 (format_directive): Used them.
30017 (pass_sprintf_length::handle_gimple_call): Same.
30018 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
30019 to avoid truncation for any argument.
30020 (extract_affine_mul): Same.
30021 * tree.c (get_file_function_name): Same.
30023 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30025 PR middle-end/77484
30026 * predict.def (PRED_INDIR_CALL): Set to 86.
30028 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30030 PR preprocessor/54124
30031 * doc/cppopts.texi: Reformat -d subtable to list the full name
30032 of the options. Add cross-reference to the docs for the general
30033 compiler -d options.
30034 * doc/invoke.texi (Developer Options): Add cross-reference to the
30035 preprocessor-specific -d option documentation.
30037 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30039 PR preprocessor/13498
30040 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
30041 redudant material, and reflect new command-line options.
30042 (System Headers): Likewise.
30044 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30046 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
30047 -isystem, and -idirafter. Copy-edit.
30048 * doc/cppopts.texi: Copy-edit. Remove contradiction about
30049 default for -ftrack-macro-expansion. Delete obsolete and
30050 badly-formatted implementation details about -fdebug-cpp output.
30051 * doc/cppwarnopts.texi: Copy-edit.
30053 2017-01-07 David Malcolm <dmalcolm@redhat.com>
30056 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30057 that the transition from a max line width >= 1<<10 to narrower
30058 lines works correctly.
30060 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
30062 * doc/options.texi (PerFunction): New.
30063 * opt-functions.awk (switch_flags): Map both Optimization and
30064 PerFunction to CL_OPTIMIZATION.
30065 * opth-gen.awk: Test for PerFunction flag along with
30067 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
30068 it only when the latter is present. Skip those that don't in
30069 the hash function generator.
30070 * common.opt (fvar-tracking): Mark as PerFunction instead of
30072 (fvar-tracking-assignments): Likewise.
30073 (fvar-tracking-assignments-toggle): Likewise.
30074 (fvar-tracking-uninit): Likewise.
30076 2017-01-07 Jakub Jelinek <jakub@redhat.com>
30078 PR translation/79018
30079 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
30082 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
30085 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
30086 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
30087 TARGET_LONG_JUMP_TABLE_OFFSETS.
30088 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
30089 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
30090 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
30091 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
30092 * config/m68k/m68k.md (tablejump expander): Likewise.
30093 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
30094 TARGET_LONG_JUMP_TABLE_OFFSETS.
30095 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
30096 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
30098 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
30099 David Holsgrove <david.holsgrove@xilinx.com>
30101 * common/config/microblaze/microblaze-common.c
30102 (TARGET_EXCEPT_UNWIND_INFO): Remove.
30103 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
30105 * config/microblaze/microblaze.c (microblaze_must_save_register)
30106 (microblaze_expand_epilogue, microblaze_return_addr): Handle
30108 (microblaze_eh_return): New function.
30109 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
30110 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
30111 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
30112 * config/microblaze/microblaze.md (eh_return): New pattern.
30114 2017-01-06 Jakub Jelinek <jakub@redhat.com>
30116 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
30117 GCC_DIAGNOSTIC_STRINGIFY): Define.
30119 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
30121 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30123 * config/arm/arm.md (<mcrr>): New.
30125 * config/arm/arm.c (arm_arch5te): New.
30126 (arm_option_override): Set arm_arch5te.
30127 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
30129 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
30130 (arm_mcrr_qualifiers): ... this. New.
30131 (MRRC_QUALIFIERS): Define to...
30132 (arm_mrrc_qualifiers): ... this. New.
30133 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
30135 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
30136 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
30137 (MRRCI, mrrc, MRRC): New.
30138 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
30139 VUNSPEC_MRRC2): New.
30141 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30143 * config/arm/arm.md (<mcr>): New.
30145 * config/arm/arm.c (arm_coproc_builtin_available): Add
30146 support for mcr, mrc, mcr2 and mrc2.
30147 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
30148 (arm_mcr_qualifiers): ... this. New.
30149 (MRC_QUALIFIERS): Define to ...
