1 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4 (tree-ssa-loop-ivopts.h): New header file.
5 (struct builtin_info): New fields.
6 (classify_builtin_1): Compute and record base and offset parts for
7 memset builtin partition by calling strip_offset.
8 (offset_cmp, fuse_memset_builtins): New functions.
9 (finalize_partitions): Fuse adjacent memset partitions by calling
11 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
13 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
15 2017-10-18 Bin Cheng <bin.cheng@arm.com>
17 PR tree-optimization/82574
18 * tree-loop-distribution.c (find_single_drs): New parameter. Check
19 that data reference must be executed exactly once per iteration
20 against the outermost loop in nest.
21 (classify_partition): Update call to above function.
23 2017-10-18 Richard Biener <rguenther@suse.de>
25 PR tree-optimization/82591
26 * graphite.c (graphite_transform_loops): Move code gen message
28 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
29 Here. Handle scop_to_isl_ast failing.
30 (scop_to_isl_ast): Limit the number of ISL operations.
32 2017-10-18 Richard Biener <rguenther@suse.de>
34 * graphite-isl-ast-to-gimple.c
35 (translate_isl_ast_to_gimple::set_rename): Simplify.
36 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
37 (graphite_copy_stmts_from_block): ... here.
38 (copy_bb_and_scalar_dependences): Simplify.
39 (add_parameters_to_ivs_params): Canonicalize.
40 (generate_entry_out_of_ssa_copies): Simplify.
41 * graphite-sese-to-poly.c (extract_affine_name): Simplify
42 by passing in ISL dimension.
43 (parameter_index_in_region_1): Rename to ...
44 (parameter_index_in_region): ... this.
45 (extract_affine): Adjust assert, pass down parameter index.
46 (add_param_constraints): Use range-info when available.
47 (build_scop_context): Adjust.
48 * sese.c (new_sese_info): Adjust.
49 (free_sese_info): Likewise.
50 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
51 Remove unused typedefs.
52 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
54 2017-10-18 Martin Liska <mliska@suse.cz>
56 * combine.c (simplify_compare_const): Add gcc_fallthrough.
58 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
60 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
61 (s390_sched_init): Do not reset s390_sched_state if we entered the
62 current basic block via a fallthru edge and all others are unlikely.
64 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
66 * config/s390/s390.c (NUM_SIDES): New variable.
67 (LONGRUNNING_THRESHOLD): New variable.
68 (LATENCY_FACTOR): New variable.
69 (s390_sched_score): Decrease score for long-running instructions on
71 (s390_sched_variable_issue): Perform bookkeeping for long-running
74 2017-10-18 Richard Biener <rguenther@suse.de>
76 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
77 Simplify with removal of the parameter rename map.
78 (set_rename): Likewise.
79 (should_copy_to_new_region): Likewise.
80 (graphite_copy_stmts_from_block): Likewise.
81 (copy_bb_and_scalar_dependences): Remove initialization of
84 (copy_internal_parameters): Likewise.
85 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
86 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
88 (parameter_index_in_region_1): Rename to ...
89 (assign_parameter_index_in_region): ... this. Assert we have
90 a parameter we handle.
91 (scan_tree_for_params): Adjust.
92 * sese.h (parameter_rename_map_t): Remove.
93 (struct sese_info_t): Remove unused parameter_rename_map and
94 copied_bb_map members.
95 * sese.c (new_sese_info): Adjust.
96 (free_sese_info): Likewise.
98 2017-10-18 Martin Liska <mliska@suse.cz>
101 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
104 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
106 * doc/invoke.texi (ffunction-sections and fdata-sections):
109 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
111 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
112 the use statement can throw internally.
114 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
116 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
117 any RTX present on the RHS of a SET.
118 * compare-elim.c (try_eliminate_compare): Restore comment.
120 2017-10-17 Jakub Jelinek <jakub@redhat.com>
122 * langhooks.h (struct lang_hooks): Document that tree_size langhook
123 may be also called on tcc_type nodes.
124 * langhooks.c (lhd_tree_size): Likewise.
126 2017-10-17 David Malcolm <dmalcolm@redhat.com>
128 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
129 format_warning_at_substring.
130 (maybe_warn): Convert source_range * param to a location_t. Pass
131 UNKNOWN_LOCATION rather than NULL to fmtwarn.
132 (format_directive): Remove code to extract source_ranges and
133 source_range * in favor of just a location_t.
134 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
136 * substring-locations.c (format_warning_va): Convert
137 source_range * param to a location_t.
138 (format_warning_at_substring): Likewise.
139 * substring-locations.h (format_warning_va): Likewise.
140 (format_warning_at_substring): Likewise.
142 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
144 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
146 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
147 and scatter/gather ops.
149 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
150 vec_gather_load and vec_scatter_store.
151 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
152 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
153 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
154 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
155 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
156 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
158 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
160 * reg-stack.c (compare_for_stack_reg): Add bool argument.
161 Detect FTST instruction and handle its register pops. Only pop
162 second operand if can_pop_second_op is true.
163 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
164 set can_pop_second_op to false in the compare_for_stack_reg call.
166 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
167 output_fp_compare for stack register operands.
168 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
169 instructions here. Do not emit stack register pops here. Assert
170 that FCOMPP pops next to top stack register. Rewrite function.
172 2017-10-17 Nathan Sidwell <nathan@acm.org>
175 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
176 use DECL_ASSEMBLER_NAME_RAW.
179 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
182 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
183 Wilco Dijkstra <wilco.dijkstra@arm.com>
185 * builtins.c (expand_builtin_update_setjmp_buf): Add a
186 converstion to Pmode from the buf_addr.
188 2017-10-17 Richard Biener <rguenther@suse.de>
190 * graphite-dependences.c (scop_get_reads_and_writes): Change
191 output parameters to references.
193 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
195 PR 71026/tree-optimization
196 * fold-const.c (distribute_real_division): Removed.
197 (fold_binary_loc): Remove calls to distribute_real_divison.
199 2017-10-17 Richard Biener <rguenther@suse.de>
201 * graphite-scop-detection.c
202 (scop_detection::stmt_has_simple_data_refs_p): Always use
203 the full nest as region.
204 (try_generate_gimple_bb): Likewise.
205 * sese.c (scalar_evolution_in_region): Simplify now that
206 SCEV can handle instantiation in regions.
207 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
208 in the non-loop part of a function if requested.
210 2017-10-17 Richard Biener <rguenther@suse.de>
212 PR tree-optimization/82563
213 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
215 (graphite_regenerate_ast_isl): Call it.
216 * graphite-scop-detection.c (build_scops): Remove entry edge split.
218 2017-10-17 Jakub Jelinek <jakub@redhat.com>
220 PR tree-optimization/82549
221 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
222 Formatting fixes. Instead of calling make_bit_field_ref with negative
225 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
227 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
228 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
229 _mm_maskz_reduce_ss): New.
230 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
231 __builtin_ia32_reducess_mask): Ditto..
232 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
233 * config/i386/sse.md (reduces<mode>): Renamed to ...
234 (reduces<mode><mask_scalar_name>): ... this.
235 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
237 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
238 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
240 2017-10-16 David Malcolm <dmalcolm@redhat.com>
242 * Makefile.in (OBJS): Add unique-ptr-tests.o.
243 * selftest-run-tests.c (selftest::run_tests): Call
244 selftest::unique_ptr_tests_cc_tests.
245 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
246 * unique-ptr-tests.cc: New file.
248 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
251 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
253 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
254 (make_hard_regno_born, make_hard_regno_dead): Update
255 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
257 2017-10-16 Jeff Law <law@redhat.com>
259 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
261 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
263 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
265 2017-10-16 Olivier Hainque <hainque@adacore.com>
267 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
268 with_cpu if we were configured for an e500v2 target cpu name.
270 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
272 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
273 * doc/invoke.texi: Document +nodsp as a valid extension for
276 2017-10-16 Martin Liska <mliska@suse.cz>
278 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
279 (test_set_range): Likewise.
280 (test_range_functions): Rename to ...
281 (test_bit_in_range): ... this.
282 (sbitmap_c_tests): Add new test.
284 2017-10-16 Tamar Christina <tamar.christina@arm.com>
286 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
287 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
288 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
290 2017-10-16 Tamar Christina <tamar.christina@arm.com>
292 * config/aarch64/aarch64-builtins.c
293 (aarch64_types_quadopu_lane_qualifiers): New.
294 (TYPES_QUADOPU_LANE): New.
295 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
296 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
297 (aarch64_<sur>dot_laneq<vsi2qi>): New.
298 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
299 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
300 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
301 (Vdottype, DOTPROD): New.
302 (sur): Add SDOT and UDOT.
304 2017-10-16 Tamar Christina <tamar.christina@arm.com>
306 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
307 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
308 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
309 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
310 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
311 (cortex-a75.cortex-a55): Likewise.
312 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
314 2017-10-16 Tamar Christina <tamar.christina@arm.com>
316 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
317 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
318 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
319 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
320 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
321 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
322 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
323 * config/arm/types.md (neon_dot, neon_dot_q): New.
324 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
326 2017-10-16 Tamar Christina <tamar.christina@arm.com>
328 * config/arm/arm.h (TARGET_DOTPROD): New.
329 * config/arm/arm.c (arm_arch_dotprod): New.
330 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
331 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
332 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
333 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
334 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
335 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
336 * doc/invoke.texi (armv8.2-a): Document dotprod
338 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
340 * i386.c (ix86_vec_cost): New function.
341 (ix86_rtx_costs): Handle vector operations better.
342 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
343 * x86-tune-costs.h: Add new costs to all tables.
345 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
347 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
349 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
350 divsd, sqrtss and sqrtsd
351 * x86-tune-costs.h: Add new entries to all costs.
352 (znver1_cost): Fix to match real instruction latencies.
354 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
355 Michael Collison <michael.collison@arm.com>
357 * compare-elim.c: Include emit-rtl.h.
358 (can_merge_compare_into_arith): New function.
359 (try_validate_parallel): Likewise.
360 (try_merge_compare): Likewise.
361 (try_eliminate_compare): Call the above when no previous clobber
363 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
366 2017-10-14 Jakub Jelinek <jakub@redhat.com>
370 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
371 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
375 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
376 to be any operand_equal_p operands. For & (B - 1) require
377 B to be power of 2. Recognize
378 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
380 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
383 * optabs.c (expand_memory_blockage): Fix call of
384 targetm.have_memory_blockage.
386 2017-10-14 Jakub Jelinek <jakub@redhat.com>
389 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
390 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
391 objects to extra_objs instead of overwriting it.
393 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
395 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
396 Use any_fp_register_operand as operand[3] predicate. Simplify
397 equality test for operands[2] and operands[4] memory location.
398 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
399 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
400 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
401 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
402 any_fp_register_operand as operand[1] predicate. Simplify
403 equality test for operands[0] and operands[3] memory location.
404 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
405 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
406 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
408 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
410 * target-insns.def: Add memory_blockage.
411 * optabs.c (expand_memory_blockage): New function.
412 (expand_asm_memory_barrier): Rename ...
413 (expand_asm_memory_blockage): ... to this.
414 (expand_mem_thread_fence): Call expand_memory_blockage
415 instead of expand_asm_memory_barrier.
416 (expand_mem_singnal_fence): Ditto.
417 (expand_atomic_load): Ditto.
418 (expand_atomic_store): Ditto.
419 * doc/md.texi (Standard Pattern Names For Generation):
420 Document memory_blockage instruction pattern.
422 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
424 * config/rl78/rl78.c (rl78_emit_libcall): New function.
425 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
426 * config/rl78/rl78.md: New define_expand "adddi3".
428 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
430 * cfghooks.c (verify_flow_info): Disable check that all probabilities
433 2017-10-13 Jeff Law <law@redhat.com>
435 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
437 2017-10-13 Jakub Jelinek <jakub@redhat.com>
440 * internal-fn.c (expand_mul_overflow): If both operands have
441 the same highpart of -1 or 0 and the topmost bit of lowpart
442 is different, overflow is if res <= 0 rather than res < 0.
444 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
446 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
447 TARGET_P9_VECTOR code for unaligned_load case.
449 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
451 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
453 2017-10-13 Nathan Sidwell <nathan@acm.org>
455 * tree-core.h (tree_contains_struct): Make bool.
456 * tree.c (tree_contains_struct): Likewise.
457 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
458 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
459 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
460 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
461 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
463 2017-10-13 Richard Biener <rguenther@suse.de>
465 * graphite-isl-ast-to-gimple.c
466 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
467 parameters and dominance check.
468 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
469 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
470 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
471 Do not update SSA form here or do intermediate IL verification.
472 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
473 (graphite_initialize): Remove check on the number of loops in
474 the function and inline into graphite_transform_loops.
475 (graphite_finalize): Inline into graphite_transform_loops.
476 (graphite_transform_loops): Perform SSA update and IL verification
478 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
480 2017-10-13 Richard Biener <rguenther@suse.de>
482 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
483 graphite_expression_type_precision): Avoid global constructor
485 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
486 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
487 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
488 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
489 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
490 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
492 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
495 * config/i386/i386.h (ix86_red_zone_size): New.
496 * config/i386/i386.md (push peephole2s): Replace
497 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
499 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
500 Alan Hayward <alan.hayward@arm.com>
501 David Sherwood <david.sherwood@arm.com>
503 * combine.c (can_change_dest_mode): Reject changes in
504 REGMODE_NATURAL_SIZE.
506 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
507 Alan Hayward <alan.hayward@arm.com>
508 David Sherwood <david.sherwood@arm.com>
510 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
511 (expand_debug_source_expr): Likewise.
512 * combine.c (combine_simplify_rtx): Likewise.
513 * cse.c (fold_rtx): Likewise.
514 * fwprop.c (canonicalize_address): Likewise.
515 * targhooks.c (default_shift_truncation_mask): Likewise.
517 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
518 Alan Hayward <alan.hayward@arm.com>
519 David Sherwood <david.sherwood@arm.com>
521 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
522 (widened_mode): Likewise.
523 (expand_unop): Likewise.
524 * ree.c (transform_ifelse): Likewise.
525 (merge_def_and_ext): Likewise.
526 (combine_reaching_defs): Likewise.
527 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
529 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
530 Alan Hayward <alan.hayward@arm.com>
531 David Sherwood <david.sherwood@arm.com>
533 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
534 * combine.c (gen_lowpart_for_combine): Likewise.
535 * dwarf2out.c (rtl_for_decl_location): Likewise.
536 * final.c (alter_subreg): Likewise.
537 * rtlhooks.c (gen_lowpart_general): Likewise.
538 (gen_lowpart_if_possible): Likewise.
540 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
541 Alan Hayward <alan.hayward@arm.com>
542 David Sherwood <david.sherwood@arm.com>
544 * calls.c (expand_call): Use subreg_lowpart_offset.
545 * cse.c (cse_insn): Likewise.
546 * regcprop.c (copy_value): Likewise.
547 (copyprop_hardreg_forward_1): Likewise.
549 2017-10-13 Jakub Jelinek <jakub@redhat.com>
552 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
553 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
554 =Q constraints to +Q and into insn condition add check
555 that operands[0] and operands[1] are equal.
556 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
557 =Q constraints to +Q and into insn condition add check
558 that operands[0] is equal to either operands[1] or operands[2].
561 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
562 instead of handling MINUS_EXPR twice (once for each argument),
563 canonicalize operand order and handle just once, use rtype where
564 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
567 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
568 any values of __C while still being pattern recognizable as a simple
571 2017-10-13 Richard Biener <rguenther@suse.de>
573 PR tree-optimization/82451
575 2017-10-02 Richard Biener <rguenther@suse.de>
577 PR tree-optimization/82355
578 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
579 a mapping for the enclosing loop but avoid generating one for
581 (copy_bb_and_scalar_dependences): Remove premature codegen
582 error on PHIs in blocks duplicated into multiple places.
583 * graphite-scop-detection.c
584 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
585 in the region use it as loop and nest to analyze the DR in.
586 (try_generate_gimple_bb): Likewise.
587 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
588 (add_loop_constraints): For blocks in a loop not in the region
589 create a dimension with a single iteration.
590 * sese.h (gbb_loop_at_index): Remove assert.
592 * cfgloop.c (loop_preheader_edge): For the loop tree root
593 return the single successor of the entry block.
594 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
595 Reset the SCEV hashtable and niters.
596 * graphite-scop-detection.c
597 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
598 assert that we only have POLYNOMIAL_CHREC that vary in loops
599 contained in the region.
600 (scop_detection::graphite_can_represent_expr): Adjust.
601 (scop_detection::stmt_has_simple_data_refs_p): For loops
602 not in the region set loop to NULL. The nest is now the
603 entry edge to the region.
604 (try_generate_gimple_bb): Likewise.
605 * sese.c (scalar_evolution_in_region): Adjust for
606 instantiate_scev change.
607 * tree-data-ref.h (graphite_find_data_references_in_stmt):
608 Make nest parameter the edge into the region.
609 (create_data_ref): Likewise.
610 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
611 entry edge into a region and adjust instantiate_scev calls.
612 (create_data_ref): Likewise.
613 (graphite_find_data_references_in_stmt): Likewise.
614 (find_data_references_in_stmt): Pass the loop preheader edge
615 from the nest argument.
616 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
617 parameter the edge into the region.
618 (instantiate_parameters): Use the loop preheader edge as entry.
619 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
621 (get_instantiated_value_entry): Make instantiate_below parameter
622 the edge into the region.
623 (instantiate_scev_name): Likewise. Adjust dominance checks,
624 when we cannot use loop-based instantiation instantiate by
625 walking use-def chains.
626 (instantiate_scev_poly): Adjust.
627 (instantiate_scev_binary): Likewise.
628 (instantiate_scev_convert): Likewise.
629 (instantiate_scev_not): Likewise.
630 (instantiate_array_ref): Remove.
631 (instantiate_scev_3): Likewise.
632 (instantiate_scev_2): Likewise.
633 (instantiate_scev_1): Likewise.
634 (instantiate_scev_r): Do not blindly handle N-operand trees.
635 Do not instantiate array-refs. Handle all constants and invariants.
636 (instantiate_scev): Make instantiate_below parameter
637 the edge into the region.
638 (resolve_mixers): Use the loop preheader edge for the region
639 parameter to instantiate_scev_r.
640 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
642 2017-10-13 Richard Biener <rguenther@suse.de>
644 PR tree-optimization/82525
645 * graphite-isl-ast-to-gimple.c
646 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
648 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
649 Fail code generation when we cannot represent the isl integer.
650 (binary_op_to_tree): Elide modulo operations that are no-ops
651 in the type we code generate. Remove now superfluous code
654 2017-10-13 Richard Biener <rguenther@suse.de>
656 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
657 (scop_detection::harmful_loop_in_region): Remove premature
659 (scop_detection::graphite_can_represent_scev): We can handle
660 pointer IVs just fine.
662 2017-10-13 Alan Modra <amodra@gmail.com>
664 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
665 "Clobbers and Scratch Registers". Add paragraph on
666 alternative to clobbers for scratch registers and OpenBLAS
669 2017-10-13 Alan Modra <amodra@gmail.com>
671 * doc/extend.texi (Clobbers): Correct vax example. Delete old
672 example of a memory input for a string of known length. Move
673 commentary out of table. Add a number of new examples
674 covering array memory inputs.
676 2017-10-12 Martin Liska <mliska@suse.cz>
678 PR tree-optimization/82493
679 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
680 (test_range_functions): New function.
681 (sbitmap_c_tests): Likewise.
682 * selftest-run-tests.c (selftest::run_tests): Run new tests.
683 * selftest.h (sbitmap_c_tests): New function.
685 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
687 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
689 * config/rs6000/amo.h: Fix spacing issue.
691 2017-10-12 Jakub Jelinek <jakub@redhat.com>
694 * config/i386/i386.md (*ashl<mode>3_mask_1,
695 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
696 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
699 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
701 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
702 (profile_probability): Set max_probability
703 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
705 * profile-count.c (profile_probability::differs_from_p): Do not
706 rely on max_probaiblity == 10000
708 2017-10-12 Jeff Law <law@redhat.com>
710 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
713 2017-10-12 Martin Sebor <msebor@redhat.com>
717 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
718 (handle_alias_pairs): Call it.
719 * common.opt (-Wattribute-alias): New option.
720 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
721 * doc/invoke.texi (-Wattribute-alias): Document.
723 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
726 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
728 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
730 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
731 (make_hard_regno_born, make_hard_regno_dead): Update
732 bb_killed_pseudos and bb_gen_pseudos.
734 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
736 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
738 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
740 * config/alpha/alpha.c (alpha_split_conditional_move):
741 Use std::swap instead of manually swapping.
742 (alpha_stdarg_optimize_hook): Ditto.
743 (alpha_canonicalize_comparison): Ditto.
745 2017-10-12 Bin Cheng <bin.cheng@arm.com>
747 * tree-loop-distribution.c (struct builtin_info): New struct.
748 (struct partition): Refactor fields into struct builtin_info.
749 (partition_free): Free struct builtin_info.
750 (build_size_arg_loc, build_addr_arg_loc): Delete.
751 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
752 information from struct builtin_info.
753 (find_single_drs): New function refactored from classify_partition.
754 Also moved builtin validity checks to this function.
755 (compute_access_range, alloc_builtin): New functions.
756 (classify_builtin_st, classify_builtin_ldst): New functions.
757 (classify_partition): Refactor code into functions find_single_drs,
758 classify_builtin_st and classify_builtin_ldst.
759 (distribute_loop): Don't do runtime alias check when distributing
761 (find_seed_stmts_for_distribution): New function.
762 (pass_loop_distribution::execute): Refactor code finding seed
763 stmts into above function. Support distribution for the innermost
764 two-level loop nest. Adjust dump information.
766 2017-10-12 Bin Cheng <bin.cheng@arm.com>
768 * tree-loop-distribution.c: Adjust the general comment.
769 (NUM_PARTITION_THRESHOLD): New macro.
770 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
771 (classify_partition): Skip builtin pattern of loop nest's inner loop.
772 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
773 in call to build_partition_graph.
774 (finalize_partitions): New parameter. Make loop distribution more
775 conservative by fusing more partitions.
776 (distribute_loop): Don't do runtime alias check in case of loop nest
778 (find_seed_stmts_for_distribution): New function.
779 (prepare_perfect_loop_nest): New function.
780 (pass_loop_distribution::execute): Refactor code finding seed stmts
781 and loop nest into above functions. Support loop nest distribution.
782 Adjust dump information accordingly.
784 2017-10-12 Bin Cheng <bin.cheng@arm.com>
786 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
787 and set PTYPE_SEQUENTIAL for merged partition.
789 2017-10-12 Richard Biener <rguenther@suse.de>
791 PR tree-optimization/69728
793 2017-09-19 Richard Biener <rguenther@suse.de>
795 PR tree-optimization/69728
796 * graphite-sese-to-poly.c (schedule_error): New global.
797 (add_loop_schedule): Handle empty domain by failing the
799 (build_original_schedule): Handle schedule_error.
801 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
802 domain by returning an unchanged schedule.
804 2017-10-12 Jakub Jelinek <jakub@redhat.com>
806 * genrecog.c (validate_pattern): For VEC_SELECT verify that
807 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
809 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
811 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
814 2017-10-12 Jakub Jelinek <jakub@redhat.com>
817 * expr.c (store_field): Don't optimize away bitsize == 0 store
818 from CALL_EXPR with addressable return type.
820 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
822 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
823 * config/rs6000/rs6000.md (sel): Delete mode attribute.
824 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
825 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
826 TARGET_ISEL instead of TARGET_ISEL<sel>.
828 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
830 * config/rs6000/rs6000.c
831 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
833 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
835 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
837 (reg_or_cint_operand): Fix comment.
838 (reg_or_zero_operand): New predicate.
839 * config/rs6000/rs6000-protos.h (output_isel): Delete.
840 * config/rs6000/rs6000.c (output_isel): Delete.
841 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
842 instead of reg_or_cint_operand. Output instruction directly (not via
844 (isel_unsigned_<mode>): Ditto.
845 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
846 gpc_reg_operand. Add an instruction alternative for this. Output
847 instruction directly.
848 (*isel_reversed_unsigned_<mode>): Ditto.
850 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
852 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
853 (TARGET_CANONICALIZE_COMPARISON): Define.
855 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
858 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
859 Check whether the dest is REG before adding REG_EQUIV note.
861 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
864 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
866 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
867 (make_hard_regno_born, make_hard_regno_dead): Update
868 bb_killed_pseudos and bb_gen_pseudos.
870 2017-10-11 Nathan Sidwell <nathan@acm.org>
872 * incpath.h (enum incpath_kind): Name enum, prefix values.
873 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
874 * incpath.c (heads, tails): Use INC_MAX.
875 (add_env_var_paths, add_standard_paths): Use incpath_kind.
876 (merge_include_chains, split_quote_chain,
877 register_include_chains): Update incpath_kind names.
878 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
879 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
881 (add_framework_path, darwin_register_objc_includes): Likewise.
882 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
884 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
886 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
887 Do not use float_operator operator predicate.
888 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
889 * config/i386/predicates.md (float_operator): Remove predicate.
891 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
893 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
894 (*jccxf_i387): Ditto.
895 (*jcc<mode>_i387): Ditto.
896 (*jccu<mode>_i387): Ditto.
897 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
898 (*jcc_*_i387 splitters): Remove.
899 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
900 * config/i386/i386.c (ix86_split_fp_branch): Remove.
901 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
904 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
906 * profile-count.h (slow_safe_scale_64bit): New function.
907 (safe_scale_64bit): New inline.
908 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
909 * profile-count.c: Include wide-int.h
910 (slow_safe_scale_64bit): New.
912 2017-10-11 Nathan Sidwell <nathan@acm.org>
914 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
915 HAS_DECL_ASSEMBLER_NAME_P.
916 * gimple-expr.c (gimple_decl_printable_name: Check
917 HAS_DECL_ASSEMBLER_NAME_P too.
918 * ipa-utils.h (type_in_anonymous_namespace_p): Check
919 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
920 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
921 * passes.c (rest_of_decl_compilation): Check
922 HAS_DECL_ASSEMBLER_NAME_P too.
923 * recog.c (verify_changes): Likewise.