30150 (arm_mrc_qualifiers): ... this. New.
30151 (MCR_QUALIFIERS): Define to ...
30152 (arm_mcr_qualifiers): ... this. New.
30153 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
30155 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
30156 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
30157 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
30158 VUNSPEC_MRC2): New.
30160 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30162 * config/arm/arm.md (*ldc): New.
30166 * config/arm/arm.c (arm_coproc_builtin_available): Add
30167 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
30168 (arm_coproc_ldc_stc_legitimate_address): New.
30169 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
30170 'qualifier_const_pointer'.
30171 (LDC_QUALIFIERS): Define to...
30172 (arm_ldc_qualifiers): ... this. New.
30173 (STC_QUALIFIERS): Define to...
30174 (arm_stc_qualifiers): ... this. New.
30175 * config/arm/arm-protos.h
30176 (arm_coproc_ldc_stc_legitimate_address): New.
30177 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
30178 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
30179 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
30180 stc2, stcl, stc2l): New.
30181 * config/arm/constraints.md (Uz): New.
30182 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
30183 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
30184 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
30185 VUNSPEC_STC2L): New.
30187 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30189 * config/arm/arm.md (<cdp>): New.
30190 * config/arm/arm.c (neon_const_bounds): Rename this ...
30191 (arm_const_bounds): ... this.
30192 (arm_coproc_builtin_available): New.
30193 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
30194 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
30195 (CDP_QUALIFIERS): Define to...
30196 (arm_cdp_qualifiers): ... this. New.
30198 (arm_expand_builtin_args): Add case for 6 arguments.
30199 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
30200 (arm_const_bounds): ... this.
30201 (arm_coproc_builtin_available): New.
30202 * config/arm/arm_acle.h (__arm_cdp): New.
30204 * config/arm/arm_acle_builtins.def (cdp): New.
30206 * config/arm/iterators.md (CDPI,CDP,cdp): New.
30207 * config/arm/neon.md: Rename all 'neon_const_bounds' to
30208 'arm_const_bounds'.
30209 * config/arm/types.md (coproc): New.
30210 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
30211 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
30212 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
30213 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
30215 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30217 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
30218 (UBINOP_QUALIFIERS): New.
30220 (acle_builtin_data): New. Change comment.
30221 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
30222 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
30223 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
30224 arm_acle_builtins.def.
30225 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
30226 (arm_init_acle_builtins): New.
30227 (CRC32_BUILTIN): Remove.
30228 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
30229 crc32cb, crc32ch and crc32cw.
30230 (arm_init_crc32_builtins): Remove.
30231 (arm_init_builtins): Use arm_init_acle_builtins rather
30232 than arm_init_crc32_builtins.
30233 (arm_expand_acle_builtin): New.
30234 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
30235 * config/arm/arm_acle_builtins.def: New.
30237 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30239 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
30240 (arm_builtin_datum): ... this.
30241 (arm_init_neon_builtin): Rename to ...
30242 (arm_init_builtin): ... this. Add a new parameters PREFIX
30243 and USE_SIG_IN_NAME.
30244 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
30245 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
30246 'arm_builtin_datum'.
30247 (arm_init_vfp_builtins): Likewise.
30248 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
30249 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
30250 (arm_expand_neon_args): Rename to ...
30251 (arm_expand_builtin_args): ... this. Rename builtin_arg
30252 enum values and differentiate between ARG_BUILTIN_MEMORY
30253 and ARG_BUILTIN_NEON_MEMORY.
30254 (arm_expand_neon_builtin_1): Rename to ...
30255 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
30256 values, arm_expand_builtin_args and add bool parameter NEON.
30257 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
30258 (arm_expand_vfp_builtin): Likewise.
30259 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
30261 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30263 PR middle-end/77484
30264 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
30265 * predict.c (tree_estimate_probability_bb): Reverse direction of
30266 polymorphic call predictor.
30268 2017-01-06 David Malcolm <dmalcolm@redhat.com>
30270 * passes.c (execute_one_pass): Split out pass-skipping logic into...
30271 (determine_pass_name_match): ...this new function and...
30272 (should_skip_pass_p): ...this new function.
30274 2017-01-06 Nathan Sidwell <nathan@acm.org>
30276 * ipa-visibility.c (function_and_variable_visibility): Reformat
30277 comments and long lines. Remove extrneous if.
30278 * symtab.c (symtab_node::make_decl_local): Fix code format.
30279 (symtab_node::set_section_for_node): Fix comment typo.
30281 2017-01-06 Martin Liska <mliska@suse.cz>
30284 * lra-constraints.c: Rename invariant to lra_invariant.
30285 * predict.c (set_even_probabilities): Initialize e to NULL.
30287 2017-01-05 Martin Sebor <msebor@redhat.com>
30289 PR tree-optimization/78910
30290 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
30291 (format_integer): Correct off-by-one error in the handling
30292 of precision with negative numbers in signed conversions..
30294 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
30296 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
30298 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30300 PR tree-optimization/71016
30301 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
30302 factor_out_conditional_conversion. Formatting fix.
30303 (factor_out_conditional_conversion): Add cond_stmt argument.
30304 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
30305 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
30308 2017-01-05 David Malcolm <dmalcolm@redhat.com>
30310 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
30311 read-rtl-function.o, and selftest-rtl.o.
30312 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
30313 (selftest::aarch64_test_loading_full_dump): New function.
30314 (selftest::aarch64_run_selftests): New function.
30315 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
30316 selftest::aarch64_run_selftests.
30317 * config/i386/i386.c
30318 (selftest::ix86_test_loading_dump_fragment_1): New function.
30319 (selftest::ix86_test_loading_call_insn): New function.
30320 (selftest::ix86_test_loading_full_dump): New function.
30321 (selftest::ix86_test_loading_unspec): New function.
30322 (selftest::ix86_run_selftests): Call the new functions.
30323 * emit-rtl.c (maybe_set_max_label_num): New function.
30324 * emit-rtl.h (maybe_set_max_label_num): New decl.
30325 * function.c (instantiate_decls): Guard call to
30326 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
30327 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
30329 * gensupport.c (gen_reader::gen_reader): Pass "false"
30330 for new "compact" param of rtx_reader.
30331 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
30332 rather than an empty string for NULL strings.
30333 * read-md.c: Potentially include config.h rather than bconfig.h.
30334 Wrap include of errors.h with #ifdef GENERATOR_FILE.
30335 (have_error): New global, copied from errors.c.
30336 (md_reader::read_name): Rename to...
30337 (md_reader::read_name_1): ...this, adding "out_loc" param,
30338 and converting "missing name or number" to returning false, rather
30340 (md_reader::read_name): Reimplement in terms of read_name_1.
30341 (md_reader::read_name_or_nil): New function.
30342 (md_reader::read_string): Handle "(nil)" by returning NULL.
30343 (md_reader::md_reader): Add new param "compact".
30344 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
30345 (md_reader::read_file): New method.
30346 * read-md.h (md_reader::md_reader): Add new param "compact".
30347 (md_reader::read_file): New method.
30348 (md_reader::is_compact): New accessor.
30349 (md_reader::read_name): Convert return type from void to file_location.
30350 (md_reader::read_name_or_nil): New decl.
30351 (md_reader::read_name_1): New decl.
30352 (md_reader::m_compact): New field.
30353 (noop_reader::noop_reader): Pass "false" for new "compact" param
30355 (rtx_reader::rtx_reader): Add new "compact" param.
30356 (rtx_reader::read_rtx_operand): Make virtual and convert return
30357 type from void to rtx.
30358 (rtx_reader::read_until): New decl.
30359 (rtx_reader::handle_any_trailing_information): New virtual function.
30360 (rtx_reader::postprocess): New virtual function.
30361 (rtx_reader::finalize_string): New virtual function.
30362 (rtx_reader::m_in_call_function_usage): New field.
30363 (rtx_reader::m_reuse_rtx_by_id): New field.