924 * tree-pretty-print.c (dump_decl_name): Likewise.
925 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
927 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
928 (SET_DECL_ASSEMBLER_NAME): Use it.
929 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
930 (COPY_DECL_ASSEMBLER_NAME): Likewise.
931 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
933 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
935 * config.gcc (i386, x86_64): Add extra objects.
936 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
937 (ix86_min_insn_size): Declare.
938 (ix86_issue_rate): Declare.
939 (ix86_adjust_cost): Declare.
940 (ia32_multipass_dfa_lookahead): Declare.
941 (ix86_macro_fusion_p): Declare.
942 (ix86_macro_fusion_pair_p): Declare.
943 (ix86_bd_has_dispatch): Declare.
944 (ix86_bd_do_dispatch): Declare.
945 (ix86_core2i7_init_hooks): Declare.
946 (ix86_atom_sched_reorder): Declare.
947 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
948 (COSTS_N_BYTES): Move to x86-tune-costs.h.
949 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
950 (rip_relative_addr_p): Rename to ...
951 (ix86_rip_relative_addr_p): ... this one; export.
952 (memory_address_length): Update.
953 (ix86_issue_rate): Move to x86-tune-sched.c.
954 (ix86_flags_dependent): Move to x86-tune-sched.c.
955 (ix86_agi_dependent): Move to x86-tune-sched.c.
956 (exact_dependency_1): Move to x86-tune-sched.c.
957 (exact_store_load_dependency): Move to x86-tune-sched.c.
958 (ix86_adjust_cost): Move to x86-tune-sched.c.
959 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
960 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
961 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
962 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
963 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
964 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
965 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
966 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
967 (min_insn_size): Rename to ...
968 (ix86_min_insn_size): ... this one; export.
969 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
970 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
971 (core2i7_first_cycle_multipass_backtrack): Move to
972 x86-tune-sched-core.c.
973 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
974 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
975 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
976 (ix86_avoid_jump_mispredicts): Update.
977 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
978 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
979 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
980 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
981 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
982 (MAX_INSN): Move to ix86-tune-sched-bd.c.
983 (MAX_IMM): Move to ix86-tune-sched-bd.c.
984 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
985 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
986 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
987 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
988 (MAX_STORE): Move to ix86-tune-sched-bd.c.
989 (BIG): Move to ix86-tune-sched-bd.c.
990 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
991 (enum insn_path): Move to ix86-tune-sched-bd.c.
992 (get_mem_group): Move to ix86-tune-sched-bd.c.
993 (is_cmp): Move to ix86-tune-sched-bd.c.
994 (dispatch_violation): Move to ix86-tune-sched-bd.c.
995 (is_branch): Move to ix86-tune-sched-bd.c.
996 (is_prefetch): Move to ix86-tune-sched-bd.c.
997 (init_window): Move to ix86-tune-sched-bd.c.
998 (allocate_window): Move to ix86-tune-sched-bd.c.
999 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
1000 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
1001 (process_end_window): Move to ix86-tune-sched-bd.c.
1002 (allocate_next_window): Move to ix86-tune-sched-bd.c.
1003 (find_constant): Move to ix86-tune-sched-bd.c.
1004 (get_num_immediates): Move to ix86-tune-sched-bd.c.
1005 (has_immediate): Move to ix86-tune-sched-bd.c.
1006 (get_insn_path): Move to ix86-tune-sched-bd.c.
1007 (get_insn_group): Move to ix86-tune-sched-bd.c.
1008 (count_num_restricted): Move to ix86-tune-sched-bd.c.
1009 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
1010 (add_insn_window): Move to ix86-tune-sched-bd.c.
1011 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
1012 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
1013 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
1014 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
1015 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
1016 (do_dispatch): Move to ix86-tune-sched-bd.c.
1017 (has_dispatch): Move to ix86-tune-sched-bd.c.
1018 * config/i386/t-i386: Add new object files.
1019 * config/i386/x86-tune-costs.h: New file.
1020 * config/i386/x86-tune-sched-atom.c: New file.
1021 * config/i386/x86-tune-sched-bd.c: New file.
1022 * config/i386/x86-tune-sched-core.c: New file.
1023 * config/i386/x86-tune-sched.c: New file.
1025 2017-10-11 Liu Hao <lh_mouse@126.com>
1027 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
1028 the generic version below instead.
1029 (should_colorize): Recognize Windows consoles as terminals
1031 * pretty-print.c [__MINGW32__] (write_all): New function.
1032 [__MINGW32__] (find_esc_head): Likewise.
1033 [__MINGW32__] (find_esc_terminator): Likewise.
1034 [__MINGW32__] (eat_esc_sequence): Likewise.
1035 [__MINGW32__] (mingw_ansi_fputs): New function that handles
1037 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
1040 2017-10-11 Richard Biener <rguenther@suse.de>
1042 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
1043 Properly call analyze_scalar_evolution with the loop of the stmt.
1045 2017-10-11 Richard Biener <rguenther@suse.de>
1047 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
1048 * tree-core.h (tree_base): Add chrec_var union member.
1049 * tree.h (CHREC_VAR): Remove.
1050 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
1051 * tree-chrec.h (build_polynomial_chrec): Adjust.
1052 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
1053 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
1055 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
1057 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
1058 * match.pd: ... here.
1059 ((T) X == (T) Y): Relax condition.
1061 2017-10-11 Bin Cheng <bin.cheng@arm.com>
1063 PR tree-optimization/82472
1064 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
1066 (break_alias_scc_partitions): Update postorder number.
1068 2017-10-11 Martin Liska <mliska@suse.cz>
1071 * opts.c (parse_no_sanitize_attribute): Do not use error_value
1073 * opts.h (parse_no_sanitize_attribute): Remove last argument.
1075 2017-10-11 Martin Liska <mliska@suse.cz>
1077 * print-rtl.c (print_insn): Move declaration of idbuf
1078 to same scope as name.
1080 2017-10-11 Martin Liska <mliska@suse.cz>
1085 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1088 2017-10-11 Martin Liska <mliska@suse.cz>
1091 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1094 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
1096 * config.gcc (default_gnu_indirect_function): Default to yes for
1097 arm*-*-linux* with glibc.
1099 2017-10-11 Richard Biener <rguenther@suse.de>
1101 * tree-scalar-evolution.c (get_scalar_evolution): Handle
1102 default-defs and types we do not want to analyze.
1103 (interpret_loop_phi): Replace unreachable code with an assert.
1104 (compute_scalar_evolution_in_loop): Remove and inline ...
1105 (analyze_scalar_evolution_1): ... here, replacing condition with
1106 what makes the intent clearer. Remove handling of cases
1107 get_scalar_evolution now handles.
1109 2017-10-10 Jim Wilson <wilson@tuliptree.org>
1111 PR rtl-optimization/81434
1112 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
1113 comment for main loop. In sched_group_found if, also add checks for
1114 pass and min_cost_group.
1116 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
1118 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
1119 (rs6000_insn_cost): New function.
1120 * config/rs6000/rs6000.md (cost): New attribute.
1122 2017-10-10 Jakub Jelinek <jakub@redhat.com>
1123 H.J. Lu <hongjiu.lu@intel.com>
1127 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
1128 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
1129 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
1130 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
1131 to OPTION_MASK_ISA_AVX512VL - builtins that have both
1132 OPTION_MASK_ISA_MMX and some other bit set require both
1133 mmx and the ISAs without the mmx bit.
1134 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
1135 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
1136 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
1137 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
1138 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
1139 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
1140 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
1141 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
1142 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
1143 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
1144 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
1145 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
1146 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
1147 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
1148 Add OPTION_MASK_ISA_MMX.
1150 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
1152 * config.gcc (armv7*-*-freebsd*): New target.
1153 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
1155 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
1157 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
1158 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
1161 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
1163 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
1164 a template parameter.
1165 (WIDE_INT_REF_FOR): Update accordingly.
1166 * tree.h (wi::int_traits <const_tree>): Delete.
1167 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
1168 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
1169 (wi::tree_to_wide_ref): New typedef.
1170 (wi::to_wide): New function.
1171 * calls.c (get_size_range): Use wi::to_wide when operating on
1173 * cgraph.c (cgraph_node::create_thunk): Likewise.
1174 * config/i386/i386.c (ix86_data_alignment): Likewise.
1175 (ix86_local_alignment): Likewise.
1176 * dbxout.c (stabstr_O): Likewise.
1177 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
1178 * expr.c (const_vector_from_tree): Likewise.
1179 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
1180 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
1181 (fold_negate_expr_1, int_const_binop_1, const_binop)
1182 (fold_convert_const_int_from_real, optimize_bit_field_compare)
1183 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
1184 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
1185 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
1186 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
1187 (fold_not_const, round_up_loc): Likewise.
1188 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
1189 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
1190 (alloca_call_type): Likewise.
1191 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
1192 * godump.c (go_output_typedef): Likewise.
1193 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
1194 * internal-fn.c (get_min_precision): Likewise.
1195 * ipa-cp.c (ipcp_store_vr_results): Likewise.
1196 * ipa-polymorphic-call.c
1197 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1198 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
1199 (ipa_modify_call_arguments): Likewise.
1200 * match.pd: Likewise.
1201 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
1202 * print-tree.c (print_node_brief, print_node): Likewise.
1203 * stmt.c (expand_case): Likewise.
1204 * stor-layout.c (layout_type): Likewise.
1205 * tree-affine.c (tree_to_aff_combination): Likewise.
1206 * tree-cfg.c (group_case_labels_stmt): Likewise.
1207 * tree-data-ref.c (dr_analyze_indices): Likewise.
1208 (prune_runtime_alias_test_list): Likewise.
1209 * tree-dump.c (dequeue_and_dump): Likewise.
1210 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
1211 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
1212 * tree-pretty-print.c (dump_generic_node): Likewise.
1213 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
1214 (simple_iv_with_niters): Likewise.
1215 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1216 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
1217 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
1218 * tree-ssa-loop-niter.c (split_to_var_and_offset)
1219 (refine_value_range_using_guard, number_of_iterations_ne_max)
1220 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
1221 (get_cst_init_from_scev, record_nonwrapping_iv)
1222 (scev_var_range_cant_overflow): Likewise.
1223 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
1224 * tree-ssa-pre.c (compute_avail): Likewise.
1225 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1226 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
1227 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
1228 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
1229 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
1230 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
1231 (dump_case_nodes, try_switch_expansion): Likewise.
1232 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
1233 (vect_do_peeling): Likewise.
1234 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
1235 * tree-vect-stmts.c (vectorizable_load): Likewise.
1236 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
1237 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
1238 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
1239 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
1240 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
1241 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
1242 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
1243 (evrp_dom_walker::before_dom_children): Likewise.
1244 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
1245 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
1246 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
1247 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
1248 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
1249 (get_range_pos_neg): Likewise.
1250 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
1251 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
1252 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
1253 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
1254 * config/avr/avr.c (avr_fold_builtin): Likewise.
1255 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
1256 * config/msp430/msp430.c (msp430_attr): Likewise.
1257 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
1258 * config/powerpcspe/powerpcspe-c.c
1259 (altivec_resolve_overloaded_builtin): Likewise.
1260 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
1261 (rs6000_expand_ternop_builtin): Likewise.
1262 * config/rs6000/rs6000-c.c
1263 (altivec_resolve_overloaded_builtin): Likewise.
1264 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
1265 (rs6000_expand_ternop_builtin): Likewise.
1266 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
1268 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1270 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
1271 when copying loop nest with only one inner loop.
1273 2017-10-10 Richard Biener <rguenther@suse.de>
1275 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
1276 blocks if SCEV is active.
1277 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
1279 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
1280 (scev_initialize): Assert we are not yet initialized.
1282 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1284 * tree-loop-distribution.c (generate_loops_for_partition): Remove
1285 inner loop's exit stmt by making it always exit the loop, otherwise
1286 we would generate an infinite empty loop.
1288 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1290 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
1291 renaming variables in new preheader if it's deleted.
1293 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1295 * tree-loop-distribution.c (struct partition): Remove unused field
1296 loops of the structure.
1297 (partition_alloc, partition_free): Ditto.
1298 (build_rdg_partition_for_vertex): Ditto.
1300 2017-10-09 Jeff Law <law@redhat.com>
1302 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
1303 return type to match prototype and documentation.
1305 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1307 * config/rs6000/rs6000.c (processor_costs): Move to ...
1308 * config/rs6000/rs6000.h: ... here.
1309 (rs6000_cost): Declare.
1311 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
1313 * except.c (setjmp_fn): New global variable.
1314 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
1315 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
1316 if DONT_USE_BUILTIN_SETJMP is defined.
1318 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1320 * target.def (insn_cost): New hook.
1321 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
1322 * doc/tm.texi: Regenerate.
1323 * rtlanal.c (insn_cost): Use the new hook.
1325 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1327 * combine.c (combine_validate_cost): Compute the new insn_cost,
1328 not just pattern_cost.
1329 (try_combine): Adjust comment.
1331 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1333 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
1335 * combine.c (uid_insn_cost): Adjust comment.
1336 (combine_validate_cost): Adjust comment. Use pattern_cost instead
1338 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
1339 * dse.c (find_shift_sequence): Ditto.
1340 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
1341 (bb_valid_for_noce_process_p): Use pattern_cost.
1342 * rtl.h (insn_rtx_cost): Delete.
1343 (pattern_cost): New prototype.
1344 (insn_cost): New prototype.
1345 * rtlanal.c (insn_rtx_cost): Rename to...
1346 (pattern_cost): ... this.
1349 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
1351 * config/i386/i386.md (*jcc_2): Remove insn pattern.
1352 (*jcc<mode>_0_r_i387): Ditto.
1353 (*jccxf_r_i387): Ditto.
1354 (*jcc<mode>_r_i387): Ditto.
1355 (*jccu<mode>_r_i387): Ditto.
1356 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
1357 (*jcc): Rename from *jcc_1.
1359 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1361 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
1362 deferred rescans after the lvx/stvx recombination pre-pass.
1364 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1366 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
1367 memory operation instruction support.
1368 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
1369 (rs6000-ibm-aix[789]*): Likewise.
1370 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
1371 Document new functions.
1373 2017-10-09 Richard Biener <rguenther@suse.de>
1375 PR tree-optimization/82397
1376 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
1377 equality only for semantically equal trees.
1379 2017-10-09 Richard Biener <rguenther@suse.de>
1381 PR tree-optimization/82449
1382 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
1383 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
1384 allow constant addresses.
1385 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
1388 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1390 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
1393 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1396 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
1399 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1402 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
1404 2017-10-09 Jakub Jelinek <jakub@redhat.com>
1407 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
1408 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
1410 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
1412 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
1413 (WI_BINARY_PREDICATE_RESULT): Likewise.
1414 (wi::binary_traits::operator_result): New type.
1415 (wi::binary_traits::predicate_result): Likewise.
1416 (generic_wide_int::operator~, unary generic_wide_int::operator-)
1417 (generic_wide_int::operator==, generic_wide_int::operator!=)
1418 (generic_wide_int::operator&, generic_wide_int::and_not)
1419 (generic_wide_int::operator|, generic_wide_int::or_not)
1420 (generic_wide_int::operator^, generic_wide_int::operator+
1421 (binary generic_wide_int::operator-, generic_wide_int::operator*):
1423 (operator~, unary operator-, operator==, operator!=, operator&)
1424 (operator|, operator^, operator+, binary operator-, operator*): New
1426 * expr.c (get_inner_reference): Use wi::bit_and_not.
1427 * fold-const.c (fold_binary_loc): Likewise.
1428 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
1429 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
1430 (bit_value_binop): Likewise.
1431 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
1432 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
1433 (extract_range_from_binary_expr_1): Likewise.
1434 (masked_increment): Likewise.
1435 (simplify_bit_ops_using_ranges): Likewise.
1437 2017-10-09 Martin Jambor <mjambor@suse.cz>
1440 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
1441 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
1442 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
1444 (hsa_fixup_mov_insn_type): New function.
1445 (hsa_op_with_type::get_in_type): Use it.
1446 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
1447 immediates in an assert.
1448 (hsa_op_with_type::extend_int_to_32bit): New method.
1449 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
1450 types. Convert to dest type if necessary.
1451 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
1452 (reg_for_gimple_ssa): Pass false as min32int to
1453 hsa_type_for_scalar_tree_type.
1454 (gen_hsa_addr): Fixup type when creating addresable temporary.
1455 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
1456 (gen_hsa_unary_operation): Extend operands and convert to dest type if
1457 necessary. Call hsa_fixup_mov_insn_type.
1458 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
1459 extend operands and convert to dest type if necessary.
1460 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
1461 to dest type if necessary.
1462 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
1463 if conversion nt necessary and size matches.
1464 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
1465 to dest type if necessary.
1466 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
1467 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
1469 (gen_hsa_clrsb): Likewise.
1470 (gen_hsa_ffs): Likewise.
1471 (gen_hsa_divmod): Extend operands and convert to dest type if
1473 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
1475 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1477 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
1478 Remove empty default arguments. Use a brace block as output
1480 (conditional return): Ditto.
1482 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
1483 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
1485 (group_ending_nop): Ditto.
1486 (doloop_end): Ditto.
1487 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
1488 (splitters for those): Ditto.
1490 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1492 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
1493 a conditional jump (and the compare for it) so that pc_rtx is the
1495 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
1496 for the deleted and renamed ctr<mode>_internal[234] patterns.
1497 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
1498 Delete second conditional return pattern.
1499 (ctr<mode>_internal2): Delete this second bdnz pattern.
1500 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
1501 (ctr<mode>_internal4): Delete this second bdz pattern.
1503 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1505 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
1506 (always_initialized_rtx_for_ssa_name_p): New predicate.
1507 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
1508 (finish_out_of_ssa): Free new field of SA.
1509 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
1510 * tree-ssa-coalesce.c: Include tree-ssa.h.
1511 (get_parm_default_def_partitions): Remove extern keyword.
1512 (get_undefined_value_partitions): New function.
1513 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
1514 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
1516 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
1518 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1520 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
1522 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1524 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
1525 for targets that preffer 128bit.
1527 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1529 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
1531 2017-10-08 Olivier Hainque <hainque@adacore.com>
1533 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
1534 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
1536 (thumb_set_return_address): Likewise.
1538 2017-10-08 Olivier Hainque <hainque@adacore.com>
1540 * common/config/arm/arm-common.c (arm_except_unwind_info):
1541 Handle DWARF2_UNWIND_INFO.
1543 2017-10-07 Michael Collison <michael.collison@arm.com>
1545 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
1548 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
1550 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
1551 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
1552 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
1553 defined, force the creation of a new block for a dispatch label.
1555 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
1557 * invoke.texi (Wsuggest-attribute=cold): Document.
1558 * common.opt (Wsuggest-attribute=cold): New
1559 * ipa-pure-const.c (warn_function_cold): New function.
1560 * predict.c (compute_function_frequency): Use it.
1561 * predict.h (warn_function_cold): Declare.
1563 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
1565 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
1568 2017-10-06 Martin Liska <mliska@suse.cz>
1570 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
1571 keyword for member functions.
1572 (struct sanopt_tree_couple): New struct.
1573 (struct sanopt_tree_couple_hash): New function.
1574 (struct sanopt_ctx): Add new hash_map.
1575 (has_dominating_ubsan_ptr_check): New function.
1576 (record_ubsan_ptr_check_stmt): Likewise.
1577 (maybe_optimize_ubsan_ptr_ifn): Likewise.
1578 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
1579 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
1581 2017-10-06 Sudakshina Das <sudi.das@arm.com>
1584 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
1585 aarch64_simd_valid_immediate on CONST_VECTORs.
1586 (aarch64_reg_or_bic_imm): Likewise.
1588 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
1590 PR rtl-optimization/82396
1591 * haifa-sched.c (ready_sort_real): Disable qsort checking.
1593 2017-10-06 Sebastian Pop <sebpop@gmail.com>
1595 * graphite-dependences.c (scop_get_reads): Move code to...
1596 (scop_get_must_writes): Move code to...
1597 (scop_get_may_writes): Move code to...
1598 (scop_get_reads_and_writes): ... here.
1599 (scop_get_dependences): Call scop_get_reads_and_writes.
1601 2017-10-06 Jakub Jelinek <jakub@redhat.com>
1603 PR tree-optimization/82434
1604 * fold-const.h (can_native_encode_type_p,
1605 can_native_encode_string_p): Remove.
1606 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
1607 don't encode anything, just return what would be otherwise returned.
1608 (native_encode_fixed, native_encode_complex, native_encode_vector):
1610 (native_encode_string): Likewise. Inline by hand
1611 can_native_encode_string_p.
1612 (can_native_encode_type_p): Remove.
1613 (can_native_encode_string_p): Remove.
1614 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
1615 STRING_CSTs using can_native_encode_string_p, test all
1616 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
1617 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
1618 argument from native_encode_expr.
1619 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
1620 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
1623 2017-10-06 Richard Biener <rguenther@suse.de>
1625 PR tree-optimization/82397
1626 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
1627 operand_equal_p but rely on data_ref_compare_tree for detecting
1629 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
1630 to match up with dr_group_sort_cmp.
1632 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1635 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
1637 * config/s390/s390-builtin-types.def: Regenerate.
1639 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1642 * config/s390/s390-builtin-types.def: Regenerate.
1643 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
1644 Change flag from B_VXE to B_VX.
1645 (s390_vec_min_dbl): Remove B_VXE flag.
1647 2017-10-06 Richard Biener <rguenther@suse.de>
1649 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
1650 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
1651 translate_isl_ast_to_gimple::is_valid_rename,
1652 translate_isl_ast_to_gimple::get_rename,
1653 translate_isl_ast_to_gimple::get_def_bb_for_const,
1654 translate_isl_ast_to_gimple::get_new_name,
1655 translate_isl_ast_to_gimple::collect_all_ssa_names,
1656 translate_isl_ast_to_gimple::copy_loop_phi_args,
1657 translate_isl_ast_to_gimple::collect_all_ssa_names,
1658 translate_isl_ast_to_gimple::copy_loop_phi_args,
1659 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
1660 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
1661 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
1662 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
1663 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
1664 translate_isl_ast_to_gimple::copy_cond_phi_args,
1665 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
1666 translate_isl_ast_to_gimple::edge_for_new_close_phis,
1667 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
1668 translate_isl_ast_to_gimple::rename_uses,
1669 translate_isl_ast_to_gimple::rename_all_uses): Remove.
1670 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
1671 (set_rename_for_each_def): Likewise.
1672 (graphite_copy_stmts_from_block): Handle debug stmt resetting
1673 here. Handle rewriting SCEV analyzable uses here.
1674 (copy_bb_and_scalar_dependences): Generate code for PHI
1676 (graphite_regenerate_ast_isl): Adjust.
1677 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
1678 (add_write, add_read): New functions.
1679 (build_cross_bb_scalars_def): Use it and simplify.
1680 (build_cross_bb_scalars_use): Likewise.
1681 (graphite_find_cross_bb_scalar_vars): Inline into...
1682 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
1683 simulating out-of-SSA. Compute liveout and add dependencies.
1684 (build_scops): Force an empty entry block.
1685 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
1687 (sese_build_liveouts): Declare.
1688 (sese_trivially_empty_bb_p): Likewise.
1689 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
1690 compute liveout and debug_liveout.
1691 (sese_bad_liveouts_use): Remove.
1692 (sese_reset_debug_liveouts_bb): Likewise.
1693 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
1694 (sese_build_liveouts): Build liveout and debug_liveout and store
1696 (new_sese_info): Adjust.
1697 (free_sese_info): Likewise.
1698 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
1699 do not build liveout here.
1700 (move_sese_in_condition): Adjust region entry.
1701 (scev_analyzable_p): Match up with chrec_apply requirements.
1702 (sese_trivially_empty_bb_p): New.
1703 * tree-into-ssa.c (get_reaching_def): Properly support generating
1704 default-defs for incremental rewrite of anonymous names.
1706 2017-10-06 Richard Biener <rguenther@suse.de>
1708 * graphite-sese-to-poly.c (extract_affine): For casts increasing
1709 precision do not perform modulo reduction.
1711 2017-10-06 Richard Biener <rguenther@suse.de>
1713 PR tree-optimization/82436
1714 * tree-vect-slp.c (vect_supported_load_permutation_p): More
1715 conservatively choose the vectorization factor when checking
1716 whether we can perform the required load permutation.
1717 (vect_transform_slp_perm_load): Assert when we may not fail.
1719 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1721 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
1722 message for incompatible -msdata=* and -mcall-* options.
1724 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1726 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
1727 rate for post-reload scheduling.
1729 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1731 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
1733 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1735 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
1736 to improve monte carlo in scimark.
1738 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1740 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
1741 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
1742 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
1743 pentium4_cost, nocona_cost): Set reassociation width to 1.
1744 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
1745 width to 2 for fp operations and 1 otherwise.
1746 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
1748 (atom_cost): Set reassociation width to 2.
1749 (slm_cost, generic_cost): Set fp reassociation width
1750 to 2 and 1 otherwise.
1751 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
1752 (core_cost): Set fp reassociation width to 4 and vector to 2.
1753 (ix86_reassociation_width): Rewrite using cost table; special case
1754 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
1755 and TARGET_AVX128_OPTIMAL.
1756 * config/i386/i386.h (processor_costs): Add
1757 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
1758 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
1759 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
1760 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
1761 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
1762 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
1764 2017-10-05 Nathan Sidwell <nathan@acm.org>
1766 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
1768 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1770 * config/arm/arm.c (arm_test_fpu_data): New.
1771 (arm_run_selftests): Call arm_test_fpu_data.
1773 2017-10-04 Nathan Sidwell <nathan@acm.org>
1775 * toplev.c (toplev::main): Remove excess parens on pretty_printer
1777 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
1779 2017-10-04 Sudakshina Das <sudi.das@arm.com>
1781 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
1782 check type for aarch64_simd_valid_immediate.
1783 (aarch64_output_simd_mov_immediate): Update prototype.
1784 (aarch64_simd_valid_immediate): Update prototype.
1785 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
1786 support for ORR-immediate.
1787 (and<mode>3): modified pattern to add support for BIC-immediate.
1788 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
1789 now checks for valid immediate for BIC and ORR based on new enum
1791 (aarch64_output_simd_mov_immediate): Function now used to output
1792 BIC/ORR imm as well based on new enum argument.
1793 * config/aarch64/constraints.md (Do): New vector immediate constraint.
1795 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
1796 (aarch64_reg_or_bic_imm): Likewise.
1798 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1800 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
1801 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
1803 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1807 PR rtl-optimization/82396
1808 * haifa-sched.c (autopref_multipass_init): Simplify
1810 (autopref_rank_data): Simplify sort order.
1811 * sched-int.h (autopref_multipass_data_): Remove
1812 multi_mem_insn_p, min_offset and max_offset.
1814 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1816 * doc/sourcebuild.texi: Document vect_peeling_profitable.
1818 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1820 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
1823 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1825 * doc/sourcebuild.texi: Document vect_long_mult.
1827 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
1829 PR tree-optimization/82413
1830 * fold-const.c (build_range_check): Use widest_int when comparing
1831 the maximum ETYPE value with HIGH.
1833 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1835 PR rtl-optimization/82396
1836 * haifa-sched.c (autopref_multipass_init): Simplify
1838 (autopref_rank_data): Simplify sort order.
1839 * sched-int.h (autopref_multipass_data_): Remove
1840 multi_mem_insn_p, min_offset and max_offset.
1842 2017-10-04 Jakub Jelinek <jakub@redhat.com>
1844 PR tree-optimization/82381
1845 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
1846 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
1849 PR tree-optimization/82374
1850 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
1851 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
1852 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
1853 current_function_decl to the new decl.
1855 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
1857 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
1858 helper macro for IEEE float128 hardware built-in functions.
1859 (SQRTF128_ODD): Add built-in functions with the round-to-odd
1861 (TRUNCF128_ODD): Likewise.
1862 (ADDF128_ODD): Likewise.
1863 (SUBF128_ODD): Likewise.
1864 (MULF128_ODD): Likewise.
1865 (DIVF128_ODD): Likewise.
1866 (FMAF128_ODD): Likewise.
1867 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
1868 UNSPEC_TRUNC_ROUND_TO_ODD.
1869 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
1870 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
1871 floating point round to odd instructions.
1872 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
1873 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
1874 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
1875 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
1876 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
1877 (trunc<mode>sf2_hw): Change the truncate with round to odd
1878 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
1879 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
1880 to odd hardware instructions.
1881 (sub<mode>3_odd): Likewise.
1882 (mul<mode>3_odd): Likewise.
1883 (div<mode>3_odd): Likewise.
1884 (sqrt<mode>2_odd): Likewise.
1885 (fma<mode>4_odd): Likewise.
1886 (fms<mode>4_odd): Likewise.
1887 (nfma<mode>4_odd): Likewise.
1888 (nfms<mode>4_odd): Likewise.
1889 (trunc<mode>df2_odd): Change the truncate with round to odd
1890 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
1892 * doc/extend.texi (PowerPC built-in functions): Update documentation
1893 for existing IEEE float128-bit built-in functions. Add built-in
1894 functions that generate the IEEE 128-bit floating point round to
1897 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
1899 PR rtl-optimization/77729
1900 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
1901 to (X&(C1&~C2))|C2 transformations.
1903 2017-10-03 Martin Jambor <mjambor@suse.cz>
1905 PR tree-optimization/82363
1906 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
1907 mismatch, mark lacc written regardless of racc.
1909 2017-10-03 Jakub Jelinek <jakub@redhat.com>
1911 PR tree-optimization/82381
1912 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
1913 stmt_to_insert nor wheather SSA_NAMEs are default defs.
1914 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
1915 fallthrough into reassoc_stmt_dominates_stmt_p.
1918 * combine.c (combine_instructions): Don't combine in unreachable
1921 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1924 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
1925 function to not use the have_cpu variable. Do not set cpu_index,
1926 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
1928 (rs6000_valid_attribute_p): Remove duplicate initializations of
1929 old_optimize and func_optimize.
1930 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
1931 (rs6000_activate_target_options): Make global.
1932 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
1935 2017-10-02 Jakub Jelinek <jakub@redhat.com>
1937 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
1938 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
1939 Set *poffset to 0 and *psize and *pmax_size to -1 if
1940 *poffset + *psize overflows in HOST_WIDE_INT.
1942 PR tree-optimization/82387
1943 PR tree-optimization/82388
1944 PR tree-optimization/82389
1945 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
1946 instead of live_bytes non-NULL.
1948 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
1951 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
1954 2017-10-02 Richard Biener <rguenther@suse.de>
1956 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
1957 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
1958 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
1960 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1962 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
1963 requested precision matches the type's.
1964 * calls.c (alloc_max_size): Calculate the new candidate size as
1965 a widest_int and use wi::to_widest when comparing it with the
1966 current candidate size.
1967 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
1968 zero rather than integer_zero_node.
1969 * match.pd: Check for a no-op conversion before using wi::add
1970 rather than after. Use tree_to_uhwi when summing small shift
1971 counts into an unsigned int.
1973 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1974 Alan Hayward <alan.hayward@arm.com>
1975 David Sherwood <david.sherwood@arm.com>
1978 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
1979 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
1980 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
1981 POINTER_AND_FP_REGS.
1983 2017-10-02 Richard Biener <rguenther@suse.de>
1985 PR tree-optimization/82355
1986 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1987 a mapping for the enclosing loop but avoid generating one for
1989 (copy_bb_and_scalar_dependences): Remove premature codegen
1990 error on PHIs in blocks duplicated into multiple places.
1991 * graphite-scop-detection.c
1992 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1993 in the region use it as loop and nest to analyze the DR in.
1994 (try_generate_gimple_bb): Likewise.
1995 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1996 (add_loop_constraints): For blocks in a loop not in the region
1997 create a dimension with a single iteration.
1998 * sese.h (gbb_loop_at_index): Remove assert.
2000 2017-10-01 Kevin Buettner <kevinb@redhat.com>
2002 * omp-expand.c (adjust_context_scope): New function.
2003 (expand_parallel_call): Call adjust_context_scope.
2005 2017-10-01 Jeff Law <law@redhat.com>
2007 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
2008 dom_opt_dom_walker class with direct access to private members.
2009 Add comments. Call test_for_singularity.
2010 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
2011 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
2012 m_dummy_cond anymore.
2013 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
2015 (pass_dominator:execute): Build the dummy_cond here and pass it
2016 to the dom_opt_dom_walker ctor.
2017 (test_for_singularity): New function.
2019 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
2020 Maya Rashish <coypu@sdf.org>
2022 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
2023 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
2024 (alpha*-*-netbsd*) Use nbsd_tm_file.
2025 (arm*-*-netbsdelf*) Likewise.
2026 (i[34567]86-*-netbsdelf*) Likewise.
2027 (x86_64-*-netbsd*) Likewise.
2028 (mips*-*-netbsd*) Likewise.
2029 (powerpc-*-netbsd*) Likewise.
2030 (sh*-*-netbsd*) Likewise.
2031 (sparc-*-netbsdelf*) Likewise.
2032 (sparc64-*-netbsd*) Likewise.
2033 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
2035 (vax-*-netbsdelf*) Likewise.
2036 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
2037 (UINT_FAST8_TYPE) Likewise.
2038 (INT_FAST16_TYPE) Check CHAR_FAST16.
2039 (UINT_FAST16_TYPE) Likewise.
2041 2017-09-30 Jakub Jelinek <jakub@redhat.com>
2044 * config/i386/i386.md
2045 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
2046 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
2047 *divmodsi4_zext_2): New define_insn_and_split.
2048 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
2049 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
2050 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
2051 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
2052 New define_insn_and_split.
2053 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
2054 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
2055 operands[1] having DImode when mode is SImode.
2057 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
2058 always SImode for DIV and MOD in REG_EQUAL notes.
2060 2017-09-29 Yury Gribov <tetra2005@gmail.com>
2063 * match.pd: Fix handling of NaNs in pattern.
2065 2017-09-29 Jeff Law <law@redhat.com>
2067 * sbitmap.c (bitmap_bit_in_range_p): New function.
2068 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
2069 * tree-ssa-dse.c (live_bytes_read): New function.
2070 (dse_classify_store): Ignore reads of dead bytes.
2072 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
2073 typos and whitespace errors.
2074 * config/i386/predicates.md (address_no_seg_operand): Likewise.
2075 * config/s390/s390.c (s390_emit_prologue): Likewise.
2077 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2080 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
2081 with a symbol for LRA.
2083 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2085 PR rtl-optimization/82338
2086 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
2088 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
2090 * genmodes.c (calc_wider_mode): Suppress qsort macro.
2091 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
2092 (qsort_chk): Declare.
2093 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
2094 (qsort_chk): New function.
2096 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2098 PR tree-optimization/82337
2099 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
2100 phi definition if the PHI result appears in an abnormal PHI.
2101 (find_basis_for_base_expr): Don't record a basis if the LHS of the
2102 basis appears in an abnormal PHI.
2104 2017-09-29 Richard Biener <rguenther@suse.de>
2106 * graphite-isl-ast-to-gimple.c
2107 (translate_isl_ast_to_gimple::set_codegen_error): New function.
2108 (binary_op_to_tree): Use it.
2109 (get_rename_from_scev): Likewise.
2110 (copy_loop_phi_nodes): Likewise.
2111 (copy_bb_and_scalar_dependences): Likewise.
2112 (translate_pending_phi_nodes): Likewise.
2114 2017-09-29 Jakub Jelinek <jakub@redhat.com>
2117 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
2118 for movabsq $(i32 << shift), r64.
2120 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
2122 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
2123 index when encoding %esp as %rsp to avoid 0x67 prefix.
2125 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
2127 * config/i386/i386.md (*movsf_internal, *movdf_internal):
2128 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
2130 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
2132 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
2133 Extensions with more than 16 double VFP registers.
2134 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
2135 to_clear_mask and all code related to it. Replace the remaining
2136 entry by a sbitmap and adapt code accordingly.
2138 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
2140 * brig-builtins.def: Change pure attributes to const.
2142 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2144 * config.gcc (default_gnu_indirect_function): Default to yes for
2145 sparc*-*-linux* with glibc.
2147 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2149 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
2150 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
2151 when creating .init_array and .fini_array sections with priority
2154 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
2157 * config/aarch64/aarch64.c
2158 (aarch64_builtin_support_vector_misalignment): Always return false
2159 when misalignment is unknown.
2161 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
2163 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
2164 this function to return false if the definition used by the swap
2165 instruction is artificial, or if the memory address from which the
2166 constant value is loaded is not represented by a base address held
2167 in a register or if the base address register is a frame or stack
2168 pointer. Additionally, return false if the base address of the
2169 loaded constant is a SYMBOL_REF but is not considered to be a
2171 (replace_swapped_load_constant): New function.
2172 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
2173 loaded constant vector with a load of a swapped constant vector.
2175 2017-09-27 Carl Love <cel@us.ibm.com>
2177 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
2178 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
2179 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
2182 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
2184 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
2185 first, always call autopref_rank_data otherwise.
2187 2017-09-27 Richard Biener <rguenther@suse.de>
2189 * graphite-scop-detection.c (find_scop_parameters): Move
2190 loop bound handling ...
2191 (gather_bbs::before_dom_children): ... here, avoiding the need
2192 to build scop_info->loop_nest.
2193 (record_loop_in_sese): Remove.
2194 * sese.h (sese_info_t::loop_nest): Remove.
2195 * sese.c (new_sese_info): Do not allocate loop_nest.
2196 (free_sese_info): Do not free loop_nest.
2198 2017-09-27 Jakub Jelinek <jakub@redhat.com>
2201 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
2202 lhs from calls if the lhs has addressable type.
2204 2017-09-27 Richard Biener <rguenther@suse.de>
2206 * graphite.h (scop::max_alias_set): New member.
2207 * graphite-scop-detection.c: Remove references to non-existing
2208 --param in comments.
2209 (build_alias_sets): Record the maximum alias set used for drs.
2210 (build_scops): Support zero as unlimited for
2211 --param graphite-max-arrays-per-scop.
2212 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
2214 (build_poly_sr_1): ... here. Compute alias set based on the
2215 maximum alias set used for drs rather than
2216 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
2218 2017-09-27 Richard Biener <rguenther@suse.de>
2220 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
2221 --param loop-block-tile-size=0 to disable tiling.
2223 2017-09-27 Richard Biener <rguenther@suse.de>
2225 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
2226 (graphite-max-nb-scop-params): Document special value zero.
2227 * domwalk.h (dom_walker::STOP): New symbolical constant.
2228 (dom_walker::dom_walker): Add optional parameter for bb to
2230 (dom_walker::~dom_walker): Declare.
2231 (dom_walker::before_dom_children): Document STOP return value.
2232 (dom_walker::m_user_bb_to_rpo): New member.
2233 (dom_walker::m_bb_to_rpo): Likewise.
2234 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
2235 mapping here if not provided by the user.
2236 (dom_walker::~dom_walker): Free bb to RPO mapping if not
2237 provided by the user.
2238 (dom_walker::STOP): Define.
2239 (dom_walker::walk): Do not compute bb to RPO mapping here.
2240 Support STOP return value from before_dom_children to stop
2242 * graphite-optimize-isl.c (optimize_isl): If the schedule
2243 is the same still generate code if -fgraphite-identity
2244 or -floop-parallelize-all are given.
2245 * graphite-scop-detection.c: Include cfganal.h.
2246 (gather_bbs::gather_bbs): Get and pass through bb to RPO
2248 (gather_bbs::before_dom_children): Return STOP for BBs
2250 (build_scops): Compute bb to RPO mapping and pass it to
2251 the domwalk. Treat --param graphite-max-nb-scop-params=0
2252 as not limiting the number of params.
2253 * graphite.c (graphite_initialize): Remove limit on the
2254 number of basic-blocks in a function.
2255 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
2256 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
2257 default value of 10.
2259 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2261 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
2262 Adjust code to eliminate needing to do the shift right 32-bits
2263 operation after XSCVDPSPN.
2265 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2267 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
2268 ((X / Y) != 0 -> X >= Y): Likewise.
2270 2017-09-26 Carl Love <cel@us.ibm.com>
2272 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
2273 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
2274 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
2275 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
2276 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
2277 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
2278 definitions and overloading.
2279 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
2280 statement for P9V_BUILTIN_XST_LEN_R.
2281 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
2282 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
2283 define_expand and define_insn for the instructions and builtins.
2284 * doc/extend.texi: Update the built-in documentation file for the new
2286 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
2287 define_insn for the instructions
2289 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
2292 * gcc/config/netbsd-protos.h: New file.
2293 * gcc/config/netbsd.c: New file.
2294 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
2295 * gcc/config/t-netbsd: New file.
2296 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
2297 (tmake_file) Add t-netbsd.
2298 (extra_objs) Add netbsd.o.
2300 2017-09-26 Janus Weil <janus@gcc.gnu.org>
2304 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
2306 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2308 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
2309 sign extension from a vector register to a GPR by doing a 32-bit
2310 direct move and then an EXTSW.
2311 (extendsi<mode>2 splitter): Likewise.
2312 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
2313 right or vector extract after doing XSCVDPSPN. Use
2314 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
2316 (movdi_from_sf_zero_ext): Likewise.
2317 (reload_gpr_from_vsxsf): Likewise.
2318 (p8_mfvsrd_4_disf): Delete, no longer used.
2319 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
2320 then needing to move the SFmode to a GPR to use the XSCVDPSP
2321 instruction instead of FRSP and XSCVDPSPN.
2322 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
2323 it is adjacent to the other XSCVSPDP insns.
2324 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
2325 SFmode to be in traditional Altivec registers.
2326 (vsx_xscvdpspn): Eliminate useless alternative constraint.
2327 (vsx_xscvspdpn): Likewise.
2328 (vsx_xscvspdpn_scalar): Likewise.
2330 2017-09-26 Martin Jambor <mjambor@suse.cz>
2332 * tree-sra.c (compare_access_positions): Put integral types first,
2333 stabilize sorting of integral types, remove conditions putting
2334 non-full-precision integers last.
2335 (sort_and_splice_var_accesses): Disable scalarization if a
2336 non-integert would be represented by a non-full-precision integer.
2338 2017-09-26 Joseph Myers <joseph@codesourcery.com>
2340 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
2341 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
2342 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
2343 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
2344 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
2345 conditionals inside the function instead of around it. Call
2346 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
2347 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
2349 2017-09-26 Richard Biener <rguenther@suse.de>
2351 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
2353 (scop_detection::build_scop_breadth): ... this. Removed.
2354 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
2355 (scop_detection::harmful_stmt_in_bb): Likewise.
2356 (scop_detection::graphite_can_represent_stmt): Likewise.
2357 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
2358 (scop_detection::can_represent_loop): Remove recursion, fold in ...
2359 (scop_detection::can_represent_loop_1): ... this. Removed.
2360 (scop_detection::harmful_loop_in_region): Simplify after inlining
2361 the above and remove more quadraticness.
2362 (build_scops): Adjust.
2363 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
2366 2017-09-26 Jakub Jelinek <jakub@redhat.com>
2369 * config/i386/i386.c (ix86_print_operand_address_as): Only test
2370 REGNO (base) == SP_REG if base is a REG.
2373 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
2374 if it is different SSA_NAME.
2375 (optimize_range_tests_cmp_bitwise): New function.
2376 (optimize_range_tests): Call it.
2378 2017-09-26 Richard Biener <rguenther@suse.de>
2380 PR tree-optimization/82321
2381 * graphite.c (canonicalize_loop_closed_ssa): Properly check
2382 for the def being inside the loop.
2384 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2386 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
2388 * config/s390/s390-builtins.def: Fix constraint on op4.
2390 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2392 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
2393 independent expanders.
2394 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
2395 ("vec_ordered", "vec_unordered"): New expanders.
2397 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2399 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
2402 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2404 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
2405 vec_unpacks_lo_v16qi.
2406 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
2408 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2410 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
2411 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
2412 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
2414 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2416 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
2418 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
2420 ("*vec_slb<mode>"): New insn pattern.
2421 ("vec_shr_<mode>"): New expander.
2422 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
2423 and force the shift count operand to V16QImode.
2424 ("vec_srb<mode>"): Set shift count mode to V16QI.
2426 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2428 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
2429 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
2430 ("vec_widen_smult_hi_<mode>"): New expander definitions.
2432 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
2435 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
2437 2017-09-26 Richard Biener <rguenther@suse.de>
2439 PR tree-optimization/82320
2440 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
2443 2017-09-25 Jeff Law <law@redhat.com>
2445 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
2446 prototype for new argument.
2447 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
2448 mostly extracted from rs6000_emit_allocate_stack.
2449 (rs6000_emit_probe_stack_range_stack_clash): New function.
2450 (rs6000_emit_allocate_stack): Call
2451 rs6000_emit_probe_stack_range_stack_clash as needed.
2452 (rs6000_emit_probe_stack_range): Add additional argument
2453 to call to gen_probe_stack_range{si,di}.
2454 (output_probe_stack_range): New.
2455 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
2456 (output_probe_stack_range_stack_clash): New.
2457 (rs6000_emit_prologue): Emit notes into dump file as requested.
2458 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
2459 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
2460 Add additional operand and pass it to output_probe_stack_range.
2462 2017-09-25 Bin Cheng <bin.cheng@arm.com>
2464 PR tree-optimization/82163
2465 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
2466 (checking_verify_loop_closed_ssa): New parameter.
2467 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
2468 (check_loop_closed_ssa_stmt): Delete.
2469 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
2470 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
2471 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
2474 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
2476 * brig-builtins.def: Treat HSAIL barrier builtins as
2477 setjmp/longjump style functions.
2479 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2481 * target.def (constant_alignment): New hook.
2482 * defaults.h (CONSTANT_ALIGNMENT): Delete.
2483 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
2484 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
2485 * doc/tm.texi: Regenerate.
2486 * targhooks.h (default_constant_alignment): Declare.
2487 (constant_alignment_word_strings): Likewise.
2488 * targhooks.c (default_constant_alignment): New function.
2489 (constant_alignment_word_strings): Likewise.
2490 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
2491 instead of CONSTANT_ALIGNMENT.
2492 * varasm.c (align_variable, get_variable_align, build_constant_desc)
2493 (force_const_mem): Likewise.
2494 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
2495 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
2496 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
2497 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2498 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
2500 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
2501 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2502 constant_alignment_word_strings.
2503 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
2504 (CONSTANT_ALIGNMENT): Likewise.
2505 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2506 (arm_constant_alignment): New function.
2507 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
2508 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2509 constant_alignment_word_strings.
2510 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
2511 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2512 constant_alignment_word_strings.
2513 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
2514 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2515 (cris_constant_alignment): New function.
2516 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
2517 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2518 (epiphany_constant_alignment): New function.
2519 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
2520 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2521 constant_alignment_word_strings.
2522 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
2523 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2524 constant_alignment_word_strings.
2525 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
2526 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2527 constant_alignment_word_strings.
2528 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
2529 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
2530 * config/i386/i386.c (ix86_constant_alignment): Make static.
2531 Use the same interface as the target hook.
2532 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2533 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
2534 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2535 constant_alignment_word_strings.
2536 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
2537 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
2538 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2539 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
2540 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2541 constant_alignment_word_strings.
2542 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
2543 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2544 constant_alignment_word_strings.
2545 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
2546 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2547 constant_alignment_word_strings.
2548 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
2549 * config/microblaze/microblaze.c (microblaze_constant_alignment):
2551 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2552 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
2553 * config/mips/mips.c (mips_constant_alignment): New function.
2554 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2555 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
2556 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
2557 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2558 (mmix_constant_alignment): Make static. Use the same interface
2560 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
2561 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2562 constant_alignment_word_strings.
2563 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
2564 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2565 constant_alignment_word_strings.
2566 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
2567 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2568 constant_alignment_word_strings.
2569 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
2570 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2571 (rs6000_constant_alignment): New function.
2572 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
2573 * config/riscv/riscv.c (riscv_constant_alignment): New function.
2574 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2575 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
2576 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2577 (rs6000_constant_alignment): New function.
2578 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
2579 * config/s390/s390.c (s390_constant_alignment): New function.
2580 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2581 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
2582 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2583 constant_alignment_word_strings.
2584 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
2585 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2586 (sparc_constant_alignment): New function.
2587 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
2588 * config/spu/spu.c (spu_constant_alignment): New function.
2589 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2590 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
2591 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2592 constant_alignment_word_strings.
2593 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
2594 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2595 constant_alignment_word_strings.
2596 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
2597 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2598 constant_alignment_word_strings.
2599 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
2600 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2601 (visium_constant_alignment): New function.
2602 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
2603 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2604 (xtensa_constant_alignment): New function.
2605 * system.h (CONSTANT_ALIGNMENT): Poison.
2607 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
2609 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2610 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
2611 (rs6000_builtin_valid_without_lhs): New helper function.
2612 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2613 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
2615 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2617 * target.h (vec_perm_indices): Use unsigned short rather than
2619 (auto_vec_perm_indices): Likewise.
2620 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
2621 Use unsigned int rather than unsigned char.
2622 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
2624 2017-09-25 Richard Biener <rguenther@suse.de>
2626 * cfgloop.h (sort_sibling_loops): Declare.
2627 * cfgloop.c (sort_sibling_loops_cmp): New helper.
2628 (sort_sibling_loops): New function sorting the sibling loop list
2630 * graphite.c (graphite_transform_loops): Sort sibling loops.
2632 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
2634 * target.def (vec_perm_const_ok): Change sel parameter to
2636 * optabs-query.c (can_vec_perm_p): Update accordingly.
2637 * doc/tm.texi: Regenerate.
2638 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
2639 auto_vec_perm_indices and remove separate nelt field.
2640 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
2641 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
2642 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
2643 (aarch64_expand_vec_perm_const): Update accordingly.
2644 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
2645 to vec_perm_indices.
2646 * config/arm/arm.c (expand_vec_perm_d): Change perm to
2647 auto_vec_perm_indices and remove separate nelt field.
2648 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
2649 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
2650 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
2652 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
2653 to vec_perm_indices.
2654 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
2655 sel to vec_perm_indices.
2656 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
2657 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
2658 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
2660 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
2663 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
2666 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
2667 on the FUNCTION_DECL function context if it has a DIE that is a
2670 2017-09-25 Richard Biener <rguenther@suse.de>
2672 PR tree-optimization/82285
2673 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
2676 2017-09-25 Tom de Vries <tom@codesourcery.com>
2680 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
2683 2017-09-25 Richard Biener <rguenther@suse.de>
2685 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
2686 ISL errors other than isl_error_quota happen. Dump if the
2687 schedule is the same.
2688 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
2689 errors instead of aborting inside ISL.
2691 2017-09-25 Iain Sandoe <iain@codesourcery.com>
2694 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
2695 of libgcc_eh for m64.
2696 * config/i386/darwin64.h: Likewise.
2698 2017-09-25 Richard Biener <rguenther@suse.de>
2701 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
2702 attribute for incomplete types nor twice for complete ones.
2704 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
2707 * config/i386/i386.c (ix86_print_operand_address_as): Encode
2708 %esp as %rsp to avoid 0x67 prefix if there is no index or base
2711 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
2714 * config/i386/i386.opt (mprefer-avx256): Use
2715 ix86_target_flags variable.
2716 * config/i386/i386.c (ix86_target_string): Move
2717 -mprefer-avx256 to flag2_opts.
2719 2017-09-22 Jakub Jelinek <jakub@redhat.com>
2722 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
2723 and x != -1 | y != -1 into (x & y) != -1.
2725 2017-09-22 Steve Ellcey <sellcey@cavium.com>
2727 * config.gcc: Add new case statement to set
2728 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
2729 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
2730 s390x-*-linux* case statements. Added aarch64 to the list of
2731 supported architectures.
2733 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2735 PR tree-optimization/82289
2736 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
2737 STMT_VINFO_RELEVANT_P.
2739 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2740 Alan Hayward <alan.hayward@arm.com>
2741 David Sherwood <david.sherwood@arm.com>
2743 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
2744 for VR_RANGE only; don't allow VR_ANTI_RANGE.