30364 * read-rtl-function.c: New file.
30365 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
30366 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
30367 (selftest::verify_three_block_rtl_cfg): New decl.
30368 * read-rtl-function.h: New file.
30369 * read-rtl.c: Potentially include config.h rather than bconfig.h.
30370 For host, include function.h, memmodel.h, and emit-rtl.h.
30371 (one_time_initialization): New function.
30372 (struct compact_insn_name): New struct.
30373 (compact_insn_names): New array.
30374 (find_code): Handle insn codes in compact dumps.
30375 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
30376 (bind_subst_iter_and_attr): Likewise.
30377 (add_condition_to_string): Likewise.
30378 (add_condition_to_rtx): Likewise.
30379 (apply_attribute_uses): Likewise.
30380 (add_current_iterators): Likewise.
30381 (apply_iterators): Likewise.
30382 (initialize_iterators): Guard usage of apply_subst_iterator with
30383 #ifdef GENERATOR_FILE.
30384 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
30385 (md_reader::read_mapping): Likewise.
30386 (add_define_attr_for_define_subst): Likewise.
30387 (add_define_subst_attr): Likewise.
30388 (read_subst_mapping): Likewise.
30389 (check_code_iterator): Likewise.
30390 (rtx_reader::read_rtx): Likewise. Move one-time initialization
30392 (one_time_initialization): New function.
30393 (rtx_reader::read_until): New method.
30394 (read_flags): New function.
30395 (parse_reg_note_name): New function.
30396 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
30397 Handle reuse_rtx ids.
30398 Wrap iterator lookup within #ifdef GENERATOR_FILE.
30399 Add parsing support for RTL dumps, mirroring the special-cases in
30400 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
30401 values, and calling handle_any_trailing_information.
30402 (rtx_reader::read_rtx_operand): Convert return type from void
30403 to rtx, returning return_rtx. Handle case 'e'. Call
30404 finalize_string on XSTR and XTMPL fields.
30405 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
30406 "(nil)" values were omitted. Call the postprocess vfunc on the
30408 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
30409 class ctor. Initialize m_in_call_function_usage. Call
30410 one_time_initialization.
30411 * rtl-tests.c (selftest::test_uncond_jump): Call
30412 set_new_first_and_last_insn.
30413 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
30414 * selftest-rtl.c: New file.
30415 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
30416 (selftest::get_insn_by_uid): New decl.
30417 * selftest-run-tests.c (selftest::run_tests): Call
30418 read_rtl_function_c_tests.
30419 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
30420 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
30423 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
30425 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
30426 operands in a special way. Assert that pos+len <= mode precision.
30428 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30430 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
30431 3 argument Alias with unlimited for the negative form.
30432 (fno-vect-cost-model): Removed.
30434 2017-01-05 Martin Liska <mliska@suse.cz>
30436 * hsa-gen.c (gen_hsa_divmod): New function.
30437 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
30439 2017-01-05 Martin Liska <mliska@suse.cz>
30442 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
30445 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30447 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
30448 small constant length operands.
30450 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30452 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
30453 between loop iterations.
30455 2017-01-05 Martin Liska <mliska@suse.cz>
30458 * gimplify.c (gimplify_decl_expr): Compare to
30459 asan_poisoned_variables instread of checking flags.
30460 (gimplify_target_expr): Likewise.
30461 (gimplify_expr): Likewise.
30462 (gimplify_function_tree): Conditionally initialize
30463 asan_poisoned_variables.
30465 2017-01-04 Jeff Law <law@redhat.com>
30467 PR tree-optimizatin/78812
30468 * rtl.h (contains_mem_rtx_p): Prototype.
30469 * ifcvt.c (containts_mem_rtx_p): Move from here to...
30470 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
30471 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
30472 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
30473 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
30475 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30477 * input.c (assert_char_at_range): Default-initialize actual_range.
30479 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30481 * df-scan.c (df_ref_create_structure): Make regno unsigned,
30482 to match the caller.
30484 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30486 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
30487 insns after final jump in test to emit dummy move.