2745 (extract_range_from_binary_expr_1): Don't call
2746 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
2748 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2749 Alan Hayward <alan.hayward@arm.com>
2750 David Sherwood <david.sherwood@arm.com>
2752 * target.def (preferred_vector_alignment): New hook.
2753 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
2755 * doc/tm.texi: Regenerate.
2756 * targhooks.h (default_preferred_vector_alignment): Declare.
2757 * targhooks.c (default_preferred_vector_alignment): New function.
2758 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
2760 (DR_TARGET_ALIGNMENT): New macro.
2761 (aligned_access_p): Update commentary.
2762 (vect_known_alignment_in_bytes): New function.
2763 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
2765 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
2766 Calculate the misalignment based on the target alignment rather than
2768 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
2769 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
2770 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
2771 the target alignment, rather than masking the element misalignment
2772 with the number of elements in a vector. Also use the target
2773 alignment when calculating the maximum number of peels.
2774 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
2775 instead of TYPE_ALIGN_UNIT.
2776 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
2777 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
2778 (vect_create_addr_base_for_vector_ref): Update call accordingly.
2779 (vect_create_data_ref_ptr): Likewise.
2780 (vect_setup_realignment): Realign by ANDing with
2781 -DR_TARGET_MISALIGNMENT.
2782 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
2783 the number of peels based on DR_TARGET_ALIGNMENT.
2784 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
2785 with the guaranteed alignment boundary when deciding whether
2787 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
2788 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
2789 (ensure_base_align): Remove stmt_info parameter. Get the
2790 target base alignment from DR_TARGET_ALIGNMENT.
2791 (vectorizable_store): Update call accordingly. Interpret
2792 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
2794 (vectorizable_load): Likewise.
2796 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2797 Alan Hayward <alan.hayward@arm.com>
2798 David Sherwood <david.sherwood@arm.com>
2800 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
2801 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
2802 (vect_enhance_data_refs_alignment): Likewise.
2804 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2806 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2807 error. Only quit immediately if parsing is complete.
2808 (BEGIN): Initialize fatal_err and parse_done.
2809 (begin fpu, end fpu): Check number of arguments.
2810 (begin arch, end arch): Likewise.
2811 (begin cpu, end cpu): Likewise.
2812 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2813 (optalias): Likewise.
2815 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2817 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2818 * config/arm/arm-isa.h: Delete. Move definitions to ...
2819 * arm-cpus.in: ... here. Use new feature and fgroup values.
2820 * config/arm/arm.c (arm_option_override): Use lower case for feature
2822 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2823 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2824 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2826 (print_isa_bits_for): New function.
2827 (gen_isa): New function.
2828 (gen_comm_data): Use print_isa_bits_for.
2829 (define feature): New keyword.
2830 (define fgroup): New keyword.
2831 * config/arm/t-arm (TM_H): Remove.
2832 (GTM_H): Add arm-isa.h.
2833 (arm-isa.h): Add rule to generate file.
2834 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2835 case for feature bit names.
2837 2017-09-22 Richard Biener <rguenther@suse.de>
2839 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
2841 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
2842 print of no dependency loops ...
2843 * graphite.c (graphite_transform_loops): ... here.
2844 (canonicalize_loop_closed_ssa_form): Work from inner to outer
2846 (same_close_phi_node, remove_duplicate_close_phi,
2847 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
2848 (canonicalize_loop_closed_ssa): ... here and simplify.
2849 * graphite-optimize-isl.c: Include tree-vectorizer.h.
2850 (optimize_isl): Use dump_printf_loc to tell when we stopped
2851 optimizing because of an ISL timeout.
2853 2017-09-22 Richard Biener <rguenther@suse.de>
2855 PR tree-optimization/82291
2856 * tree-if-conv.c (predicate_mem_writes): Make sure to
2857 remove writes in blocks predicated with false.
2859 2017-09-22 Richard Biener <rguenther@suse.de>
2861 * sese.c: Include cfganal.h.
2862 (if_region_set_false_region): Remove.
2863 (create_if_region_on_edge): Likewise.
2864 (move_sese_in_condition): Re-implement without destroying
2867 2017-09-22 Richard Biener <rguenther@suse.de>
2869 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
2870 Verify both BBs contain loop PHI nodes before dispatching to
2872 (graphite_regenerate_ast_isl): Do not recompute dominators,
2873 do not verify three times. Restructure for clarity.
2874 * graphite-scop-detection.c (same_close_phi_node,
2875 remove_duplicate_close_phi, make_close_phi_nodes_unique,
2876 defined_in_loop_p, canonicalize_loop_closed_ssa,
2877 canonicalize_loop_closed_ssa_form): Simplify, remove excess
2878 checking and SSA rewrite, move to ...
2879 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
2880 (graphite_initialize): Do not pass in ctx, do not reset the
2881 SCEV cache, compute only dominators.
2882 (graphite_transform_loops): Allocate ISL ctx after
2883 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
2884 Maintain post-dominators only around build_scops.
2885 * sese.c (if_region_set_false_region): Make static. Free
2886 and recompute dominators.
2887 (move_sese_in_condition): Assert we don't get called with
2888 post-dominators computed.
2889 * sese.h (if_region_set_false_region): Remove.
2891 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
2893 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
2894 mode attribute for TARGET_AVX512VL.
2896 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
2898 * config/i386/i386.opt (mprefer-avx256): New option.
2899 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
2901 (ix86_preferred_simd_mode): Return 256-bit AVX modes
2902 for TARGET_PREFER_AVX256.
2903 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
2905 2017-09-21 Jeff Law <law@redhat.com>
2907 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
2908 Fix dump output if the only stack space is for pushed registers.
2910 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2912 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
2915 2017-09-21 Martin Sebor <msebor@redhat.com>
2918 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
2919 code (in C++) or code that triggers warnings.
2921 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
2923 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
2925 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
2927 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
2928 * sched-rgn.c: Ditto.
2929 * sel-sched-ir.c: Ditto.
2931 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
2933 * toplev.h (set_random_seed): Adjust return type.
2934 * toplev.c (init_local_tick): Move eager initialization of random_seed
2935 to get_random_seed. Adjust comment.
2936 (init_random_seed): Inline to get_random_seed, delete.
2937 (get_random_seed): Initialize random_seed lazily.
2938 (set_random_seed): Do not return previous value.
2939 (print_switch_value): Do not call get_random_seed.
2941 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
2943 * cgraph.c (delete_function_version): New, broken out from...
2944 (cgraph_node::delete_function_version): ...here. Rename to
2945 cgraph_node::delete_function_version_by_decl. Update all uses.
2946 (cgraph_node::remove): Call delete_function_version.
2948 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2951 * tree-inline.c (expand_call_inline): Emit clobber stmts for
2952 VAR_DECLs to which addressable non-volatile parameters are mapped
2953 and for id->retvar after the return value assignment. Clear
2954 id->retval and id->retbnd after inlining.
2956 2017-09-21 Richard Biener <rguenther@suse.de>
2958 PR tree-optimization/82276
2959 PR tree-optimization/82244
2960 * tree-vrp.c (build_assert_expr_for): Set
2961 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
2963 (remove_range_assertions): Revert earlier change.
2965 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
2968 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
2970 2017-09-21 Richard Biener <rguenther@suse.de>
2972 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2973 Restore valid IL after code generation errors.
2974 * graphite.c (graphite_transform_loops): Diagnose code
2975 generation issues as MSG_MISSED_OPTIMIZATION and continue
2976 with processing SCOPs.
2978 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2979 Alan Hayward <alan.hayward@arm.com>
2980 David Sherwood <david.sherwood@arm.com>
2982 * calls.c (compute_argument_addresses): Use simplify_gen_binary
2983 rather than choosing between plus_constant and gen_rtx_<CODE>.
2984 * expr.c (emit_push_insn): Likewise.
2985 (expand_expr_real_2): Likewise.
2987 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2988 Alan Hayward <alan.hayward@arm.com>
2989 David Sherwood <david.sherwood@arm.com>
2991 * loop-unroll.c (split_iv): Call copy_rtx on the step.
2993 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2994 Alan Hayward <alan.hayward@arm.com>
2995 David Sherwood <david.sherwood@arm.com>
2997 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
2998 calling tree_to_uhwi.
3000 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3001 Alan Hayward <alan.hayward@arm.com>
3002 David Sherwood <david.sherwood@arm.com>
3004 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
3005 INTEGER_CST rather than a negative test for ADDR_EXPR.
3007 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3008 Alan Hayward <alan.hayward@arm.com>
3009 David Sherwood <david.sherwood@arm.com>
3011 * tree-vrp.c (extract_range_from_binary_expr_1): Check
3012 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
3014 2017-09-21 Richard Biener <rguenther@suse.de>
3016 PR tree-optimization/71351
3017 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
3018 graphite_create_new_loop_guard): Remove, fold remaining parts
3020 (translate_isl_ast_node_for): ... here and simplify.
3022 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3025 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
3026 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
3027 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
3028 alternative always use QI mode, for -Os imov (=R,R) alternative
3029 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
3032 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3033 Jeff Law <law@redhat.com>
3035 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
3036 (allocate_stack_space): New function, partially extracted from
3038 (s390_emit_prologue): Track offset to most recent stack probe.
3039 Code to allocate space moved into allocate_stack_space.
3040 Dump actions when no stack is allocated.
3041 (s390_prologue_plus_offset): New function.
3042 (s390_emit_stack_probe): Likewise.
3044 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
3046 * common.opt (Wa, Wl, Wp, g, gz=): Add
3048 (gno-column-info): Remove.
3049 (gcolumn-info): Drop RejectNegative.
3051 (gno-record-gcc-switches): Remove.
3052 (grecord-gcc-switches): Drop RejectNegative.
3053 (gno-split-dwarf): Remove.
3054 (gsplit-dwarf): Drop RejectNegative.
3055 (gno-strict-dwarf): Remove.
3056 (gstrict-dwarf): Drop RejectNegative.
3057 * config/darwin.opt (gfull, gused): Add RejectNegative.
3058 * dwarf2out.c (gen_producer_string): Drop
3059 gno-record-gcc-switches handler.
3060 * optc-gen.awk: Add g to prefixes with negative forms.
3061 * opts-common.c (remapping_prefix_p): New.
3062 (find_opt): Check it.
3063 (generate_canonical_option): Test g prefix.
3064 (option_map): Add -gno- mapping.
3065 (add_misspelling_candidates): Check remapping_prefix_p.
3067 2017-09-20 Jeff Law <law@redhat.com>
3069 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
3070 thinko in stack clash protection support.
3072 * explow.c (compute_stack_clash_protection_loop_data): Use
3073 CONST_INT_P instead of explicit test. Verify object is a
3074 CONST_INT_P before looking at INTVAL.
3075 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
3076 instead of explicit test.
3078 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3081 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
3082 address instead of to r1 and r11.
3084 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
3086 * config.gcc: Support "knm".
3087 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
3088 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3090 * config/i386/i386.c (m_KNM): Define.
3091 (processor_target_table): Add "knm".
3093 (ix86_option_override_internal): Add "knm".
3094 (ix86_issue_rate): Add PROCESSOR_KNM.
3095 (ix86_adjust_cost): Ditto.
3096 (ia32_multipass_dfa_lookahead): Ditto.
3097 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
3098 (fold_builtin_cpu): Add M_INTEL_KNM.
3099 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
3100 (processor_type): Add PROCESSOR_KNM.
3101 * config/i386/x86-tune.def: Add m_KNM.
3102 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
3104 2017-09-20 Richard Biener <rguenther@suse.de>
3106 PR tree-optimization/80213
3107 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
3108 are allowed in empty BBs as well.
3109 (canonicalize_loop_closed_ssa): Also look for other complex
3111 (scop_detection::get_sese): Include the loop-closed PHI block
3113 (scop_detection::merge_sese): Remove code adding extra blocks.
3114 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
3115 (build_scops): Assert the final returned scop is invalid.
3117 2017-09-20 Richard Biener <rguenther@suse.de>
3119 PR tree-optimization/82264
3120 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
3122 (vn_phi_lookup): Likewise.
3123 (vn_phi_insert): Likewise.
3125 2017-09-20 Jakub Jelinek <jakub@redhat.com>
3127 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
3128 that fits into uhwi or shwi, add DW_AT_const_value regardless
3129 of early_dwarf without going through RTL, using add_AT_unsigned
3132 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
3133 (DEBUG_LTO_ABBREV_SECTION): Likewise.
3134 (DEBUG_LTO_MACINFO_SECTION): Likewise.
3135 (DEBUG_MACRO_SECTION): Likewise.
3136 (DEBUG_LTO_MACRO_SECTION): Likewise.
3137 (DEBUG_STR_DWO_SECTION): Likewise.
3138 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
3139 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
3140 (DEBUG_LTO_DWO_LINE_SECTION): Define.
3141 (DEBUG_LTO_LINE_STR_SECTION): Define.
3142 (init_sections_and_labels): Initialize debug_line_str_section
3143 variable. Initialize debug_loc_section for -gdwarf-5 to
3144 DEBUG_LOCLISTS_SECTION. Formatting fixes.
3146 2017-09-20 Richard Biener <rguenther@suse.de>
3148 * graphite-sese-to-poly.c (extract_affine): Properly handle
3149 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
3151 2017-09-20 Richard Biener <rguenther@suse.de>
3153 PR tree-optimization/81373
3154 * graphite-scop-detection.c (build_cross_bb_scalars_def):
3155 Force SESE live-out defs to be handled even if they are
3158 2017-09-19 Jeff Law <law@redhat.com>
3160 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
3161 nothing for stack adjustments with REG_STACK_CHECK.
3162 * sched-deps.c (parse_add_or_inc): Reject insns with
3163 REG_STACK_CHECK from dependency breaking.
3164 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
3165 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
3166 * reg-notes.def (STACK_CHECK): New note.
3168 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
3169 (ix86_expand_prologue): Dump stack clash info as needed.
3170 Call ix86_adjust_stack_and_probe_stack_clash as needed.
3172 * function.c (dump_stack_clash_frame_info): New function.
3173 * function.h (dump_stack_clash_frame_info): Prototype.
3174 (enum stack_clash_probes): New enum.
3176 * config/alpha/alpha.c (alpha_expand_prologue): Also check
3177 flag_stack_clash_protection.
3178 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
3179 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
3180 (arm_frame_pointer_required): Likewise.
3181 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
3182 (ia64_expand_prologue): Likewise.
3183 * config/mips/mips.c (mips_expand_prologue): Likewise.
3184 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
3185 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3186 (sparc_flat_expand_prologue): Likewise.
3187 * config/spu/spu.c (spu_expand_prologue): Likewise.
3189 * explow.c: Include "params.h".
3190 (anti_adjust_stack_and_probe_stack_clash): New function.
3191 (get_stack_check_protect): Likewise.
3192 (compute_stack_clash_protection_loop_data): Likewise.
3193 (emit_stack_clash_protection_loop_start): Likewise.
3194 (emit_stack_clash_protection_loop_end): Likewise.
3195 (allocate_dynamic_stack_space): Use get_stack_check_protect.
3196 Use anti_adjust_stack_and_probe_stack_clash.
3197 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
3198 (emit_stack_clash_protection_loop_start): Likewise.
3199 (emit_stack_clash_protection_loop_end): Likewise.
3200 * rtl.h (get_stack_check_protect): Prototype.
3201 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
3202 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
3203 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
3205 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
3207 * doc/tm.texi: Rebuilt.
3208 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3209 get_stack_check_protect.
3210 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
3211 * config/arm/arm.c (arm_expand_prologue): Likewise.
3212 (arm_frame_pointer_required): Likewise.
3213 * config/i386/i386.c (ix86_expand_prologue): Likewise.
3214 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
3215 * config/mips/mips.c (mips_expand_prologue): Likewise.
3216 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
3217 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
3218 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3219 (sparc_flat_expand_prologue): Likewise.
3221 * common.opt (-fstack-clash-protection): New option.
3222 * flag-types.h (enum stack_check_type): Note difference between
3223 -fstack-check= and -fstack-clash-protection.
3224 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
3225 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
3226 * toplev.c (process_options): Issue warnings/errors for cases
3227 not handled with -fstack-clash-protection.
3228 * doc/invoke.texi (-fstack-clash-protection): Document new option.
3229 (-fstack-check): Note additional problem with -fstack-check=generic.
3230 Note that -fstack-check is primarily for Ada and refer users
3231 to -fstack-clash-protection for stack-clash-protection.
3232 Document new params for stack clash protection.
3234 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
3236 * config/i386/i386.c (ix86_split_long_move): Do not handle
3237 address used for LEA in a special way.
3239 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3241 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
3243 2017-09-19 Martin Sebor <msebor@redhat.com>
3246 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
3247 of incompatible types.
3249 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
3251 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3252 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
3253 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3254 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
3256 2017-09-19 Richard Biener <rguenther@suse.de>
3258 PR tree-optimization/82244
3259 * tree-vrp.c (remove_range_assertions): Do not propagate
3260 a constant to abnormals but replace the assert with a copy.
3262 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
3264 PR rtl-optimization/57878
3265 PR rtl-optimization/68988
3266 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
3267 avoidance test involving non_reload_pseudos. Move frequency test
3268 below the general fragmentation avoidance test.
3270 2017-09-19 Richard Biener <rguenther@suse.de>
3272 PR tree-optimization/69728
3273 * graphite-sese-to-poly.c (schedule_error): New global.
3274 (add_loop_schedule): Handle empty domain by failing the
3276 (build_original_schedule): Handle schedule_error.
3278 2017-09-19 Richard Biener <rguenther@suse.de>
3280 * graphite-scop-detection.c (scop_detection::can_represent_loop):
3281 Do not iterate to sibling loops but only to siblings of inner
3284 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
3287 * config/m68k/m68k.md (moveq feeding equality comparison): Check
3288 that the registers are different.
3290 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
3292 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
3293 to processor_model and "amdfam17h" to arch_names_table.
3294 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
3296 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3299 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
3301 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3302 Alan Hayward <alan.hayward@arm.com>
3303 David Sherwood <david.sherwood@arm.com>
3305 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
3307 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
3309 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
3310 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
3311 vect_schedule_slp_instance.
3312 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
3313 Update call to vect_slp_analyze_node_operations. Simplify return
3315 (vect_slp_analyze_bb_1): Update call accordingly.
3316 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
3317 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
3318 (vect_schedule_slp): Update call accordingly.
3320 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3321 Alan Hayward <alan.hayward@arm.com>
3322 David Sherwood <david.sherwood@arm.com>
3324 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
3325 with types that aren't in fact scalar.
3327 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3329 * tree-vect-slp.c (vect_record_max_nunits): New function,
3331 (vect_build_slp_tree_1): ...here.
3332 (vect_build_slp_tree_2): Call it for phis too.
3334 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3336 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
3337 to vect_get_vec_def_for_operand when getting the mask operand.
3339 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3340 Alan Hayward <alan.hayward@arm.com>
3341 David Sherwood <david.sherwood@arm.com>
3343 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
3346 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3347 Alan Hayward <alan.hayward@arm.com>
3348 David Sherwood <david.sherwood@arm.com>
3350 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
3351 calculation for vector booleans.
3353 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3354 Alan Hayward <alan.hayward@arm.com>
3355 David Sherwood <david.sherwood@arm.com>
3357 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
3359 (vect_transform_stmt): ...here.
3360 (vect_analyze_stmt): Use it instead of calling
3361 vectorizable_live_operation directly.
3363 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
3365 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
3366 non-SIMT targets in acc vector loops.
3368 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
3370 * configure.ac: Add arc and check if assembler supports gdwarf2.
3371 * configure: Regenerate.
3373 2017-09-18 Richard Biener <rguenther@suse.de>
3375 PR tree-optimization/82220
3376 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
3377 epilogue niters from the min_profitable_iters compute.
3379 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3382 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
3383 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
3384 (ix86_init_pic_reg): Revert 2017-09-01 changes.
3386 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
3389 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
3390 switching to a new text section.
3392 2017-09-18 Richard Biener <rguenther@suse.de>
3394 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
3396 (build_alias_set): Reject aliases with no access function.
3398 2017-09-18 Richard Biener <rguenther@suse.de>
3400 PR tree-optimization/79622
3401 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
3403 (build_cross_bb_scalars_use): Likewise.
3405 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
3407 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
3409 2017-09-18 Alan Modra <amodra@gmail.com>
3412 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
3413 stack_pointer_rtx for count 0. Update comments. Break up
3414 large rtl expression.
3416 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
3418 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
3419 Increase to 20 bytes.
3420 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
3421 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
3422 or avx version of the stub.
3424 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
3427 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
3428 compute the minimum stack alignment. Also update preferred stack
3429 boundary for leaf functions.
3431 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
3433 PR tree-optimization/82228
3434 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
3437 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3439 * common/config/nds32/nds32-common.c
3440 (nds32_option_optimization_table): Refine formatting.
3441 (nds32_option_optimization_table): Use -fsched-pressure and
3442 -fomit-frame-pointer for specific optimization level.
3444 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3446 * config/nds32/nds32.c: Refine formatting and comments.
3447 * config/nds32/nds32.h: Likewise.
3448 * config/nds32/nds32.md: Likewise.
3449 * config/nds32/nds32-cost.c: Likewise.
3450 * config/nds32/nds32-isr.c: Likewise.
3451 * config/nds32/nds32-md-auxiliary.c: Likewise.
3452 * config/nds32/nds32-multiple.md: Likewise.
3453 * config/nds32/nds32-predicates.c: Likewise.
3455 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
3456 Jakub Jelinek <jakub@redhat.com>
3458 Add support for -std=c++2a.
3459 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
3461 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
3463 2017-09-15 Steve Ellcey <sellcey@cavium.com>
3466 * doc/extend.texi (Common Function Attributes): Add
3467 references to ARM, AArch64, and S/390 specific attributes.
3468 (Function Specific Option Pragmas): Add AArch64 and S/390
3469 to list of back ends that support the target pragma.
3471 2017-09-15 Nathan Sidwell <nathan@acm.org>
3473 * doc/standards.texi: Fix C++17 description. Update URLs for
3476 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
3478 * common.opt (Wcast-align=strict): New warning option.
3479 * doc/invoke.texi: Document -Wcast-align=strict.
3481 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
3483 * cgraph.h (cgraph_thunk_info): Add comments.
3484 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
3485 assert for VIRTUAL_* arguments stricter.
3487 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
3489 PR tree-optimization/71026
3490 * match.pd: Move RDIV patterns from fold-const.c
3491 * fold-const.c (distribute_real_division): Removed.
3492 (fold_binary_loc): Remove calls to distribute_real_divison.
3494 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3496 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
3497 c++1z and gnu++1z as deprecated. Change other references to
3498 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
3499 Change -Wc++1z-compat to -Wc++17-compat.
3500 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
3501 * dwarf2out.c (highest_c_language): Handle C++17.
3502 (gen_compile_unit_die): Likewise.
3504 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3506 PR rtl-optimization/82192
3507 * combine.c (make_extraction): Don't look through non-paradoxical
3508 SUBREGs or TRUNCATE if pos + len is or might be bigger than
3511 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3512 Alan Hayward <alan.hayward@arm.com>
3513 David Sherwood <david.sherwood@arm.com>
3515 * target.def (function_arg_offset): New hook.
3516 * targhooks.h (default_function_arg_offset): Declare.
3517 * targhooks.c (default_function_arg_offset): New function.
3518 * function.c (locate_and_pad_parm): Use
3519 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
3520 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
3521 (TARGET_FUNCTION_ARG_OFFSET): ...this.
3522 * doc/tm.texi: Regenerate.
3523 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
3524 * config/spu/spu.c (spu_function_arg_offset): New function.
3525 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
3526 * system.h (FUNCTION_ARG_OFFSET): Poison.
3528 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3529 Alan Hayard <alan.hayward@arm.com>
3530 David Sherwood <david.sherwood@arm.com>
3532 * target.def (truly_noop_truncation): New hook.
3533 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
3534 than TRULY_NOOP_TRUNCATION.
3535 * hooks.h (hook_bool_uint_uint_true): Declare.
3536 * hooks.c (hook_bool_uint_uint_true): New function.
3537 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
3538 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
3539 * doc/tm.texi: Regenerate.
3540 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
3541 rather than TRULY_NOOP_TRUNCATION in comments.
3542 (simplify_comparison): Likewise.
3543 (record_truncated_value): Likewise.
3544 * expmed.c (extract_bit_field_1): Likewise.
3545 (extract_split_bit_field): Likewise.
3546 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
3547 instead of TRULY_NOOP_TRUNCATION.
3548 * function.c (assign_parm_setup_block): Likewise.
3549 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
3550 * rtlhooks.c: Include target.h.
3551 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
3552 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
3553 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
3554 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
3555 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
3556 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
3557 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
3558 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
3559 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
3560 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
3561 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
3562 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
3563 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
3564 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
3565 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
3566 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
3567 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
3568 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
3569 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
3570 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
3571 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
3572 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
3573 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
3574 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
3575 * config/mips/mips.c (mips_truly_noop_truncation): New function.
3576 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3577 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3578 rather than TRULY_NOOP_TRUNCATION in comments.
3579 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
3580 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
3581 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
3582 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
3583 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
3584 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
3585 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
3586 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
3587 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
3588 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
3589 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
3590 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3591 rather than TRULY_NOOP_TRUNCATION in comments.
3592 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
3593 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
3594 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
3595 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
3596 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
3597 TRULY_NOOP_TRUNCATION condition.
3598 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
3599 (TRULY_NOOP_TRUNCATION): Delete.
3600 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
3601 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
3602 * config/spu/spu.c (spu_truly_noop_truncation): New function.
3603 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3604 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
3605 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
3606 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
3607 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3608 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3609 rather than TRULY_NOOP_TRUNCATION in comments.
3610 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
3611 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
3612 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
3613 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
3614 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
3615 * system.h (TRULY_NOOP_TRUNCATION): Poison.
3617 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
3620 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
3621 (*cmp_ior): Likewise.
3622 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
3623 (*ior_scc_scc_cmp): Likewise.
3624 (*and_scc_scc): Likewise.
3625 (*and_scc_scc_cmp): Likewise.
3627 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3628 Alan Hayard <alan.hayward@arm.com>
3629 David Sherwood <david.sherwood@arm.com>
3631 * target.def (can_change_mode_class): New hook.
3632 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
3633 (hard_regno_nregs): Likewise.
3634 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
3635 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
3636 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
3637 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
3638 (LOAD_EXTEND_OP): Update accordingly.
3639 * doc/tm.texi: Regenerate.
3640 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
3641 CANNOT_CHANGE_MODE_CLASS.
3642 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
3643 (REG_CAN_CHANGE_MODE_P): ...this new macro.
3644 * combine.c (simplify_set): Update accordingly.
3645 * emit-rtl.c (validate_subreg): Likewise.
3646 * recog.c (general_operand): Likewise.
3647 * regcprop.c (mode_change_ok): Likewise.
3648 * reload1.c (choose_reload_regs): Likewise.
3649 (inherit_piecemeal_p): Likewise.
3650 * rtlanal.c (simplify_subreg_regno): Likewise.
3651 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
3652 instead of CANNOT_CHANGE_MODE_CLASS.
3653 (reload_cse_simplify_operands): Likewise.
3654 * reload.c (push_reload): Use targetm.can_change_mode_class
3655 instead of CANNOT_CHANGE_MODE_CLASS.
3656 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
3657 REG_CANNOT_CHANGE_MODE_P.
3658 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3659 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
3660 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3661 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3662 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3663 (arm_can_change_mode_class): New function.
3664 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3665 than CANNOT_CHANGE_MODE_CLASS in comments.
3666 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3667 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
3668 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
3669 (ix86_can_change_mode_class): ...this new function, inverting the
3670 sense of the return value.
3671 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3672 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3673 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3674 (ia64_can_change_mode_class): New function.
3675 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3676 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
3677 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
3678 (m32c_can_change_mode_class): ...this new function, inverting the
3679 sense of the return value.
3680 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3681 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3682 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
3683 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
3684 (mips_can_change_mode_class): ...this new function, inverting the
3685 sense of the return value.
3686 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3687 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3688 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3689 (msp430_can_change_mode_class): New function.
3690 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3691 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
3692 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3693 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3694 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3695 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
3696 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3697 (pa_cannot_change_mode_class): Replace with...
3698 (pa_can_change_mode_class): ...this new function, inverting the
3699 sense of the return value.
3700 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3701 than CANNOT_CHANGE_MODE_CLASS in comments.
3702 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3703 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
3704 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3705 (pdp11_cannot_change_mode_class): Replace with...
3706 (pdp11_can_change_mode_class): ...this new function, inverting the
3707 sense of the return value.
3708 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3709 * config/powerpcspe/powerpcspe-protos.h
3710 (rs6000_cannot_change_mode_class_ptr): Delete.
3711 * config/powerpcspe/powerpcspe.c
3712 (rs6000_cannot_change_mode_class_ptr): Delete.
3713 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3714 (rs6000_option_override_internal): Assign to
3715 targetm.can_change_mode_class instead of
3716 rs6000_cannot_change_mode_class_ptr.
3717 (rs6000_cannot_change_mode_class): Replace with...
3718 (rs6000_can_change_mode_class): ...this new function, inverting the
3719 sense of the return value.
3720 (rs6000_debug_cannot_change_mode_class): Replace with...
3721 (rs6000_debug_can_change_mode_class): ...this new function.
3722 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3723 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
3724 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3725 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3726 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
3728 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
3729 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3730 (rs6000_option_override_internal): Assign to
3731 targetm.can_change_mode_class instead of
3732 rs6000_cannot_change_mode_class_ptr.
3733 (rs6000_cannot_change_mode_class): Replace with...
3734 (rs6000_can_change_mode_class): ...this new function, inverting the
3735 sense of the return value.
3736 (rs6000_debug_cannot_change_mode_class): Replace with...
3737 (rs6000_debug_can_change_mode_class): ...this new function.
3738 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3739 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
3740 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
3741 (s390_can_change_mode_class): ...this new function, inverting the
3742 sense of the return value.
3743 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3744 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3745 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
3746 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3747 (sh_cannot_change_mode_class): Replace with...
3748 (sh_can_change_mode_class): ...this new function, inverting the
3749 sense of the return value.
3750 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3751 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3752 (sparc_can_change_mode_class): New function.
3753 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3754 * config/spu/spu.c (spu_can_change_mode_class): New function.
3755 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3756 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3757 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3758 (visium_can_change_mode_class): New function.
3759 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
3761 2017-09-15 Richard Biener <rguenther@suse.de>
3763 PR tree-optimization/82217
3764 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
3765 but not undefined case.
3767 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3770 * postreload.c (reload_cse_simplify_operands): Skip
3771 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
3773 2017-09-15 Richard Biener <rguenther@suse.de>
3775 PR tree-optimization/68823
3776 * graphite-scop-detection.c (build_alias_set): If we have a
3777 possible dependence check whether we can handle them by just
3778 looking at the DRs DR_ACCESS_FNs.
3779 (build_scops): If build_alias_set fails, fail the SCOP.
3781 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3783 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
3784 to support float128 built-in functions that require the ISA 3.0
3786 (BU_FLOAT128_3_HW): Likewise.
3787 (SQRTF128): Add support for the IEEE 128-bit square root and fma
3789 (FMAF128): Likewise.
3791 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
3792 support for built-in functions that need the ISA 3.0 IEEE 128-bit
3793 floating point instructions.
3794 (rs6000_invalid_builtin): Likewise.
3795 (rs6000_builtin_mask_names): Likewise.
3796 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
3797 (RS6000_BTM_FLOAT128_HW): Likewise.
3798 (RS6000_BTM_COMMON): Likewise.
3799 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
3801 * doc/extend.texi (RS/6000 built-in functions): Document the
3802 IEEE 128-bit floating point square root and fused multiply-add
3805 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
3807 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
3808 reg (r2) isn't in the set of registers defined in the prologue.
3810 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3811 Alan Hayward <alan.hayward@arm.com>
3812 David Sherwood <david.sherwood@arm.com>
3814 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
3815 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
3816 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
3817 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
3818 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
3820 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
3821 max_vectorization_factor.
3822 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
3824 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3825 Alan Hayward <alan.hayward@arm.com>
3826 David Sherwood <david.sherwood@arm.com>
3828 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
3829 (vect_worthwhile_without_simd_p): Declare.
3830 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
3831 (vectorizable_reduction): Use it.
3832 * tree-vect-stmts.c (vectorizable_shift): Likewise.
3833 (vectorizable_operation): Likewise.
3835 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3836 Alan Hayward <alan.hayward@arm.com>
3837 David Sherwood <david.sherwood@arm.com>
3839 * tree-vectorizer.h (vect_get_num_copies): New function.
3840 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
3841 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3842 (vectorizable_induction): Likewise.
3843 (vectorizable_live_operation): Likewise.
3844 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3845 (vectorizable_bswap): Likewise.
3846 (vectorizable_call): Likewise.
3847 (vectorizable_conversion): Likewise.
3848 (vectorizable_assignment): Likewise.
3849 (vectorizable_shift): Likewise.
3850 (vectorizable_operation): Likewise.
3851 (vectorizable_store): Likewise.
3852 (vectorizable_load): Likewise.
3853 (vectorizable_condition): Likewise.
3854 (vectorizable_comparison): Likewise.
3855 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
3857 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3858 Alan Hayward <alan.hayward@arm.com>
3859 David Sherwood <david.sherwood@arm.com>
3861 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
3862 of vect_init_vector.
3864 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3865 Alan Hayward <alan.hayward@arm.com>
3866 David Sherwood <david.sherwood@arm.com>
3868 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
3869 an inline wrapper that provides a location.
3870 (gimple_build_vector): Likewise.
3871 * gimple-fold.c (gimple_build_vector_from_val): New function.
3872 (gimple_build_vector): Likewise.
3873 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
3874 functions to build the initial value. Always return a gimple value.
3875 (get_initial_defs_for_reduction): Likewise. Only compute
3877 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
3878 vect_init_vector on the results from get_initial_def(s)_for_reduction.
3879 (vectorizable_induction): Use gimple_build_vector rather than
3882 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3883 Alan Hayward <alan.hayward@arm.com>
3884 David Sherwood <david.sherwood@arm.com>
3886 * target.h (vec_perm_indices): New typedef.
3887 (auto_vec_perm_indices): Likewise.
3888 * optabs-query.h: Include target.h
3889 (can_vec_perm_p): Take a vec_perm_indices *.
3890 * optabs-query.c (can_vec_perm_p): Likewise.
3891 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
3892 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3893 * tree-vect-generic.c (lower_vec_perm): Likewise.
3894 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
3895 (vect_grouped_load_supported): Likewise.
3896 (vect_shift_permute_load_chain): Likewise.
3897 (vect_permute_store_chain): Use auto_vec_perm_indices.
3898 (vect_permute_load_chain): Likewise.
3899 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
3900 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
3901 Update uses of can_vec_perm_p.
3902 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
3903 mode with a number of elements. Take a vec_perm_indices *.
3904 (vect_create_epilog_for_reduction): Update accordingly.
3905 Use auto_vec_perm_indices.
3906 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
3907 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
3908 (vect_transform_slp_perm_load): Likewise.
3909 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
3910 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
3911 (vect_gen_perm_mask_checked): Likewise.
3912 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
3913 (vect_gen_perm_mask_checked): Likewise.
3914 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
3915 (vectorizable_store): Likewise.
3916 (vectorizable_load): Likewise.
3917 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
3918 (vectorizable_bswap): Likewise.
3920 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3921 Alan Hayward <alan.hayward@arm.com>
3922 David Sherwood <david.sherwood@arm.com>
3924 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
3925 * tree.c (build_vector): Likewise.
3926 (build_vector_from_ctor): Update accordingly.
3927 (build_vector_from_val): Likewise.
3928 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
3929 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3930 * tree-vect-generic.c (add_rshift): Likewise.
3931 (expand_vector_divmod): Likewise.
3932 (optimize_vector_constructor): Likewise.
3933 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3934 (vect_transform_slp_perm_load): Likewise.
3935 (vect_schedule_slp_instance): Likewise.
3936 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
3937 (vectorizable_call): Likewise.
3938 (vect_gen_perm_mask_any): Likewise. Add elements in order.
3939 * expmed.c (make_tree): Likewise.
3940 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
3941 a vector passed to build_vector.
3942 (fold_convert_const): Likewise.
3943 (exact_inverse): Likewise.
3944 (fold_ternary_loc): Likewise.
3945 (fold_relational_const): Likewise.
3946 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
3947 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
3948 (const_unop): Likewise. Store the reduction accumulator in a
3949 variable rather than an array.
3950 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
3951 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
3952 the new vector, rather than constructing it after the input arrays.
3953 (native_interpret_vector): Use auto_vec<tree> when building
3954 a vector passed to build_vector. Add elements in order.
3955 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
3956 auto_vec<tree> when building a vector passed to build_vector.
3957 (vect_create_epilog_for_reduction): Likewise.
3958 (vectorizable_induction): Likewise.
3959 (get_initial_def_for_reduction): Likewise. Fix indentation of
3961 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
3963 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
3964 passed to build_vector.
3966 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3967 Alan Hayward <alan.hayward@arm.com>
3968 David Sherwood <david.sherwood@arm.com>
3970 * tree-core.h (tree_base::u): Add an "nelts" field.
3971 (tree_vector): Use VECTOR_CST_NELTS as the length.
3972 * tree.c (tree_size): Likewise.
3973 (make_vector): Initialize VECTOR_CST_NELTS.
3974 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
3975 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
3976 TYPE_VECTOR_SUBPARTS.
3977 * expr.c (const_vector_mask_from_tree): Consistently use "units"
3978 as the number of units, setting it from VECTOR_CST_NELTS.
3979 (const_vector_from_tree): Likewise.
3980 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
3981 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
3982 (fold_negate_expr_1): Likewise.
3983 (fold_convert_const): Likewise.
3984 (const_binop): Likewise. Differentiate the number of output and
3986 (const_unop): Likewise.
3987 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
3988 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
3989 in cases that did the opposite.
3991 2017-09-14 Richard Biener <rguenther@suse.de>
3993 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
3996 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
3998 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
4000 2017-09-14 Jakub Jelinek <jakub@redhat.com>
4003 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
4004 if and where to split a bb, except for splitting before debug insn
4005 sequences followed by non-label real insn. Delete debug insns
4006 in between basic blocks.
4008 * combine.c (make_compound_operation_int): Formatting fixes.
4010 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
4011 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4012 * config/netbsd.h (LINK_EH_SPEC): Likewise.
4013 * config/sol2.h (LINK_EH_SPEC): Likewise.
4014 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4015 * config/s390/linux.h (LINK_SPEC): Likewise.
4016 * config/freebsd.h (LINK_EH_SPEC): Likewise.
4017 * config/openbsd.h (LINK_EH_SPEC): Likewise.
4018 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4019 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
4020 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
4021 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4022 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
4023 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
4025 2017-09-13 Jakub Jelinek <jakub@redhat.com>
4027 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
4029 (ENDFILE_LINUX_SPEC): Likewise.
4030 (LINK_EH_SPEC): Likewise.
4031 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
4032 (LINK_OS_LINUX_SPEC32): Likewise.
4033 (LINK_OS_LINUX_SPEC64): Likewise.
4034 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
4035 (LINK_OS_LINUX_SPEC): Likewise.
4037 2017-09-13 Martin Liska <mliska@suse.cz>
4040 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
4041 CASE_HIGH is NULL_TREE.
4043 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4044 Alan Hayward <alan.hayward@arm.com>
4045 David Sherwood <david.sherwood@arm.com>
4047 * target.def (secondary_memory_needed): New hook.
4048 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
4049 instead of SECONDARY_MEMORY_NEEDED.
4050 (secondary_memory_needed_mode): Likewise.
4051 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
4052 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
4053 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
4054 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
4055 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
4056 * doc/tm.texi: Regenerate.
4057 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
4058 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
4059 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4060 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
4061 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
4062 * config/i386/i386.c (inline_secondary_memory_needed): Put the
4063 mode argument first and change the reg_class arguments to reg_class_t.
4064 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
4065 Make static. Update the call to inline_secondary_memory_needed.
4066 (ix86_register_move_cost): Update the call to
4067 inline_secondary_memory_needed.
4068 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4069 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
4071 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
4072 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4074 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
4075 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
4076 * config/mips/mips.c (mips_secondary_memory_needed): Make static
4077 and match hook interface. Add comment from mips.h.
4078 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4079 * config/mmix/mmix.md (truncdfsf2): Refer to
4080 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4082 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
4083 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
4084 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4085 (pa_secondary_memory_needed): New function.
4086 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
4087 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
4088 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4089 (pdp11_secondary_memory_needed): Make static and match hook interface.
4090 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
4091 * config/powerpcspe/powerpcspe-protos.h
4092 (rs6000_secondary_memory_needed_ptr): Delete.
4093 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
4095 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4096 (rs6000_option_override_internal): Assign to
4097 targetm.secondary_memory_needed rather than
4098 rs6000_secondary_memory_needed_ptr.
4099 (rs6000_secondary_memory_needed): Match hook interface.
4100 (rs6000_debug_secondary_memory_needed): Likewise.
4101 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
4102 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
4103 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
4104 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4105 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
4106 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
4108 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
4109 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4110 (rs6000_option_override_internal): Assign to
4111 targetm.secondary_memory_needed rather than
4112 rs6000_secondary_memory_needed_ptr.
4113 (rs6000_secondary_memory_needed): Match hook interface.
4114 (rs6000_debug_secondary_memory_needed): Likewise.
4115 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
4116 * config/s390/s390.c (s390_secondary_memory_needed): New function.
4117 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4118 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
4119 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4120 (sparc_secondary_memory_needed): New function.
4121 * lra-constraints.c (check_and_process_move): Refer to
4122 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4124 (curr_insn_transform): Likewise.
4125 (process_alt_operands): Use targetm.secondary_memory_needed
4126 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4127 (check_secondary_memory_needed_p): Likewise.
4128 (choose_split_class): Likewise.
4129 * reload.c: Unconditionally include code that was previously
4130 conditional on SECONDARY_MEMORY_NEEDED.
4131 (push_secondary_reload): Use targetm.secondary_memory_needed
4132 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4133 (push_reload): Likewise.
4134 * reload1.c: Unconditionally include code that was previously
4135 conditional on SECONDARY_MEMORY_NEEDED.
4136 (choose_reload_regs): Use targetm.secondary_memory_needed
4137 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4138 (gen_reload): Likewise.
4139 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
4141 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4142 Alan Hayward <alan.hayward@arm.com>
4143 David Sherwood <david.sherwood@arm.com>
4145 * target.def (secondary_memory_needed_mode): New hook:
4146 * targhooks.c (default_secondary_memory_needed_mode): Declare.
4147 * targhooks.h (default_secondary_memory_needed_mode): New function.
4148 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
4149 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
4150 * doc/tm.texi: Regenerate.
4151 * lra-constraints.c (check_and_process_move): Use
4152 targetm.secondary_memory_needed_mode instead of
4153 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
4154 (curr_insn_transform): Likewise.
4155 * reload.c (get_secondary_mem): Likewise.
4156 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4157 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
4159 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4160 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4161 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
4162 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4163 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
4165 * config/powerpcspe/powerpcspe-protos.h
4166 (rs6000_secondary_memory_needed_mode): Delete.
4167 * config/powerpcspe/powerpcspe.c
4168 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4169 (rs6000_secondary_memory_needed_mode): Make static.
4170 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4171 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
4173 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4175 (rs6000_secondary_memory_needed_mode): Make static.
4176 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4177 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
4178 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4179 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4180 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4182 (sparc_secondary_memory_needed_mode): New function.
4183 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
4185 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
4187 * config/aarch64/constraints.md (Umq): New constraint.
4188 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
4190 (mov<mode>): Update condition.
4192 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4194 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
4195 when bitposition is the same.
4197 2017-09-13 Richard Biener <rguenther@suse.de>
4199 * dwarf2out.c (output_die_symbol): Remove.
4200 (output_die): Do not output a DIEs symbol.
4202 2017-09-13 Richard Biener <rguenther@suse.de>
4205 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
4206 default-def to avoid breaking iterator update with the weird
4207 interaction with cgraph_update_edges_for_call_stmt_node.
4209 2017-09-13 Richard Biener <rguenther@suse.de>
4211 * tree-cfg.c (verify_gimple_assign_binary): Add verification
4212 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
4213 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
4214 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
4216 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
4218 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
4219 Disable pc relative literal load irrespective of
4220 TARGET_FIX_ERR_A53_84341 for default.
4222 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
4224 * config/sparc/sparc.c (output_return): Output the source location of
4225 the insn in the delay slot, if any.
4226 (output_sibcall): Likewise.
4228 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
4231 * common.opt (-static-pie): New alias.
4232 (shared): Negate static-pie.
4233 (-no-pie): Update help text.
4235 (static-pie): New option.
4236 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
4237 -static-pie support.
4238 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
4239 (LINK_EH_SPEC): Likewise.
4240 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4241 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4242 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4243 * gcc.c (LINK_COMMAND_SPEC): Likewise.
4244 (init_gcc_specs): Likewise.
4245 (init_spec): Likewise.
4246 (display_help): Update help message for -pie.
4247 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
4250 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
4252 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
4253 (movdi_aarch64): Likewise.
4254 (movti_aarch64): Likewise.
4256 2017-09-12 Simon Wright <simon@pushface.org>
4259 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
4260 calculation of the minor version, always output as 0.
4262 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4265 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4266 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
4267 on it early, rather than manual conversion late. For
4268 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
4269 instead of performing manual conversion.
4271 2017-09-12 Carl Love <cel@us.ibm.com>
4273 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
4274 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
4276 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4277 VMULOSW): Add definitions.
4278 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4279 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4280 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4281 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4282 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4284 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4286 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
4288 (movti_aarch64): Likewise.
4289 (movdf_aarch64): Likewise.
4290 (movtf_aarch64): Likewise.
4291 (load_pairdi): Likewise.
4292 (store_pairdi): Likewise.
4293 (load_pairdf): Likewise.
4294 (store_pairdf): Likewise.
4295 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
4296 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
4297 (ldr_got_small_<mode>): Likewise.
4298 (ldr_got_small_28k_<mode>): Likewise.
4299 (ldr_got_tiny): Likewise.
4300 * config/aarch64/iterators.md (ldst_sz): New.
4301 (ldpstp_sz): Likewise.
4302 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
4304 (thunderx_load): Split load_8 to load_16.
4305 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
4307 (thunderx2t99_storepair_basic): Split store_8 to store_16.
4308 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
4309 (xgene1_store_pair): Split store_8 to store_16.
4310 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
4311 (falkor_st_0_st_sd): Split store_8 to store_16.
4313 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4315 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
4316 and store1/2/3/4 to store_4/8/12/16.
4317 * config/aarch64/aarch64.md: Update for rename.
4318 * config/arm/arm.md: Likewise.: Likewise.
4319 * config/arm/arm.c: Likewise.
4320 * config/arm/thumb1.md: Likewise.
4321 * config/arm/thumb2.md: Likewise.
4322 * config/arm/vfp.md: Likewise.
4323 * config/arm/arm-generic.md: Likewise.
4324 * config/arm/arm1020e.md: Likewise.
4325 * config/arm/arm1026ejs.md: Likewise.
4326 * config/arm/arm1136jfs.md: Likewise.
4327 * config/arm/arm926ejs.md: Likewise.
4328 * config/arm/cortex-a15.md: Likewise.
4329 * config/arm/cortex-a17.md: Likewise.
4330 * config/arm/cortex-a5.md: Likewise.
4331 * config/arm/cortex-a53.md: Likewise.
4332 * config/arm/cortex-a57.md: Likewise.
4333 * config/arm/cortex-a7.md: Likewise.
4334 * config/arm/cortex-a8.md: Likewise.
4335 * config/arm/cortex-a9.md: Likewise.
4336 * config/arm/cortex-m4.md: Likewise.
4337 * config/arm/cortex-m7.md: Likewise.
4338 * config/arm/cortex-r4.md: Likewise.
4339 * config/arm/exynos-m1.md: Likewise.
4340 * config/arm/fa526.md: Likewise.
4341 * config/arm/fa606te.md: Likewise.
4342 * config/arm/fa626te.md: Likewise.
4343 * config/arm/fa726te.md: Likewise.
4344 * config/arm/fmp626.md: Likewise.
4345 * config/arm/iwmmxt.md: Likewise.
4346 * config/arm/ldmstm.md: Likewise.
4347 * config/arm/marvell-pj4.md: Likewise.
4348 * config/arm/xgene1.md: Likewise.
4349 * config/aarch64/thunderx.md: Likewise.
4350 * config/aarch64/thunderx2t99.md: Likewise.
4351 * config/aarch64/falkor.md: Likewise.
4353 2017-09-12 Martin Liska <mliska@suse.cz>
4355 * attribs.c (private_lookup_attribute): New function.
4356 * attribs.h (private_lookup_attribute): Declared here.
4357 (lookup_attribute): Called from this place.
4359 2017-09-12 Richard Biener <rguenther@suse.de>
4361 PR tree-optimization/82157
4362 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
4363 stmts with side-effects.
4365 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4366 Alan Hayward <alan.hayward@arm.com>
4367 David Sherwood <david.sherwood@arm.com>
4369 * target.def (hard_regno_nregs): New hook.
4370 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
4371 * targhooks.h (default_hard_regno_nregs): Declare.
4372 * targhooks.c (default_hard_regno_nregs): New function.
4373 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
4374 (TARGET_HARD_REGNO_NREGS): ...this hook.
4375 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
4376 (CLASS_MAX_NREGS): Likewise.
4377 * doc/tm.texi: Regenerate.
4378 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
4379 instead of HARD_REGNO_NREGS.
4380 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
4381 HARD_REGNO_NREGS in the comment.
4382 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
4383 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
4384 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
4385 Return an unsigned int.
4386 (TARGET_HARD_REGNO_NREGS): Redefine.
4387 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
4388 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
4389 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4390 (arc_hard_regno_nregs): New function.
4391 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
4392 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
4393 (arm_hard_regno_nregs): New function.
4394 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
4395 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
4396 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
4397 (TARGET_HARD_REGNO_NREGS): Redefine.
4398 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
4399 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
4400 (HARD_REGNO_NREGS): Delete.
4401 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
4402 (cr16_hard_regno_nregs): New function.
4403 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
4404 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
4405 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
4406 (cris_hard_regno_nregs): New function.
4407 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
4408 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
4409 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4410 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
4411 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
4412 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
4413 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
4414 (frv_hard_regno_nregs): Make static. Take and return an
4416 (frv_class_max_nregs): Remove outdated copy of documentation.
4417 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
4418 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
4419 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
4420 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
4421 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
4422 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
4423 (TARGET_HARD_REGNO_NREGS): Redefine.
4424 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
4425 (CLASS_MAX_NREGS): Update comment.
4426 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
4427 (ia64_hard_regno_nregs): New function.
4428 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
4429 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
4430 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
4431 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
4432 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
4434 (m32c_hard_regno_nregs): Likewise. Make static.
4435 (TARGET_HARD_REGNO_NREGS): Redefine.
4436 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
4437 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
4438 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
4439 (m68k_hard_regno_nregs): New function.
4440 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
4441 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
4442 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
4443 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
4444 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
4445 Take and return an unsigned int.
4446 (TARGET_HARD_REGNO_NREGS): Redefine.
4447 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
4448 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4449 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
4450 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
4451 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
4452 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
4453 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
4454 (msp430_hard_regno_nregs): Make static. Take and return an
4456 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
4457 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
4458 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
4459 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
4460 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
4461 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
4462 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
4463 (TARGET_HARD_REGNO_NREGS): Redefine.
4464 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
4465 (PA_HARD_REGNO_NREGS): ...this.
4466 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
4467 (PA_HARD_REGNO_NREGS): ...this.
4468 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4469 (pa_hard_regno_nregs): New function.
4470 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
4471 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
4472 (pdp11_hard_regno_nregs): New function.
4473 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
4474 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
4475 (rs6000_hard_regno_nregs_hook): New function.