30489 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30491 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
30492 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
30494 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30496 * multiple_target.c (create_dispatcher_calls): Init e_next.
30497 * tree-ssa-loop-split.c (split_loop): Init border.
30498 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
30501 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
30506 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
30507 (altivec_register_operand): Do not return true if the operand
30508 contains a SUBREG mixing SImode and SFmode.
30509 (vsx_register_operand): Likewise.
30510 (vsx_reg_sfsubreg_ok): New predicate.
30511 (vfloat_operand): Do not return true if the operand contains a
30512 SUBREG mixing SImode and SFmode.
30513 (vint_operand): Likewise.
30514 (vlogical_operand): Likewise.
30515 (gpc_reg_operand): Likewise.
30516 (int_reg_operand): Likewise.
30517 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
30518 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
30519 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
30521 (rs6000_emit_move_si_sf_subreg): New helper function.
30522 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
30523 fixup SUBREGs involving SImode and SFmode.
30524 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
30525 numbers for the new peephole2 optimization.
30526 (peephole2 for SFmode unions): New peephole2 to optimize cases in
30527 the GLIBC math library that do AND/IOR/XOR operations on single
30528 precision floating point.
30529 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
30530 target macros to say whether we need to avoid SUBREGs mixing
30532 (TARGET_ALLOW_SF_SUBREG): Likewise.
30533 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
30534 (UNSPEC_SI_FROM_SF): Likewise.
30535 (iorxor): Change spacing.
30536 (and_ior_xor): New iterator for AND, IOR, and XOR.
30537 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
30538 (movdi_from_sf_zero_ext): Likewise.
30539 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
30540 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
30541 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
30542 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
30543 (fms<mode>4): Likewise.
30544 (fnma<mode>4): Likewise.
30545 (fnms<mode>4): Likewise.
30546 (nfma<mode>4): Likewise.
30547 (nfms<mode>4): Likewise.
30549 2017-01-04 Marek Polacek <polacek@redhat.com>
30552 * doc/invoke.texi: Document -Wpointer-compare.
30554 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30556 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
30559 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
30560 descriptions for -gdwarf-5 and emit them as uleb128 instead of
30563 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
30566 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
30567 documentation of the powerpc_popcntb_ok attribute.
30568 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30569 code to issue warning messages if a requested CPU configuration is
30570 not supported by the binary (assembler and loader) toolchain.
30571 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
30572 made to define a built-in function that has been disabled.
30573 (paired_init_builtins): Add assertion to prevent ICE if attempt is
30574 made to define a built-in function that has been disabled.
30575 (altivec_init_builtins): Add comment explaining why definition
30576 of the DST built-in functions is not preceded by an assertion
30577 check. Add assertions to prevent ICE if attempts are made to
30578 define an altivec predicate or an abs* built-in function that has
30580 (htm_init_builtins): Add comment explaining why definition of the
30581 htm built-in functions is not preceded by an assertion check.
30583 2017-01-04 Jeff Law <law@redhat.com>
30585 PR tree-optimizatin/67955
30586 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
30587 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
30588 the points-to solution does not include pt_null. Use DECL_PT_UID
30591 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
30593 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
30594 Use gen_int_mode instead of gen_lopwart for const_int operands.
30596 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30598 PR tree-optimization/71563
30599 * match.pd: Simplify X << Y into X if Y is known to be 0 or
30600 out of range value - has low bits known to be zero.
30602 2017-01-04 Alan Modra <amodra@gmail.com>
30604 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
30605 * configure: Regenerate.
30606 * config.in: Regenerate.
30608 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30611 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
30612 a substring of the message, but strcmp with the whole message. Ifdef
30613 ENABLE_NLS, translate the message first using dgettext.
30615 2017-01-03 Jeff Law <law@redhat.com>
30617 PR tree-optimizatin/78856
30618 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
30619 (mark_threaded_blocks): Remove code to truncate thread paths that
30620 cross multiple loop headers. Instead invalidate the cached loop
30621 iteration information and handle case of a thread path walking
30622 into an irreducible region.