4476 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
4477 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4478 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
4479 Take and return an unsigned int. Move earlier in file.
4480 (TARGET_HARD_REGNO_NREGS): Redefine.
4481 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
4482 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
4483 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
4484 (rl78_hard_regno_nregs): Make static. Take and return an
4486 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
4487 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
4488 (rs6000_hard_regno_nregs_hook): New function.
4489 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
4490 * config/rx/rx.c (rx_hard_regno_nregs): New function.
4491 (TARGET_HARD_REGNO_NREGS): Redefine.
4492 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
4493 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
4494 instead of HARD_REGNO_NREGS.
4495 (s390_hard_regno_nregs): New function.
4496 (s390_hard_regno_mode_ok): Add comment from s390.h.
4497 (TARGET_HARD_REGNO_NREGS): Redefine.
4498 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
4499 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
4500 (sh_hard_regno_nregs): New function.
4501 (sh_pass_in_reg_p): Use it.
4502 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
4503 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4504 (sparc_hard_regno_nregs): New function.
4505 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
4506 * config/spu/spu.c (spu_hard_regno_nregs): New function.
4507 (spu_function_arg_advance): Use it, supplying a valid register number.
4508 (TARGET_HARD_REGNO_NREGS): Redefine.
4509 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
4510 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
4511 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
4512 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
4513 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
4514 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
4515 (CLASS_MAX_NREGS): Remove copy of old documentation.
4516 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
4517 (visium_hard_regno_nregs): New function.
4518 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
4519 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
4520 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4521 xtensa_hard_regno_nregs): New function.
4522 * system.h (HARD_REGNO_NREGS): Poison.
4524 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4526 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
4527 hard_regno_nregs instead of HARD_REGNO_NREGS.
4528 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
4529 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
4530 (c6x_expand_epilogue): Likewise.
4531 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
4532 (frv_read_iacc_argument): Likewise.
4533 * config/sh/sh.c: Include regs.h.
4534 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4535 (regs_used): Likewise.
4536 (output_stack_adjust): Likewise.
4537 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
4538 * expmed.c: Include regs.h.
4539 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4540 * ree.c: Include regs.h.
4541 (combine_reaching_defs): Use hard_regno_nregs instead of
4543 (add_removable_extension): Likewise.
4545 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4547 * regs.h (hard_regno_nregs): Turn into a function.
4548 (end_hard_regno): Update accordingly.
4549 * caller-save.c (setup_save_areas): Likewise.
4550 (save_call_clobbered_regs): Likewise.
4551 (replace_reg_with_saved_mem): Likewise.
4552 (insert_restore): Likewise.
4553 (insert_save): Likewise.
4554 * combine.c (can_change_dest_mode): Likewise.
4555 (move_deaths): Likewise.
4556 (distribute_notes): Likewise.
4557 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
4558 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
4559 (rs6000_split_multireg_move): Likewise.
4560 (rs6000_register_move_cost): Likewise.
4561 (rs6000_memory_move_cost): Likewise.
4562 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
4563 (rs6000_split_multireg_move): Likewise.
4564 (rs6000_register_move_cost): Likewise.
4565 (rs6000_memory_move_cost): Likewise.
4566 * cselib.c (cselib_reset_table): Likewise.
4567 (cselib_lookup_1): Likewise.
4568 * emit-rtl.c (set_mode_and_regno): Likewise.
4569 * function.c (aggregate_value_p): Likewise.
4570 * ira-color.c (setup_profitable_hard_regs): Likewise.
4571 (check_hard_reg_p): Likewise.
4572 (calculate_saved_nregs): Likewise.
4573 (assign_hard_reg): Likewise.
4574 (improve_allocation): Likewise.
4575 (calculate_spill_cost): Likewise.
4576 * ira-emit.c (modify_move_list): Likewise.
4577 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
4578 (ira_hard_reg_in_set_p): Likewise.
4579 * ira.c (setup_reg_mode_hard_regset): Likewise.
4580 (clarify_prohibited_class_mode_regs): Likewise.
4581 (check_allocation): Likewise.
4582 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4583 (lra_setup_reg_renumber): Likewise.
4584 (setup_try_hard_regno_pseudos): Likewise.
4585 (spill_for): Likewise.
4586 (assign_hard_regno): Likewise.
4587 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
4588 * lra-constraints.c (in_class_p): Likewise.
4589 (lra_constraint_offset): Likewise.
4590 (simplify_operand_subreg): Likewise.
4591 (lra_constraints): Likewise.
4592 (split_reg): Likewise.
4593 (split_if_necessary): Likewise.
4594 (invariant_p): Likewise.
4595 (inherit_in_ebb): Likewise.
4596 * lra-lives.c (process_bb_lives): Likewise.
4597 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
4598 (get_hard_regs): Likewise.
4599 (do_remat): Likewise.
4600 * lra-spills.c (assign_spill_hard_regs): Likewise.
4601 * mode-switching.c (create_pre_exit): Likewise.
4602 * postreload.c (reload_combine_recognize_pattern): Likewise.
4603 * recog.c (peep2_find_free_register): Likewise.
4604 * regcprop.c (kill_value_regno): Likewise.
4605 (set_value_regno): Likewise.
4606 (copy_value): Likewise.
4607 (maybe_mode_change): Likewise.
4608 (find_oldest_value_reg): Likewise.
4609 (copyprop_hardreg_forward_1): Likewise.
4610 * regrename.c (check_new_reg_p): Likewise.
4611 (regrename_do_replace): Likewise.
4612 * reload.c (push_reload): Likewise.
4613 (combine_reloads): Likewise.
4614 (find_dummy_reload): Likewise.
4615 (operands_match_p): Likewise.
4616 (find_reloads): Likewise.
4617 (find_equiv_reg): Likewise.
4618 (reload_adjust_reg_for_mode): Likewise.
4619 * reload1.c (count_pseudo): Likewise.
4620 (count_spilled_pseudo): Likewise.
4621 (find_reg): Likewise.
4622 (clear_reload_reg_in_use): Likewise.
4623 (free_for_value_p): Likewise.
4624 (allocate_reload_reg): Likewise.
4625 (choose_reload_regs): Likewise.
4626 (reload_adjust_reg_for_temp): Likewise.
4627 (emit_reload_insns): Likewise.
4628 (delete_output_reload): Likewise.
4629 * rtlanal.c (subreg_get_info): Likewise.
4630 * sched-deps.c (sched_analyze_reg): Likewise.
4631 * sel-sched.c (init_regs_for_mode): Likewise.
4632 (mark_unavailable_hard_regs): Likewise.
4633 (choose_best_reg_1): Likewise.
4634 (verify_target_availability): Likewise.
4635 * valtrack.c (dead_debug_insert_temp): Likewise.
4636 * var-tracking.c (track_loc_p): Likewise.
4637 (emit_note_insn_var_location): Likewise.
4638 * varasm.c (make_decl_rtl): Likewise.
4639 * reginfo.c (choose_hard_reg_mode): Likewise.
4640 (init_reg_modes_target): Refer directly to
4641 this_target_regs->x_hard_regno_nregs.
4643 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4645 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
4646 instead of hard_regno_nregs.
4648 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4650 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
4651 end_hard_regno instead of hard_regno_nregs.
4652 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
4653 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
4654 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
4655 * ira-color.c (improve_allocation): Likewise.
4656 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4657 * lra-lives.c (mark_regno_live): Likewise.
4658 (mark_regno_dead): Likewise.
4659 * lra-remat.c (operand_to_remat): Likewise.
4660 * lra.c (collect_non_operand_hard_regs): Likewise.
4661 * postreload.c (reload_combine_note_store): Likewise.
4662 (move2add_valid_value_p): Likewise.
4663 * reload.c (regno_clobbered_p): Likewise.
4665 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4667 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
4669 * config/v850/v850.c (v850_reorg): Likewise.
4670 * reload.c (refers_to_regno_for_reload_p): Likewise.
4671 (find_equiv_reg): Likewise.
4672 * reload1.c (reload_reg_reaches_end_p): Likewise.
4674 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4676 * caller-save.c (add_used_regs): Use REG_NREGS instead of
4678 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
4679 * config/arm/arm.c (output_move_neon): Likewise.
4680 (arm_attr_length_move_neon): Likewise.
4681 (neon_split_vcombine): Likewise.
4682 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
4683 (c6x_mark_reg_written): Likewise.
4684 (c6x_dwarf_register_span): Likewise.
4685 * config/i386/i386.c (ix86_save_reg): Likewise.
4686 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
4687 (rws_access_reg): Likewise.
4688 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
4689 * mode-switching.c (create_pre_exit): Likewise.
4690 * ree.c (combine_reaching_defs): Likewise.
4691 (add_removable_extension): Likewise.
4692 * regcprop.c (find_oldest_value_reg): Likewise.
4693 (copyprop_hardreg_forward_1): Likewise.
4694 * reload.c (reload_inner_reg_of_subreg): Likewise.
4695 (push_reload): Likewise.
4696 (combine_reloads): Likewise.
4697 (find_dummy_reload): Likewise.
4698 (reload_adjust_reg_for_mode): Likewise.
4699 * reload1.c (find_reload_regs): Likewise.
4700 (forget_old_reloads_1): Likewise.
4701 (reload_reg_free_for_value_p): Likewise.
4702 (reload_adjust_reg_for_temp): Likewise.
4703 (emit_reload_insns): Likewise.
4704 (delete_output_reload): Likewise.
4705 * sel-sched.c (choose_best_reg_1): Likewise.
4706 (choose_best_pseudo_reg): Likewise.
4708 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4709 Alan Hayward <alan.hayward@arm.com>
4710 David Sherwood <david.sherwood@arm.com>
4712 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
4713 * target.def (slow_unaligned_access): New hook.
4714 * targhooks.h (default_slow_unaligned_access): Declare.
4715 * targhooks.c (default_slow_unaligned_access): New function.
4716 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
4717 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
4718 * doc/tm.texi: Regenerate.
4719 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
4720 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
4721 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
4723 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
4724 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
4726 (rs6000_slow_unaligned_access): New function.
4727 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4728 (expand_block_compare): Likewise.
4729 (expand_strn_compare): Likewise.
4730 (rs6000_rtx_costs): Likewise.
4731 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
4732 (riscv_slow_unaligned_access): Likewise.
4733 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
4734 (riscv_slow_unaligned_access_p): ...this and make static.
4735 (riscv_option_override): Update accordingly.
4736 (riscv_slow_unaligned_access): New function.
4737 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4738 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
4739 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4740 (rs6000_slow_unaligned_access): New function.
4741 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4742 (rs6000_rtx_costs): Likewise.
4743 * config/rs6000/rs6000-string.c (expand_block_compare)
4744 (expand_strn_compare): Use targetm.slow_unaligned_access instead
4745 of SLOW_UNALIGNED_ACCESS.
4746 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
4747 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
4748 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
4749 of SLOW_UNALIGNED_ACCESS.
4750 * expmed.c (simple_mem_bitfield_p): Likewise.
4751 * expr.c (alignment_for_piecewise_move): Likewise.
4752 (emit_group_load_1): Likewise.
4753 (emit_group_store): Likewise.
4754 (copy_blkmode_from_reg): Likewise.
4755 (emit_push_insn): Likewise.
4756 (expand_assignment): Likewise.
4757 (store_field): Likewise.
4758 (expand_expr_real_1): Likewise.
4759 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
4760 * lra-constraints.c (simplify_operand_subreg): Likewise.
4761 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4762 * gimple-ssa-store-merging.c: Likewise in block comment at start
4764 * tree-ssa-strlen.c: Include target.h.
4765 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
4766 of SLOW_UNALIGNED_ACCESS.
4767 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
4769 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4771 PR rtl-optimization/82185
4772 * expmed.c (emit_store_flag_int): Only test tem if it has been
4775 2017-09-12 Richard Biener <rguenther@suse.de>
4778 * match.pd ((FTYPE) N CMP CST): Fix typo.
4780 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
4782 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
4784 (mips_near_type_p): Add 'short_call' attribute as a synonym
4786 * doc/extend.texi (short_call): Document new function attribute.
4788 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4791 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
4792 assertion check that in the condition.
4793 (get_atomic_generic_size): Likewise. Before testing if parameter
4794 has pointer type, if it has array type, call for C++
4795 default_conversion to perform array-to-pointer conversion.
4797 2017-09-12 Richard Biener <rguenther@suse.de>
4799 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
4800 for operations we cannot scalarize.
4802 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
4804 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
4805 vectors heap vectors. Clean up comments.
4806 Make visited_bbs a reference.
4807 (profitable_jump_thread_path): Make GC
4808 vectors heap vectors. Clean up comments.
4810 (convert_and_register_jump_thread_path): Make GC vectors heap
4812 (check_subpath_and_update_thread_path): Same. Clean up comments.
4813 Make visited_bbs a reference.
4814 (handle_phi): Abstract common code to to
4815 register_jump_thread_path_if_profitable.
4816 Rename VAR_BB to DEF_BB.
4818 Make GC vectors heap vectors.
4819 Make visited_bbs a reference.
4820 (handle_assignment): Same.
4821 (register_jump_thread_path_if_profitable): New.
4822 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
4824 Make GC vectors heap vectors. Clean up comments.
4825 Make visited_bbs a reference.
4826 (find_jump_threads_backwards): Make visited_bbs live in the stack.
4827 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
4830 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
4833 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
4834 words of E_DImode object are reachable by xtensa_uimm8x4 access.
4836 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
4838 Revert r251800 and r251799.
4840 2017-09-11 Martin Jambor <mjambor@suse.cz>
4843 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
4844 arguments in advance.
4845 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
4846 use it to find predecessor edges.
4847 (naive_outof_ssa): Collect vector of predecessors.
4849 2017-09-08 Jason Merrill <jason@redhat.com>
4851 PR c++/70029 - ICE with ref-qualifier and -flto
4852 * langhooks.h (struct lang_hooks_for_types): Add
4853 copy_lang_qualifiers.
4854 * attribs.c (build_type_attribute_qual_variant): Use it.
4855 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
4857 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
4858 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
4860 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
4863 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
4864 (*mulsi3_sp64): New instruction.
4865 (mulsi3): New expander.
4867 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
4869 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
4871 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4873 * sancov.c: Include memmodel.h.
4875 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
4878 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
4881 2017-09-07 Carl Love <cel@us.ibm.com>
4883 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
4884 the sldi instruction.
4886 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
4888 * sancov.c: Include tm_p.h.
4890 2017-09-07 Jakub Jelinek <jakub@redhat.com>
4893 * output.h (switch_to_other_text_partition): New declaration.
4894 * varasm.c (switch_to_other_text_partition): New function.
4895 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
4896 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
4897 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
4898 to the other text partition before emitting LCL label and switch back
4899 after emitting the word after it.
4901 2017-09-07 Richard Biener <rguenther@suse.de>
4903 * passes.def (pass_split_crit_edges): Remove instance before PRE.
4904 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
4905 critical edges here, after loop init.
4906 (pass_data_pre): Remove PROP_no_crit_edges flags.
4907 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
4908 for valueization of call args to avoid leaking VN_TOP.
4909 (visit_use): Assert we do not visit default defs.
4910 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
4911 Use error_mark_node to more easily detect leaking VN_TOP.
4912 All default-defs are varying, not VN_TOP. Mark them visited.
4913 (run_scc_vn): Make code match comment.
4915 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4917 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
4918 OPTION_MASK_FLOAT128_KEYWORD.
4919 (POWERPC_MASKS): Likewise.
4920 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
4921 support for the -mfloat128-type option, and make -mfloat128
4922 default on PowerPC Linux systems. Define or undefine
4923 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
4924 Define __float128 to be __ieee128 if IEEE 128-bit support is
4925 enabled, or undefine it.
4926 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
4927 Delete defining __FLOAT128_TYPE__.
4928 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
4929 -mfloat128-type option and make -mfloat128 default on PowerPC
4931 (TARGET_FLOAT128_TYPE): Likewise.
4932 (-mfloat128-type): Likewise.
4933 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4934 Delete the -mfloat128-type option and make -mfloat128 default on
4935 PowerPC Linux systems. Always use __ieee128 to be the keyword for
4936 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
4937 128-bit floating point is enabled. Change tests from using
4938 -mfloat128-type to -mfloat128.
4939 (rs6000_mangle_type): Use the correct mangling for the __float128
4940 type even if normal long double is restricted to 64-bits.
4941 (floatn_mode): Enable the _Float128 type by default on VSX Linux
4943 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
4944 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
4945 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
4947 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
4948 documentation for -mfloat128.
4950 2017-09-06 Olivier Hainque <hainque@adacore.com>
4952 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
4954 2017-09-06 Wish Wu <wishwu007@gmail.com>
4955 Jakub Jelinek <jakub@redhat.com>
4957 * asan.c (initialize_sanitizer_builtins): Add
4958 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
4959 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
4960 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
4961 BT_FN_VOID_UINT64_PTR variables.
4962 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
4963 (BT_FN_VOID_UINT16_UINT16): Likewise.
4964 (BT_FN_VOID_UINT32_UINT32): Likewise.
4965 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
4966 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
4967 (BT_FN_VOID_UINT64_PTR): Likewise.
4968 * common.opt (flag_sanitize_coverage): New variable.
4969 (fsanitize-coverage=trace-pc): Remove.
4970 (fsanitize-coverage=): Add.
4971 * flag-types.h (enum sanitize_coverage_code): New enum.
4972 * fold-const.c (fold_range_test): Disable non-short-circuit
4973 optimization if flag_sanitize_coverage.
4974 (fold_truth_andor): Likewise.
4975 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
4976 * opts.c (COVERAGE_SANITIZER_OPT): Define.
4977 (coverage_sanitizer_opts): New array.
4978 (get_closest_sanitizer_option): Add OPTS argument, handle also
4979 OPT_fsanitize_coverage_.
4980 (parse_sanitizer_options): Adjusted to also handle
4981 OPT_fsanitize_coverage_.
4982 (common_handle_option): Add OPT_fsanitize_coverage_.
4983 * sancov.c (instrument_comparison, instrument_switch): New function.
4984 (sancov_pass): Add trace-cmp support.
4985 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
4986 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
4987 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
4988 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
4989 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
4990 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
4991 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
4992 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
4993 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
4994 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
4996 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4998 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4999 error. Only quit immediately if parsing is complete.
5000 (BEGIN): Initialize fatal_err and parse_done.
5001 (begin fpu, end fpu): Check number of arguments.
5002 (begin arch, end arch): Likewise.
5003 (begin cpu, end cpu): Likewise.
5004 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5005 (optalias): Likewise.
5007 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5009 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5010 * config/arm/arm-isa.h: Delete. Move definitions to ...
5011 * arm-cpus.in: ... here. Use new feature and fgroup values.
5012 * config/arm/arm.c (arm_option_override): Use lower case for feature
5014 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5015 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5016 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5018 (print_isa_bits_for): New function.
5019 (gen_isa): New function.
5020 (gen_comm_data): Use print_isa_bits_for.
5021 (define feature): New keyword.
5022 (define fgroup): New keyword.
5023 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
5024 (arm-isa.h): Add rule to generate file.
5025 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5026 case for feature bit names.
5028 2017-09-06 Richard Biener <rguenther@suse.de>
5030 * tree-ssa-pre.c (NECESSARY): Remove.
5031 (create_expression_by_pieces): Do not touch pass-local flags.
5032 (insert_into_preds_of_block): Likewise.
5033 (do_pre_regular_insertion): Likewise.
5034 (eliminate_insert): Likewise.
5035 (eliminate_dom_walker::before_dom_children): Likewise.
5036 (fini_eliminate): Do not look at inserted_exprs.
5037 (mark_operand_necessary): Remove.
5038 (remove_dead_inserted_code): Replace with simple work-list
5039 algorithm based on inserted_exprs and SSA uses.
5040 (pass_pre::execute): Re-order fini_eliminate and
5041 remove_dead_inserted_code.
5043 2017-09-06 Olivier Hainque <hainque@adacore.com>
5045 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
5046 for VxWorks 7. Adjust surrounding comments.
5048 2017-09-06 Richard Biener <rguenther@suse.de>
5050 * gimple-ssa-strength-reduction.c
5051 (find_candidates_dom_walker::before_dom_children): Also allow
5054 2017-09-06 Richard Biener <rguenther@suse.de>
5056 PR tree-optimization/82108
5057 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
5058 for gap in the non-permutation SLP case.
5060 2017-09-06 Martin Jambor <mjambor@suse.cz>
5062 PR tree-optimization/82078
5063 * tree-sra.c (sort_and_splice_var_accesses): Move call to
5064 add_access_to_work_queue...
5065 (build_accesses_from_assign): ...here.
5066 (propagate_all_subaccesses): Make sure racc is the group
5067 representative, if there is one.
5069 2017-09-06 Jakub Jelinek <jakub@redhat.com>
5072 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
5075 2017-09-06 Richard Biener <rguenther@suse.de>
5077 * gimple-ssa-strength-reduction.c
5078 (find_candidates_dom_walker::before_doom_children): Use a
5079 type and not a mode check.
5081 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5084 * config/arm/predicates.md (arm_general_adddi_operand): Create new
5086 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
5088 2017-09-05 Jeff Law <law@redhat.com>
5090 PR tree-optimization/64910
5091 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
5092 cases where we have 3 or more operands.
5094 2017-09-05 Jakub Jelinek <jakub@redhat.com>
5097 * omp-low.c (lower_omp_for): Recompute tree invariant if
5098 gimple_omp_for_initial/final is ADDR_EXPR.
5101 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
5102 into gimple val before gimplification fo the COND_EXPR.
5104 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
5106 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
5107 REGION_COPY argument.
5108 (thread_through_all_blocks): Remove unused argument to
5109 duplicate_thread_path.
5111 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5112 Alan Hayward <alan.hayward@arm.com>
5113 David Sherwood <david.sherwood@arm.com>
5115 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
5116 Take a scalar_mode rather than a machine_mode.
5117 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5118 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
5119 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5120 (aarch64_gen_adjusted_ldpstp): Likewise.
5121 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
5123 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5124 Alan Hayward <alan.hayward@arm.com>
5125 David Sherwood <david.sherwood@arm.com>
5127 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
5128 Take a scalar_int_mode instead of a machine_mode.
5129 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5130 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5131 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5132 (aarch64_simd_attr_length_rglist): Delete.
5133 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
5134 a scalar_int_mode instead of a machine_mode.
5135 (aarch64_add_offset): Likewise.
5136 (aarch64_internal_mov_immediate): Likewise
5137 (aarch64_add_constant_internal): Likewise.
5138 (aarch64_add_constant): Likewise.
5139 (aarch64_movw_imm): Likewise.
5140 (aarch64_rtx_arith_op_extract_p): Likewise.
5141 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5142 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5143 Remove assert that the mode isn't a vector.
5144 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5145 (aarch64_expand_mov_immediate): Update calls after above changes.
5146 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
5147 (aarch64_and_bitmask_imm): Check for scalar integer modes.
5148 (aarch64_move_imm): Likewise.
5149 (aarch64_can_const_movi_rtx_p): Likewise.
5150 (aarch64_strip_extend): Likewise.
5151 (aarch64_extr_rtx_p): Likewise.
5152 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
5153 a CONST_INT when the mode parameter is VOIDmode.
5154 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
5156 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5158 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
5159 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5160 (bitwise_type_for_mode): Update accordingly.
5162 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5164 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
5165 * stor-layout.c (mode_for_size_tree): Likewise.
5166 (mode_for_array): Update accordingly.
5167 (layout_decl): Likewise.
5168 (compute_record_mode): Likewise. Only set the mode once.
5170 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5172 * target.def (get_mask_mode): Change return type to opt_mode.
5174 * doc/tm.texi: Regenerate.
5175 * targhooks.h (default_get_mask_mode): Return an opt_mode.
5176 * targhooks.c (default_get_mask_mode): Likewise.
5177 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5178 * optabs-query.c (can_vec_mask_load_store_p): Update use of
5179 targetm.get_mask_mode.
5180 * tree.c (build_truth_vector_type): Likewise.
5182 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5184 * machmode.h (mode_for_vector): Return an opt_mode.
5185 * stor-layout.c (mode_for_vector): Likewise.
5186 (mode_for_int_vector): Update accordingly.
5187 (layout_type): Likewise.
5188 * config/i386/i386.c (emit_memmov): Likewise.
5189 (ix86_expand_set_or_movmem): Likewise.
5190 (ix86_expand_vector_init): Likewise.
5191 (ix86_get_mask_mode): Likewise.
5192 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
5194 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
5195 * expmed.c (extract_bit_field_1): Likewise.
5196 * expr.c (expand_expr_real_2): Likewise.
5197 * optabs-query.c (can_vec_perm_p): Likewise.
5198 (can_vec_mask_load_store_p): Likewise.
5199 * optabs.c (expand_vec_perm): Likewise.
5200 * targhooks.c (default_get_mask_mode): Likewise.
5201 * tree-vect-stmts.c (vectorizable_store): Likewise.
5202 (vectorizable_load): Likewise.
5203 (get_vectype_for_scalar_type_and_size): Likewise.
5205 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5207 * machmode.h (mode_for_int_vector): New function.
5208 * stor-layout.c (mode_for_int_vector): Likewise.
5209 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
5210 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
5211 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5212 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5213 (s390_expand_vcond): Likewise.
5215 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5217 * machmode.h (opt_machine_mode): New type.
5218 (opt_mode<T>): Allow construction from anything that can be
5220 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
5221 (mode_for_size): Return an opt_machine_mode.
5222 * stor-layout.c (mode_for_size): Likewise.
5223 (mode_for_size_tree): Update call accordingly.
5224 (bitwise_mode_for_mode): Likewise.
5225 (make_fract_type): Likewise.
5226 (make_accum_type): Likewise.
5227 * caller-save.c (replace_reg_with_saved_mem): Update call
5229 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5230 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5231 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5232 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5233 * expmed.c (extract_bit_field_1): Likewise.
5234 * reload.c (get_secondary_mem): Likewise.
5235 * varasm.c (assemble_integer): Likewise.
5236 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
5239 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5241 * machmode.h (decimal_float_mode_for_size): New function.
5242 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
5243 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
5244 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
5245 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
5246 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
5247 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
5249 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5251 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
5252 (get_builtin_sync_mode): Likewise.
5253 (expand_ifn_atomic_compare_exchange): Likewise.
5254 (expand_builtin_atomic_clear): Likewise.
5255 (expand_builtin_atomic_test_and_set): Likewise.
5256 (fold_builtin_atomic_always_lock_free): Likewise.
5257 * calls.c (compute_argument_addresses): Likewise.
5258 (emit_library_call_value_1): Likewise.
5259 (store_one_arg): Likewise.
5260 * combine.c (combine_instructions): Likewise.
5261 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
5262 * config/arm/arm.c (arm_function_value): Likewise.
5263 (aapcs_allocate_return_reg): Likewise.
5264 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
5265 * config/i386/i386.c (construct_container): Likewise.
5266 (ix86_gimplify_va_arg): Likewise.
5267 (ix86_expand_sse_cmp): Likewise.
5268 (emit_memmov): Likewise.
5269 (emit_memset): Likewise.
5270 (expand_small_movmem_or_setmem): Likewise.
5271 (ix86_expand_pextr): Likewise.
5272 (ix86_expand_pinsr): Likewise.
5273 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
5274 * config/microblaze/microblaze.c (microblaze_block_move_straight):
5276 * config/mips/mips.c (mips_function_value_1) Likewise.
5277 (mips_block_move_straight): Likewise.
5278 (mips_expand_ins_as_unaligned_store): Likewise.
5279 * config/powerpcspe/powerpcspe.c
5280 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5281 (rs6000_darwin64_record_arg_flush): Likewise.
5282 * config/rs6000/rs6000.c
5283 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5284 (rs6000_darwin64_record_arg_flush): Likewise.
5285 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
5286 (sparc_function_value_1): Likewise.
5287 * config/spu/spu.c (adjust_operand): Likewise.
5288 (spu_emit_branch_or_set): Likewise.
5289 (arith_immediate_p): Likewise.
5290 * emit-rtl.c (gen_lowpart_common): Likewise.
5291 * expr.c (expand_expr_real_1): Likewise.
5292 * function.c (assign_parm_setup_block): Likewise.
5293 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5294 * reload1.c (alter_reg): Likewise.
5295 * stor-layout.c (mode_for_vector): Likewise.
5296 (layout_type): Likewise.
5298 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5300 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
5301 (spu_convert_move): Likewise.
5302 * lower-subreg.c (resolve_simple_move): Likewise.
5304 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5307 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
5308 define_insn to a define_expand.
5309 (altivec_vsum2sws_direct): New define_insn.
5310 (altivec_vsumsws): Convert from a define_insn to a define_expand.
5312 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5314 * config/arm/arm.c (arm_option_params_internal): Improve setting of
5317 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
5322 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
5323 realign stack if stack alignment needed is less than incoming
5326 2017-09-05 Marek Polacek <polacek@redhat.com>
5329 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
5332 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5334 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
5336 2017-09-05 Richard Biener <rguenther@suse.de>
5338 PR tree-optimization/82084
5339 * fold-const.c (can_native_encode_string_p): Handle wide characters.
5341 2017-09-05 Richard Biener <rguenther@suse.de>
5343 PR tree-optimization/82102
5344 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
5346 2017-09-05 Martin Liska <mliska@suse.cz>
5348 PR tree-optimization/82032
5349 * tree-cfg.c (generate_range_test): New function.
5350 * tree-cfg.h (generate_range_test): Declared here.
5351 * tree-cfgcleanup.c (convert_single_case_switch): New function.
5352 (cleanup_control_expr_graph): Use it.
5353 * tree-switch-conversion.c (try_switch_expansion): Remove
5355 (emit_case_nodes): Use generate_range_test.
5357 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5360 * config/i386/i386.md (*<btsc><mode>_mask): Add
5361 TARGET_USE_BT to insn constraint.
5362 (*btr<mode>_mask): Ditto.
5364 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
5366 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
5367 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
5369 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5372 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
5373 TARGET_NEON and TARGET_IWMMXT.
5374 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
5375 TARGET_NEON and TARGET_IWMMXT.
5376 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
5378 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5380 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
5381 (ix86_rewrite_tls_address): Ditto.
5382 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
5383 (ix86_rewrite_tls_address_1): Ditto.
5384 (ix86_rewrite_tls_address): Ditto.
5385 * config/i386/predicates.md (tls_address_pattern): New predicate.
5386 * config/i386/i386.md (TLS address splitter): New splitter.
5388 2017-09-04 Richard Biener <rguenther@suse.de>
5390 PR tree-optimization/82084
5391 * fold-const.h (can_native_encode_string_p): Declare.
5392 * fold-const.c (can_native_encode_string_p): Factor out from ...
5393 (native_encode_string): ... here.
5394 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
5395 vectorizing stores from constants we later cannot handle.
5397 2017-09-04 Marek Polacek <polacek@redhat.com>
5400 * doc/invoke.texi: Update -Wtautological-compare documentation.
5402 2017-09-04 Jeff Law <law@redhat.com>
5404 PR tree-optimization/64910
5405 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
5406 swap the first and last operand if the last is a constant.
5408 2017-09-04 Marek Polacek <polacek@redhat.com>
5411 * convert.c (do_narrow): When sanitizing signed integer overflows,
5412 bail out for signed types.
5413 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
5415 2017-09-04 Richard Biener <rguenther@suse.de>
5417 PR tree-optimization/82060
5418 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
5419 Move devirtualization after stmt folding and before EH/AB/noreturn
5420 cleanup to get the stmt refs canonicalized. Use a bool instead
5421 of gimple_modified_p since that doesn't work for NOPs. Schedule
5422 NOPs generated by folding for removal.
5424 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5425 Alan Hayward <alan.hayward@arm.com>
5426 David Sherwood <david.sherwood@arm.com>
5428 * coretypes.h (pad_direction): New enum.
5429 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
5430 (FUNCTION_ARG_PADDING): Likewise.
5431 * target.def (function_arg_padding): New hook.
5432 * targhooks.h (default_function_arg_padding): Declare.
5433 * targhooks.c (default_function_arg_padding): New function.
5434 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
5435 (TARGET_FUNCTION_ARG_PADDING): ...this.
5436 * doc/tm.texi: Regenerate.
5437 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
5438 instead of direction.
5439 (compute_argument_addresses): Likewise.
5440 (load_register_parameters): Likewise.
5441 (emit_library_call_value_1): Likewise.
5442 (store_one_arg): Use targetm.calls.function_arg_padding instead
5443 of FUNCTION_ARG_PADDING.
5444 (must_pass_in_stack_var_size_or_pad): Likewise.
5445 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
5446 (emit_group_store): Likewise.
5447 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
5448 instead of FUNCTION_ARG_PADDING.
5449 (emit_push_insn): Likewise, and propagate enum change throughout
5451 * function.h (direction): Delete.
5452 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
5454 * function.c (assign_parm_find_stack_rtl): Likewise.
5455 (assign_parm_setup_block_p): Likewise.
5456 (assign_parm_setup_block): Likewise.
5457 (gimplify_parameters): Likewise.
5458 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
5459 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
5461 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
5462 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5463 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
5464 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
5465 (aarch64_function_arg_padding): ...this new function.
5466 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
5467 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5468 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
5469 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5470 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
5471 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5472 (arm_pad_arg_upward): Replace with...
5473 (arm_function_arg_padding): ...this new function.
5474 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
5476 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
5477 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
5478 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5479 (ia64_hpux_function_arg_padding): Replace with...
5480 (ia64_function_arg_padding): ...this new function. Use pad_direction
5481 instead of direction. Check for TARGET_HPUX.
5482 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
5483 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5484 (iq2000_function_arg_padding): New function.
5485 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
5486 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
5487 (mips_function_arg_padding): ...this new function.
5488 (mips_pad_reg_upward): Update accordingly.
5489 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5490 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
5491 targetm.calls.function_arg_padding.
5492 (FUNCTION_ARG_PADDING): Delete.
5493 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5494 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
5495 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5496 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
5497 (nios2_block_reg_padding): Return pad_direction instead of direction.
5498 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
5499 instead of direction.
5500 (nios2_function_arg_padding): Likewise. Make static.
5501 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5502 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
5503 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
5504 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
5505 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5506 (pa_function_arg_padding): Make static. Return pad_direction instead
5508 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
5509 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5510 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
5511 instead of direction. Use targetm.calls.function_arg_padding.
5512 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
5513 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5514 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
5515 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
5516 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
5518 (function_arg_padding): Rename to...
5519 (rs6000_function_arg_padding): ...this. Make static. Return
5520 pad_direction instead of direction.
5521 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5522 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
5523 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5524 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
5525 instead of direction. Use targetm.calls.function_arg_padding.
5526 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
5527 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5528 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
5529 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
5530 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5531 (function_arg_padding): Rename to...
5532 (rs6000_function_arg_padding): ...this. Make static. Return
5533 pad_direction instead of direction.
5534 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5535 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
5536 * config/s390/s390.c (s390_function_arg_padding): New function.
5537 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5538 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
5539 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
5540 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5541 (function_arg_padding): Rename to...
5542 (sparc_function_arg_padding): ...this. Make static. Return
5543 pad_direction instead of direction.
5544 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
5545 * config/spu/spu.c (spu_function_arg_padding): New function.
5546 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5547 * system.h (FUNCTION_ARG_PADDING): Poison.
5549 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5550 Alan Hayward <alan.hayward@arm.com>
5551 David Sherwood <david.sherwood@arm.com>
5553 * target.def (modes_tieable_p): New hook.
5554 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
5555 (TARGET_MODES_TIEABLE_P): ...this.
5556 * doc/tm.texi.in: Regenerate.
5557 * hooks.h (hook_bool_mode_mode_true): Declare.
5558 * hooks.c (hook_bool_mode_mode_true): New function.
5559 * combine.c (subst): Use targetm.modes_tieable_p instead of
5561 * dse.c (find_shift_sequence): Likewise.
5562 * expmed.c (extract_low_bits): Likewise.
5563 * lower-subreg.c: Include target.h.
5564 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
5566 * rtlanal.c (rtx_cost): Likewise.
5567 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
5568 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
5569 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
5570 (TARGET_MODES_TIEABLE_P): Redefine.
5571 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
5572 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
5573 (TARGET_MODES_TIEABLE_P): Redefine.
5574 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
5575 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
5576 (arc_modes_tieable_p): New function.
5577 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
5578 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
5579 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
5580 (arm_modes_tieable_p): Make static.
5581 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
5582 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
5583 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
5584 (TARGET_MODES_TIEABLE_P): Redefine.
5585 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
5586 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
5587 (TARGET_MODES_TIEABLE_P): Redefine.
5588 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
5589 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
5590 (cr16_modes_tieable_p): New function.
5591 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
5592 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
5593 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
5594 (TRULY_NOOP_TRUNCATION): Update comment.
5595 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
5596 (TRULY_NOOP_TRUNCATION): Update comment.
5597 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
5598 (frv_modes_tieable_p): New function.
5599 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
5600 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
5601 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
5602 (TARGET_MODES_TIEABLE_P): Redefine.
5603 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
5604 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
5605 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
5606 (TARGET_MODES_TIEABLE_P): Redefine.
5607 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
5608 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
5609 (ia64_modes_tieable_p): New function.
5610 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
5611 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
5612 (iq2000_modes_tieable_p): New function.
5613 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
5614 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
5615 (lm32_modes_tieable_p): New function.
5616 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
5617 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
5618 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
5619 (TARGET_MODES_TIEABLE_P): Redefine.
5620 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
5621 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
5622 (m32r_modes_tieable_p): New function.
5623 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
5624 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
5625 (m68k_modes_tieable_p): New function.
5626 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
5627 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
5628 (mcore_modes_tieable_p): New function.
5629 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
5630 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
5632 (TARGET_MODES_TIEABLE_P): Redefine.
5633 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
5634 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
5635 * config/mips/mips.c (mips_modes_tieable_p): Make static.
5636 (TARGET_MODES_TIEABLE_P): Redefine.
5637 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
5638 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
5639 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
5640 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
5641 (mn10300_modes_tieable_p): ...this and make static.
5642 (TARGET_MODES_TIEABLE_P): Redefine.
5643 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
5644 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
5645 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
5646 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
5647 (msp430_modes_tieable_p): Make static.
5648 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
5649 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
5650 (TARGET_MODES_TIEABLE_P): Redefine.
5651 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
5652 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
5653 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
5654 (TARGET_MODES_TIEABLE_P): Redefine.
5655 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
5656 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
5657 * config/pa/pa.c (pa_modes_tieable_p): Make static.
5658 (TARGET_MODES_TIEABLE_P): Redefine.
5659 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
5660 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
5661 (pdp11_modes_tieable_p): New function.
5662 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
5663 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
5664 (rs6000_modes_tieable_p): New function.
5665 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
5666 * config/powerpcspe/powerpcspe.md: Update comment.
5667 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
5668 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
5669 (TARGET_MODES_TIEABLE_P): Redefine.
5670 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
5671 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
5672 (rl78_modes_tieable_p): New function.
5673 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
5674 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
5675 (rs6000_modes_tieable_p): New function.
5676 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
5677 * config/rs6000/rs6000.md: Update comment.
5678 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
5679 * config/rx/rx.c (rx_modes_tieable_p): New function.
5680 (TARGET_MODES_TIEABLE_P): Redefine.
5681 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
5682 * config/s390/s390.c (s390_modes_tieable_p): New function.
5683 (TARGET_MODES_TIEABLE_P): Redefine.
5684 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
5685 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
5686 (sh_modes_tieable_p): New function.
5687 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
5688 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
5689 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
5690 (sparc_modes_tieable_p): Make static.
5691 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
5692 * config/spu/spu.c (spu_modes_tieable_p): New function.
5693 (TARGET_MODES_TIEABLE_P): Redefine.
5694 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
5695 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
5696 (TARGET_MODES_TIEABLE_P): Redefine.
5697 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
5698 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
5699 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
5700 * config/v850/v850.c (v850_modes_tieable_p): New function.
5701 (TARGET_MODES_TIEABLE_P): Redefine.
5702 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
5703 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
5704 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
5705 (visium_modes_tieable_p): New function.
5706 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
5707 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
5708 (xtensa_modes_tieable_p): New function.
5709 * system.h (MODES_TIEABLE_P): Poison.
5711 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5712 Alan Hayward <alan.hayward@arm.com>
5713 David Sherwood <david.sherwood@arm.com>
5715 * target.def (hard_regno_mode_ok): New hook.
5716 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
5717 (TARGET_HARD_REGNO_MODE_OK): ...this.
5718 * doc/tm.texi.in: Regenerate.
5719 * hooks.h (hook_bool_uint_mode_true): Declare.
5720 * hooks.c (hook_bool_uint_mode_true): New function.
5721 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
5723 * genpreds.c (write_insn_preds_c): Add an include of target.h.
5724 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
5725 instead of HARD_REGNO_MODE_OK.
5726 * caller-save.c: Include target.h.
5727 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
5729 * combine.c (can_combine_p): Likewise.
5730 (combinable_i3pat): Likewise.
5731 (can_change_dest_mode): Likewise.
5732 * expr.c (init_expr_target): Likewise.
5733 (convert_move): Likewise.
5734 (convert_modes): Likewise.
5735 * ira.c (setup_prohibited_class_mode_regs): Likewise.
5736 (setup_prohibited_mode_move_regs): Likewise.
5737 * ira.h (target_ira): Likewise.
5738 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5739 * lra-constraints.c (process_alt_operands): Likewise.
5740 (split_reg): Likewise.
5741 * recog.c (peep2_find_free_register): Likewise.
5742 * ree.c (combine_reaching_defs): Likewise.
5743 * regcprop.c (maybe_mode_change): Likewise.
5744 * reginfo.c (init_reg_sets_1): Likewise.
5745 (choose_hard_reg_mode): Likewise.
5746 (simplifiable_subregs): Likewise.
5747 * regrename.c (check_new_reg_p): Likewise.
5748 * reload.c (find_valid_class): Likewise.
5749 (find_valid_class_1): Likewise.
5750 (reload_inner_reg_of_subreg): Likewise.
5751 (push_reload): Likewise.
5752 (combine_reloads): Likewise.
5753 (find_dummy_reload): Likewise.
5754 (find_reloads): Likewise.
5755 * reload1.c (find_reg): Likewise.
5756 (set_reload_reg): Likewise.
5757 (allocate_reload_reg): Likewise.
5758 (choose_reload_regs): Likewise.
5759 (reload_adjust_reg_for_temp): Likewise.
5760 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
5761 (simplify_subreg_regno): Likewise.
5762 * sel-sched.c (init_regs_for_mode): Likewise.
5763 * varasm.c (make_decl_rtl): Likewise.
5764 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
5765 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
5767 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
5768 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
5769 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5770 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
5771 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
5772 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5773 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
5774 (arc_mode_class): Delete.
5775 (HARD_REGNO_MODE_OK): Delete.
5776 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5777 (arc_hard_regno_mode_ok): Rename old array to...
5778 (arc_hard_regno_mode_ok_modes): ...this.
5779 (arc_conditional_register_usage): Update accordingly.
5780 (arc_mode_class): Make static.
5781 (arc_hard_regno_mode_ok): New function.
5782 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
5783 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
5784 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5785 (arm_hard_regno_mode_ok): Make static.
5786 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
5788 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
5789 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
5790 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
5792 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5793 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
5794 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
5795 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
5796 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
5797 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5798 * config/bfin/predicates.md (valid_reg_operand): Use
5799 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
5800 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
5801 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
5802 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5803 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
5804 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
5805 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5806 (cr16_hard_regno_mode_ok): Make static and return a bool.
5807 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
5808 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5809 (cris_hard_regno_mode_ok): New function.
5810 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
5811 (epiphany_mode_class): Delete.
5812 (HARD_REGNO_MODE_OK): Delete.
5813 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
5814 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5815 (hard_regno_mode_ok): Rename to...
5816 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
5817 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
5818 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
5820 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
5821 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
5822 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5823 (frv_hard_regno_mode_ok): Make static and return a bool.
5824 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
5826 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
5827 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
5828 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
5829 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
5831 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5832 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
5833 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
5834 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
5836 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5837 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
5838 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5839 (ia64_hard_regno_mode_ok): New function.
5840 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
5841 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5842 (iq2000_hard_regno_mode_ok): New function.
5843 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
5844 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5845 (lm32_hard_regno_mode_ok): New function.
5846 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
5847 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
5848 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
5849 instead of HARD_REGNO_MODE_OK.
5850 (m32c_hard_regno_ok): Rename to...
5851 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
5852 (m32c_cannot_change_mode_class): Update accordingly.
5853 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5854 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
5855 (m32r_mode_class): Delete.
5856 (HARD_REGNO_MODE_OK): Delete.
5857 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5858 (m32r_hard_regno_mode_ok): Rename to...
5859 (m32r_hard_regno_modes): ...this.
5860 (m32r_mode_class): Make static.
5861 (m32r_hard_regno_mode_ok): New function.
5862 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
5863 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
5864 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5865 (m68k_hard_regno_mode_ok): Make static.
5866 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
5867 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5868 (mcore_hard_regno_mode_ok): New function.
5869 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
5870 (HARD_REGNO_MODE_OK): Delete.
5871 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
5873 (microblaze_hard_regno_mode_ok_p): ...this and make static.
5874 (microblaze_hard_regno_mode_ok): New function.
5875 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5876 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
5877 (mips_hard_regno_mode_ok): Delete.
5878 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
5879 (mips_hard_regno_mode_ok_p): ...this and make static.
5880 (mips_hard_regno_mode_ok_p): Rename to...
5881 (mips_hard_regno_mode_ok_uncached): ...this.
5882 (mips_hard_regno_mode_ok): New function.
5883 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
5884 of HARD_REGNO_MODE_OK.
5885 (mips_option_override): Update after above name changes.
5886 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5887 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
5888 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
5889 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
5890 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
5891 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5892 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
5893 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
5894 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
5895 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5896 (msp430_hard_regno_mode_ok): Make static and return a bool.
5897 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
5898 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
5899 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
5901 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5902 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
5903 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
5904 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
5905 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
5906 (PA_HARD_REGNO_MODE_OK): ...this
5907 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
5908 (PA_HARD_REGNO_MODE_OK): ...this.
5909 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5910 (pa_hard_regno_mode_ok): New function.
5911 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
5912 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5913 (pdp11_hard_regno_mode_ok): New function.
5914 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
5915 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
5917 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
5919 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5920 (rs6000_hard_regno_mode_ok): Rename to...
5921 (rs6000_hard_regno_mode_ok_uncached): ...this.
5922 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5923 (rs6000_hard_regno_mode_ok): New function.
5924 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
5925 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
5926 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
5927 (riscv_hard_regno_mode_ok): ...this and make static.
5928 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5929 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
5930 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
5931 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5932 (rl78_hard_regno_mode_ok): Make static and return bool.
5933 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
5934 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
5936 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
5937 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5938 (rs6000_hard_regno_mode_ok): Rename to...
5939 (rs6000_hard_regno_mode_ok_uncached): ...this.
5940 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5941 (rs6000_hard_regno_mode_ok): New function.
5942 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
5943 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
5944 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5945 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
5946 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
5947 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
5948 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5949 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
5950 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
5951 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5952 (sh_hard_regno_mode_ok): Make static.
5953 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
5954 instead of HARD_REGNO_MODE_OK.
5955 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
5956 (sparc_mode_class): Delete.
5957 (HARD_REGNO_MODE_OK): Delete.
5958 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5959 (hard_regno_mode_classes): Make static.
5960 (sparc_mode_class): Likewise.
5961 (sparc_hard_regno_mode_ok): New function.
5962 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
5963 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
5964 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
5966 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5967 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
5968 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
5969 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
5970 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
5971 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5972 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
5973 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
5974 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5975 (visium_hard_regno_mode_ok): New function.
5976 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
5977 instead of HARD_REGNO_MODE_OK.
5978 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
5979 (HARD_REGNO_MODE_OK): Delete.
5980 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
5981 (xtensa_hard_regno_mode_ok_p): ...this and make static.
5982 (xtensa_option_override): Update accordingly.
5983 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5984 (xtensa_hard_regno_mode_ok): New function.
5985 * system.h (HARD_REGNO_MODE_OK): Poison.
5987 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5988 Alan Hayward <alan.hayward@arm.com>
5989 David Sherwood <david.sherwood@arm.com>
5991 * target.def (hard_regno_call_part_clobbered): New hook.
5992 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
5993 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
5994 * doc/tm.texi: Regenerate.
5995 * hooks.h (hook_bool_uint_mode_false): Declare.
5996 * hooks.c (hook_bool_uint_mode_false): New function.
5997 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5998 * cselib.c (cselib_process_insn): Use
5999 targetm.hard_regno_call_part_clobbered instead of
6000 HARD_REGNO_CALL_PART_CLOBBERED.
6001 * ira-conflicts.c (ira_build_conflicts): Likewise.
6002 * ira-costs.c (ira_tune_allocno_costs): Likewise.
6003 * lra-constraints.c (need_for_call_save_p): Likewise.
6004 * lra-lives.c: Include target.h.
6005 (check_pseudos_live_through_calls): Use
6006 targetm.hard_regno_call_part_clobbered instead of
6007 HARD_REGNO_CALL_PART_CLOBBERED.
6008 * regcprop.c: Include target.h.
6009 (copyprop_hardreg_forward_1): Use
6010 targetm.hard_regno_call_part_clobbered instead of
6011 HARD_REGNO_CALL_PART_CLOBBERED.
6012 * reginfo.c (choose_hard_reg_mode): Likewise.
6013 * regrename.c (check_new_reg_p): Likewise.
6014 * reload.c (find_equiv_reg): Likewise.
6015 * reload1.c (emit_reload_insns): Likewise.
6016 * sched-deps.c (deps_analyze_insn): Likewise.
6017 * sel-sched.c (init_regs_for_mode): Likewise.
6018 (mark_unavailable_hard_regs): Likewise.
6019 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
6020 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6021 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
6023 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6024 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6025 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
6027 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
6029 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6030 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6031 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
6033 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6034 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6035 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
6037 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6038 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
6040 * config/powerpcspe/powerpcspe.c
6041 (rs6000_hard_regno_call_part_clobbered): New function.
6042 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6043 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6044 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
6046 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6047 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6048 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
6050 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6051 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6052 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
6054 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6055 Alan Hayward <alan.hayward@arm.com>
6056 David Sherwood <david.sherwood@arm.com>
6058 * rtl.h (subreg_memory_offset): Declare.
6059 * emit-rtl.c (subreg_memory_offset): New function.
6060 * expmed.c (store_bit_field_1): Use it.
6061 * expr.c (undefined_operand_subword_p): Likewise.
6062 * simplify-rtx.c (simplify_subreg): Likewise.
6064 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
6066 PR rtl-optimization/57448
6069 * optabs.c (expand_atomic_load): Place compiler memory barriers if
6070 using atomic_load pattern.
6071 (expand_atomic_store): Likewise.
6073 2017-09-04 Jakub Jelinek <jakub@redhat.com>
6076 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
6077 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
6078 handling. Use replace_call_with_value with NULL instead of
6079 gsi_replace, unlink_stmt_vdef and release_defs.
6081 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
6084 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
6086 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6089 * rtl.h (emit_library_call_value_1): Declare.
6090 (emit_library_call): Replace declaration with a series of overloads.
6091 Remove the parameter count argument.
6092 (emit_library_call_value): Likewise.
6093 * calls.c (emit_library_call_value_1): Make global. Replace varargs
6094 with an "rtx_mode_t *".
6095 (emit_library_call_value): Delete.
6096 (emit_library_call): Likewise.
6097 * asan.c (asan_emit_stack_protection): Update calls accordingly.
6098 (asan_emit_allocas_unpoison): Likewise.
6099 * builtins.c (expand_builtin_powi): Likewise.
6100 (expand_asan_emit_allocas_unpoison): Likewise.
6101 * cfgexpand.c (expand_main_function): Likewise.
6102 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
6103 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
6104 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
6105 * config/arm/arm.c (arm_trampoline_init): Likewise.
6106 (arm_call_tls_get_addr): Likewise.