30624 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
30627 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
30628 assertions. Add support for doing the signbit if the IEEE 128-bit
30629 floating point value is in a GPR.
30630 * config/rs6000/rs6000.md (Fsignbit): Delete.
30631 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
30632 Update the length attribute if the value is in a GPR.
30633 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
30634 the sign or zero extension instruction, since the value is always 0/1.
30635 (signbit<mode>2_dm2): Delete using <Fsignbit>.
30638 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
30639 extracting SImode to a GPR register so that we can generate a
30640 store, limit the vector to be in a traditional Altivec register
30641 for the vextuwrx instruction.
30643 2017-01-03 Ian Lance Taylor <iant@google.com>
30645 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
30647 2017-01-03 Martin Sebor <msebor@redhat.com>
30649 PR tree-optimization/78696
30650 * gimple-ssa-sprintf.c (format_floating): Correct handling of
30651 precision. Use MPFR for %f for greater fidelity. Correct handling
30653 (pass_sprintf_length::compute_format_length): Set width and precision
30654 specified by asrerisk to void_node for vararg functions.
30655 (try_substitute_return_value): Adjust dump output.
30657 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
30659 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
30661 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
30663 * doc/invoke.texi (SPARC options): Document -mlra as the default.
30664 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
30665 -mlra/-mno-lra was passed to the compiler.
30667 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
30669 PR rtl-optimization/65618
30670 * emit-rtl.c (try_split): Move initialization of "before" and
30671 "after" to just before the call to emit_insn_after_setloc.
30673 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
30675 * doc/md.texi (Standard Names): Remove reference to Java frontend.
30677 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
30679 * dwarf2out.c (gen_enumeration_type_die): When
30680 -gno-strict-dwarf, add a DW_AT_encoding attribute.
30682 2017-01-03 Jakub Jelinek <jakub@redhat.com>
30684 PR tree-optimization/78965
30685 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
30686 Change first argument from const call_info & to call_info &. For %n
30687 set info.nowrite to false.
30689 PR middle-end/78901
30690 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
30691 possibly throwing calls.
30693 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
30694 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
30695 and fns handling, rather than in a separate case SSA_NAME.
30697 2017-01-02 Jeff Law <law@redhat.com>
30699 * config/darwin-driver.c (darwin_driver_init): Const-correctness
30700 fixes for first_period and second_period variables.
30702 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
30705 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
30706 (*insvqi_1): New insn pattern.
30707 (*insvqi_1_mem_rex64): Ditto.
30708 (*insvqi_2): Ditto.
30709 (*insvqi_3): Rename from *insvqi.
30711 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
30713 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
30715 * doc/cfg.texi (Edges): Remove reference to Java.
30716 (Maintaining the CFG): Ditto.
30718 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30720 PR middle-end/77674
30721 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
30722 transparent aliases.
30724 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30726 PR middle-end/77484
30727 * predict.def (PRED_CALL): Update hitrate.
30728 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
30729 * predict.c (tree_estimate_probability_bb): Split CALL predictor
30730 into direct/indirect/polymorphic variants.
30732 2017-01-01 Jakub Jelinek <jakub@redhat.com>
30734 Update copyright years.
30736 * gcc.c (process_command): Update copyright notice dates.
30737 * gcov-dump.c (print_version): Ditto.
30738 * gcov.c (print_version): Ditto.
30739 * gcov-tool.c (print_version): Ditto.
30740 * gengtype.c (create_file): Ditto.
30741 * doc/cpp.texi: Bump @copying's copyright year.
30742 * doc/cppinternals.texi: Ditto.
30743 * doc/gcc.texi: Ditto.
30744 * doc/gccint.texi: Ditto.
30745 * doc/gcov.texi: Ditto.
30746 * doc/install.texi: Ditto.
30747 * doc/invoke.texi: Ditto.
30749 Copyright (C) 2017 Free Software Foundation, Inc.
30751 Copying and distribution of this file, with or without modification,
30752 are permitted in any medium without royalty provided the copyright
30753 notice and this notice are preserved.