6107 (arm_expand_divmod_libfunc): Likewise.
6108 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
6109 (smulsi3_highpart): Likewise.
6110 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
6111 (c6x_expand_compare): Likewise.
6112 (c6x_expand_movmem): Likewise.
6113 * config/frv/frv.c (frv_trampoline_init): Likewise.
6114 * config/i386/i386.c (ix86_trampoline_init): Likewise.
6115 (ix86_expand_divmod_libfunc): Likewise.
6116 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
6117 (ia64_expand_compare): Likewise.
6118 (ia64_profile_hook): Likewise.
6119 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
6120 (nonlocal_goto): Likewise.
6121 (restore_stack_nonlocal): Likewise.
6122 * config/m32r/m32r.c (block_move_call): Likewise.
6123 (m32r_trampoline_init): Likewise.
6124 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
6125 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
6126 (m68k_call_m68k_read_tp): Likewise.
6127 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
6128 (microblaze_expand_divide): Likewise.
6129 * config/mips/mips.h (mips_args): Likewise.
6130 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
6131 (MIPS_ICACHE_SYNC): Likewise.
6132 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
6133 (nios2_trampoline_init): Likewise.
6134 * config/pa/pa.c (hppa_tls_call): Likewise.
6135 (pa_trampoline_init): Likewise.
6136 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
6137 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
6138 (expand_strn_compare): Likewise.
6139 (rs6000_generate_compare): Likewise.
6140 (rs6000_expand_float128_convert): Likewise.
6141 (output_profile_hook): Likewise.
6142 (rs6000_trampoline_init): Likewise.
6143 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
6144 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
6145 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
6146 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
6147 (rs6000_generate_compare): Likewise.
6148 (rs6000_expand_float128_convert): Likewise.
6149 (output_profile_hook): Likewise.
6150 (rs6000_trampoline_init): Likewise.
6151 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
6152 * config/sh/sh.c (sh_trampoline_init): Likewise.
6153 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
6154 (sparc_emit_float_lib_cmp): Likewise.
6155 (sparc32_initialize_trampoline): Likewise.
6156 (sparc64_initialize_trampoline): Likewise.
6157 (sparc_profile_hook): Likewise.
6158 * config/spu/spu.c (ea_load_store): Likewise.
6159 * config/spu/spu.md (floatunssidf2): Likewise.
6160 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
6161 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
6162 * config/visium/visium.c (expand_block_move_4): Likewise.
6163 (expand_block_move_2): Likewise.
6164 (expand_block_move_1): Likewise.
6165 (expand_block_set_4): Likewise.
6166 (expand_block_set_2): Likewise.
6167 (expand_block_set_1): Likewise.
6168 (visium_trampoline_init): Likewise.
6169 (visium_profile_hook): Likewise.
6170 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
6171 (xtensa_setup_frame_addresses): Likewise.
6172 (xtensa_trampoline_init): Likewise.
6173 * except.c (sjlj_emit_function_enter): Likewise.
6174 (sjlj_emit_function_exit): Likewise.
6175 * explow.c (allocate_dynamic_stack_space): Likewise.
6176 (probe_stack_range): Likewise.
6177 * expr.c (convert_mode_scalar): Likewise.
6178 * optabs.c (expand_binop): Likewise.
6179 (expand_twoval_binop_libfunc): Likewise.
6180 (expand_unop): Likewise.
6181 (prepare_cmp_insn): Likewise.
6182 (prepare_float_lib_cmp): Likewise.
6183 (expand_float): Likewise.
6184 (expand_fix): Likewise.
6185 (expand_fixed_convert): Likewise.
6186 (maybe_emit_sync_lock_test_and_set): Likewise.
6187 (expand_atomic_compare_and_swap): Likewise.
6188 (expand_mem_thread_fence): Likewise.
6189 (expand_atomic_fetch_op): Likewise.
6191 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
6193 * doc/generic.texi (OpenACC): Adjust URL.
6194 * doc/invoke.texi (C Dialect Options): Ditto.
6196 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
6198 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
6199 predicate for operand 1. Add (m,<S>) constraint.
6200 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
6201 Prevent memory operand 1 with register operand 2.
6203 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
6205 PR rtl-optimization/82024
6206 * combine.c (try_combine): If the combination result is a PARALLEL,
6207 and we only need to retain the SET in there that would be placed
6208 at I2, check that we can place that at I3 instead, before doing so.
6210 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6213 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
6215 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
6216 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
6219 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
6220 Jeff Law <law@redhat.com>
6222 * varasm.c (bss_initializer_p): Do not put constants into .bss
6223 (categorize_decl_for_section): Handle bss_initializer_p returning
6224 false when DECL_INITIAL is NULL.
6226 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6229 * config/s390/s390.c (s390_can_inline_p): New function.
6231 2017-09-01 Jeff Law <law@redhat.com>
6233 PR tree-optimization/82052
6234 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
6235 Always initialize the returned slot after a hash table miss
6236 when INSERT is true.
6238 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
6240 * config/s390/s390.md (mem_signal_fence): Remove.
6241 * doc/md.texi (mem_signal_fence): Remove.
6242 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
6244 * target-insns.def (mem_signal_fence): Remove.
6246 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6249 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
6252 * asan.c (create_odr_indicator): Strip name encoding from assembler
6253 name before appending it after __odr_asan_.
6255 2017-09-01 Martin Liska <mliska@suse.cz>
6257 PR tree-optimization/82059
6258 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
6259 frequency only when an edge is redirected.
6261 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6263 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
6264 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
6265 (arc_conditional_register_usage): Remove ARC600 lp_count
6267 (arc_file_start): Emit Tag_ARC_CPU_variation.
6268 (arc_can_use_doloop_p): New conditions to use ZOLs.
6269 (hwloop_fail): New function.
6270 (hwloop_optimize): Likewise.
6271 (hwloop_pattern_reg): Likewise.
6272 (arc_doloop_hooks): New struct, to be used with reorg_loops.
6273 (arc_reorg_loops): New function, calls reorg_loops.
6274 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
6275 (arc600_corereg_hazard): Remove ZOL checking, case handled by
6277 (arc_loop_hazard): Remove function, functionality moved into
6279 (arc_hazard): Remove arc_loop_hazard call.
6280 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
6281 into hwloop_optimize.
6282 (arc_label_align): Remove ZOL handling.
6283 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
6284 * config/arc/arc.md (doloop_begin): Remove pattern.
6285 (doloop_begin_i): Likewise.
6286 (doloop_end_i): Likewise.
6287 (doloop_fallback): Likewise.
6288 (doloop_fallback_m): Likewise.
6289 (doloop_end): Reimplement expand.
6290 (arc_lp): New pattern for LP instruction.
6291 (loop_end): New pattern.
6292 (loop_fail): Likewise.
6293 (decrement_and_branch_until_zero): Likewise.
6294 * config/arc/arc.opt (mlpc-width): New option.
6295 * doc/invoke.texi (mlpc-width): Document option.
6297 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6299 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
6300 (arc_ccfsm_advance): Fix checking for delay slots.
6301 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
6303 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6305 * config/arc/arc.md (movqi_insn): Add stores to save constant long
6307 (movhi_insn): Update store instruction constraint which are saving
6308 6-bit short immediates.
6309 (movsi_insn): Consider also short scaled load operations.
6310 (zero_extendhisi2_i): Use Usd constraint instead of T.
6311 (extendhisi2_i): Add q constraint.
6312 (arc_clzsi2): Add type and length attributes.
6313 (arc_ctzsi2): Likewise.
6314 * config/arc/constraints.md (Usc): Update constraint, the
6315 assembler can parse two relocations for a single instruction.
6317 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6319 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
6320 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
6322 2017-08-31 Olivier Hainque <hainque@adacore.com>
6324 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
6325 match as powerpc-wrs-vxworks*.
6327 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
6329 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
6330 register constraint for by-element operand.
6331 (aarch64_mls_elt_merge<mode>): Likewise.
6333 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6335 * config/arc/arc.c (arc_can_follow_jump): Check for short
6338 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6340 * config.gcc: Use g.opt for arc.
6341 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
6342 functionality moved to ...
6343 (legitimate_scaled_address_p): New function, ...here.
6344 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
6345 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
6346 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
6348 (arc_override_options): Handle G option.
6349 (arc_output_pic_addr_const): Correct function definition.
6350 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
6351 (arc_decl_anon_ns_mem_p): Delete.
6352 (arc_in_small_data_p): Overhaul this function to take into
6353 consideration the value given via G option.
6354 (arc_rewrite_small_data_1): Renamed and corrected old
6355 arc_rewrite_small_data function.
6356 (arc_rewrite_small_data): New function.
6357 (small_data_pattern): Don't use pic_offset_table_rtx.
6358 * config/arc/arc.h (CC1_SPEC): Recognize G option.
6359 * config/arc/simdext.md (movmisalignv2hi): Use
6360 prepare_move_operands function.
6362 (movmisalign*): Likewise.
6363 * doc/invoke.texi (ARC options): Document -G option.
6365 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6367 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
6369 * config/arc/arc.c (arc_print_operand): Output scalled address for
6370 sdata whenever is possible.
6371 (arc_in_small_data_p): Allow sdata for 64bit datum when double
6372 load/stores are available.
6373 (compact_sda_memory_operand): Check for the alignment required by
6374 code density instructions.
6375 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
6377 * config/arc/constraints.md (Usd): Update constraint.
6378 (Us0): New constraint.
6379 (Usc): Update constraint.
6381 2017-08-31 Richard Biener <rguenther@suse.de>
6384 * dwarf2out.c (dwarf2out_early_global_decl): Process each
6387 2017-08-31 Tamar Christina <tamar.christina@arm.com>
6389 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
6390 Resize type_signature.
6392 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6393 Alan Hayward <alan.hayward@arm.com>
6394 David Sherwood <david.sherwood@arm.com>
6396 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
6397 subregs whose inner modes can be stored in GPRs.
6398 (aarch64_classify_index): Likewise.
6400 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6401 Alan Hayward <alan.hayward@arm.com>
6402 David Sherwood <david.sherwood@arm.com>
6404 * config/aarch64/iterators.md (V_cmp_result): Rename to...
6405 (V_INT_EQUIV): ...this.
6406 (v_cmp_result): Rename to...
6407 (v_int_equiv): ...this.
6408 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
6409 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
6410 (copysign<mode>3): Likewise.
6411 (aarch64_simd_bsl<mode>_internal): Likewise.
6412 (aarch64_simd_bsl<mode>): Likewise.
6413 (vec_cmp<mode><mode>): Likewise.
6414 (vcond<mode><mode>): Likewise.
6415 (vcond<v_cmp_mixed><mode>): Likewise.
6416 (vcondu<mode><v_cmp_mixed>): Likewise.
6417 (aarch64_cm<optab><mode>): Likewise.
6418 (aarch64_cmtst<mode>): Likewise.
6419 (aarch64_fac<optab><mode>): Likewise.
6420 (vec_perm_const<mode>): Likewise.
6421 (vcond_mask_<mode><v_cmp_result>): Rename to...
6422 (vcond_mask_<mode><v_int_equiv>): ...this.
6423 (vec_cmp<mode><v_cmp_result>): Rename to...
6424 (vec_cmp<mode><v_int_equiv>): ...this.
6426 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6427 Alan Hayward <alan.hayward@arm.com>
6428 David Sherwood <david.sherwood@arm.com>
6430 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
6432 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
6433 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
6434 (UNSPEC_LD4_DREG): New unspecs.
6435 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
6436 (aarch64_ld2<mode>_dreg_be): Replace with...
6437 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
6439 (aarch64_ld3<mode>_dreg_le)
6440 (aarch64_ld3<mode>_dreg_be): Replace with...
6441 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
6443 (aarch64_ld4<mode>_dreg_le)
6444 (aarch64_ld4<mode>_dreg_be): Replace with...
6445 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
6448 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6450 PR tree-optimization/81987
6451 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
6452 insert an initializer in a location not dominated by the stride
6455 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6457 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
6458 on the entire header of the finally block in the fallthru case.
6460 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6462 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
6464 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
6466 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
6467 rs6000_emit_move_from_cr and call renamed function.
6468 (rs6000_emit_prologue): Call renamed functions.
6469 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
6470 movesi_from_cr, remove volatile CRs.
6472 2017-08-30 Jon Beniston <jon@beniston.com>
6473 Richard Biener <rguenther@suse.de>
6475 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
6476 of VECTOR_MODE_P check.
6477 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
6478 element vector types.
6480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6482 * df.h (df_read_modify_subreg_p): Remove in favor of...
6483 * rtl.h (read_modify_subreg_p): ...this new function. Take a
6484 const_rtx instead of an rtx.
6485 * cprop.c (local_cprop_find_used_regs): Update accordingly.
6486 * df-problems.c (df_word_lr_mark_ref): Likewise.
6487 * ira-lives.c (mark_pseudo_reg_live): Likewise.
6488 (mark_pseudo_reg_dead): Likewise.
6489 (mark_ref_dead): Likewise.
6490 * reginfo.c (init_subregs_of_mode): Likewise.
6491 * sched-deps.c (sched_analyze_1): Likewise.
6492 * df-scan.c (df_def_record_1): Likewise.
6493 (df_uses_record): Likewise.
6494 (df_read_modify_subreg_p): Remove in favor of...
6495 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
6496 const_rtx instead of an rtx.
6498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6499 Alan Hayward <alan.hayward@arm.com>
6500 David Sherwood <david.sherwood@arm.com>
6502 * rtl.h (partial_subreg_p): New function.
6503 * caller-save.c (save_call_clobbered_regs): Use it.
6504 * calls.c (expand_call): Likewise.
6505 * combine.c (combinable_i3pat): Likewise.
6506 (simplify_set): Likewise.
6507 (make_extraction): Likewise.
6508 (make_compound_operation_int): Likewise.
6509 (gen_lowpart_or_truncate): Likewise.
6510 (force_to_mode): Likewise.
6511 (make_field_assignment): Likewise.
6512 (reg_truncated_to_mode): Likewise.
6513 (record_truncated_value): Likewise.
6514 (move_deaths): Likewise.
6515 * cse.c (record_jump_cond): Likewise.
6516 (cse_insn): Likewise.
6517 * cselib.c (cselib_lookup_1): Likewise.
6518 * expmed.c (extract_bit_field_using_extv): Likewise.
6519 * function.c (assign_parm_setup_reg): Likewise.
6520 * ifcvt.c (noce_convert_multiple_sets): Likewise.
6521 * ira-build.c (create_insn_allocnos): Likewise.
6522 * lra-coalesce.c (merge_pseudos): Likewise.
6523 * lra-constraints.c (match_reload): Likewise.
6524 (simplify_operand_subreg): Likewise.
6525 (curr_insn_transform): Likewise.
6526 * lra-lives.c (process_bb_lives): Likewise.
6527 * lra.c (new_insn_reg): Likewise.
6528 (lra_substitute_pseudo): Likewise.
6529 * regcprop.c (mode_change_ok): Likewise.
6530 (maybe_mode_change): Likewise.
6531 (copyprop_hardreg_forward_1): Likewise.
6532 * reload.c (push_reload): Likewise.
6533 (find_reloads): Likewise.
6534 (find_reloads_subreg_address): Likewise.
6535 * reload1.c (alter_reg): Likewise.
6536 (eliminate_regs_1): Likewise.
6537 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6539 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
6541 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
6542 back to if statements, including unpack.
6544 2017-08-30 Martin Liska <mliska@suse.cz>
6547 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
6549 (func_checker::compare_asm_inputs_outputs): ... this function.
6550 (func_checker::compare_gimple_asm): Use the function to compare
6551 also ASM constrains.
6552 * ipa-icf-gimple.h: Rename the function.
6554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6555 Alan Hayward <alan.hayward@arm.com>
6556 David Sherwood <david.sherwood@arm.com>
6558 * coretypes.h (complex_mode): New type.
6559 * gdbhooks.py (build_pretty_printer): Handle it.
6560 * machmode.h (complex_mode): New class.
6561 (complex_mode::includes_p): New function.
6562 (is_complex_int_mode): Likewise.
6563 (is_complex_float_mode): Likewise.
6564 * genmodes.c (get_mode_class): Handle complex mode classes.
6565 * function.c (expand_function_end): Use is_complex_int_mode.
6567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6568 Alan Hayward <alan.hayward@arm.com>
6569 David Sherwood <david.sherwood@arm.com>
6571 * coretypes.h (scalar_mode_pod): New typedef.
6572 * gdbhooks.py (build_pretty_printer): Handle it.
6573 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
6574 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
6575 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
6576 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
6579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6580 Alan Hayward <alan.hayward@arm.com>
6581 David Sherwood <david.sherwood@arm.com>
6583 * machmode.h (mode_for_vector): Take a scalar_mode instead
6585 * stor-layout.c (mode_for_vector): Likewise.
6586 * explow.c (promote_mode): Use as_a <scalar_mode>.
6587 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
6589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6590 Alan Hayward <alan.hayward@arm.com>
6591 David Sherwood <david.sherwood@arm.com>
6593 * target.def (preferred_simd_mode): Take a scalar_mode
6594 instead of a machine_mode.
6595 * targhooks.h (default_preferred_simd_mode): Likewise.
6596 * targhooks.c (default_preferred_simd_mode): Likewise.
6597 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
6598 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
6599 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
6600 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
6601 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
6602 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
6603 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
6604 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
6606 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
6607 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
6608 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
6609 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
6610 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
6611 * doc/tm.texi: Regenerate.
6612 * optabs-query.c (can_vec_mask_load_store_p): Return false for
6615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6616 Alan Hayward <alan.hayward@arm.com>
6617 David Sherwood <david.sherwood@arm.com>
6619 * target.def (scalar_mode_supported_p): Take a scalar_mode
6620 instead of a machine_mode.
6621 * targhooks.h (default_scalar_mode_supported_p): Likewise.
6622 * targhooks.c (default_scalar_mode_supported_p): Likewise.
6623 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
6624 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6625 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
6626 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
6627 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
6628 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
6629 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
6630 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
6631 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
6632 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
6633 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
6634 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
6636 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
6637 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
6638 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6639 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6640 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
6642 * doc/tm.texi: Regenerate.
6644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6645 Alan Hayward <alan.hayward@arm.com>
6646 David Sherwood <david.sherwood@arm.com>
6648 * coretypes.h (opt_scalar_mode): New typedef.
6649 * gdbhooks.py (build_pretty_printers): Handle it.
6650 * machmode.h (mode_iterator::get_2xwider): Add overload for
6652 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
6654 * expr.c (convert_mode_scalar): Likewise.
6655 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6656 * optabs.c (expand_float): Likewise.
6657 (expand_fix): Likewise.
6658 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6661 Alan Hayward <alan.hayward@arm.com>
6662 David Sherwood <david.sherwood@arm.com>
6664 * optabs.c (expand_float): Explicitly check for scalars before
6665 using a branching expansion.
6666 (expand_fix): Likewise.
6668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6669 Alan Hayward <alan.hayward@arm.com>
6670 David Sherwood <david.sherwood@arm.com>
6672 * expr.c (convert_mode): Split scalar handling out into...
6673 (convert_mode_scalar): ...this new function. Treat the modes
6676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6677 Alan Hayward <alan.hayward@arm.com>
6678 David Sherwood <david.sherwood@arm.com>
6680 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
6682 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
6684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6685 Alan Hayward <alan.hayward@arm.com>
6686 David Sherwood <david.sherwood@arm.com>
6688 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
6689 rather than a machine_mode.
6690 (fixed_from_string): Likewise.
6691 (fixed_convert): Likewise.
6692 (fixed_convert_from_int): Likewise.
6693 (fixed_convert_from_real): Likewise.
6694 (real_convert_from_fixed): Likewise.
6695 * fixed-value.c (fixed_from_double_int): Likewise.
6696 (fixed_from_string): Likewise.
6697 (fixed_convert): Likewise.
6698 (fixed_convert_from_int): Likewise.
6699 (fixed_convert_from_real): Likewise.
6700 (real_convert_from_fixed): Likewise.
6701 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
6703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6704 Alan Hayward <alan.hayward@arm.com>
6705 David Sherwood <david.sherwood@arm.com>
6707 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
6708 of separate mode class checks. Do not allow vector modes here.
6709 (immed_wide_int_const): Use as_a <scalar_mode>.
6710 * explow.c (trunc_int_for_mode): Likewise.
6711 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
6712 (wi::shwi): Likewise.
6713 (wi::min_value): Likewise.
6714 (wi::max_value): Likewise.
6715 * dwarf2out.c (loc_descriptor): Likewise.
6716 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
6719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6720 Alan Hayward <alan.hayward@arm.com>
6721 David Sherwood <david.sherwood@arm.com>
6723 * tree.h (SCALAR_TYPE_MODE): New macro.
6724 * expr.c (expand_expr_addr_expr_1): Use it.
6725 (expand_expr_real_2): Likewise.
6726 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
6727 (fold_convert_const_fixed_from_int): Likewise.
6728 (fold_convert_const_fixed_from_real): Likewise.
6729 (native_encode_fixed): Likewise
6730 (native_encode_complex): Likewise
6731 (native_encode_vector): Likewise.
6732 (native_interpret_fixed): Likewise.
6733 (native_interpret_real): Likewise.
6734 (native_interpret_complex): Likewise.
6735 (native_interpret_vector): Likewise.
6736 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
6737 (simd_clone_adjust_argument_types): Likewise.
6738 (simd_clone_init_simd_arrays): Likewise.
6739 (simd_clone_adjust): Likewise.
6740 * stor-layout.c (layout_type): Likewise.
6741 * tree.c (build_minus_one_cst): Likewise.
6742 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6743 * tree-inline.c (estimate_move_cost): Likewise.
6744 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
6745 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
6746 (vectorizable_reduction): Likewise.
6747 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
6748 (vect_recog_mixed_size_cond_pattern): Likewise.
6749 (check_bool_pattern): Likewise.
6750 (adjust_bool_pattern): Likewise.
6751 (search_type_for_mask_1): Likewise.
6752 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
6753 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6754 (vectorizable_load): Likewise.
6755 (vectorizable_store): Likewise.
6756 * ubsan.c (ubsan_encode_value): Likewise.
6757 * varasm.c (output_constant): Likewise.
6759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6760 Alan Hayward <alan.hayward@arm.com>
6761 David Sherwood <david.sherwood@arm.com>
6763 * coretypes.h (scalar_mode): New class.
6764 * machmode.h (scalar_mode): Likewise.
6765 (scalar_mode::includes_p): New function.
6766 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
6767 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
6768 * genmodes.c (get_mode_class): Handle remaining scalar modes.
6769 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
6770 * expmed.c (store_bit_field_1): Likewise.
6771 (extract_bit_field_1): Likewise.
6772 * expr.c (write_complex_part): Likewise.
6773 (read_complex_part): Likewise.
6774 (emit_move_complex_push): Likewise.
6775 (expand_expr_real_2): Likewise.
6776 * function.c (assign_parm_setup_reg): Likewise.
6777 (assign_parms_unsplit_complex): Likewise.
6778 * optabs.c (expand_binop): Likewise.
6779 * rtlanal.c (subreg_get_info): Likewise.
6780 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6781 * varasm.c (output_constant_pool_2): Likewise.
6783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6784 Alan Hayward <alan.hayward@arm.com>
6785 David Sherwood <david.sherwood@arm.com>
6787 * expmed.c (extract_high_half): Use scalar_int_mode and remove
6789 (expmed_mult_highpart_optab): Likewise.
6790 (expmed_mult_highpart): Likewise.
6792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6793 Alan Hayward <alan.hayward@arm.com>
6794 David Sherwood <david.sherwood@arm.com>
6796 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
6797 instead of a machine_mode.
6798 (builtin_memset_read_str): Likewise.
6799 * builtins.c (c_readstr): Likewise.
6800 (builtin_memcpy_read_str): Likewise.
6801 (builtin_strncpy_read_str): Likewise.
6802 (builtin_memset_read_str): Likewise.
6803 (builtin_memset_gen_str): Likewise.
6804 (expand_builtin_signbit): Use scalar_int_mode for local variables.
6805 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
6806 instead of a machine_mode.
6807 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
6809 (make_extraction): Likewise.
6810 (try_widen_shift_mode): Take and return scalar_int_modes instead
6812 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
6813 a scalar_int_mode instead of a machine_mode.
6814 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
6815 (avr_addr_space_pointer_mode): Likewise.
6816 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
6817 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
6818 (msp430_unwind_word_mode): Likewise.
6819 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
6820 (spu_addr_space_pointer_mode): Likewise.
6821 (spu_addr_space_address_mode): Likewise.
6822 (spu_libgcc_cmp_return_mode): Likewise.
6823 (spu_libgcc_shift_count_mode): Likewise.
6824 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
6825 (rl78_addr_space_pointer_mode): Likewise.
6826 (fl78_unwind_word_mode): Likewise.
6827 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
6829 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
6830 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
6831 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
6832 (mips_valid_pointer_mode): Likewise.
6833 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
6834 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
6835 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
6837 (ft32_addr_space_address_mode): Likewise.
6838 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
6839 scalar_int_mode instead of a machine_mode.
6840 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
6842 (m32c_addr_space_address_mode): Likewise.
6843 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
6844 (rs6000_eh_return_filter_mode): Likewise.
6845 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
6846 (rs6000_eh_return_filter_mode): Likewise.
6847 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
6848 (s390_libgcc_shift_count_mode): Likewise.
6849 (s390_unwind_word_mode): Likewise.
6850 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
6852 * target.def (mode_rep_extended): Likewise.
6853 (valid_pointer_mode): Likewise.
6854 (addr_space.valid_pointer_mode): Likewise.
6855 (eh_return_filter_mode): Return a scalar_int_mode rather than
6857 (libgcc_cmp_return_mode): Likewise.
6858 (libgcc_shift_count_mode): Likewise.
6859 (unwind_word_mode): Likewise.
6860 (addr_space.pointer_mode): Likewise.
6861 (addr_space.address_mode): Likewise.
6862 * doc/tm.texi: Regenerate.
6863 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
6865 (do_jump): Use scalar_int_mode for local variables.
6866 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
6867 rather than a machine_mode.
6868 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
6869 (scompare_loc_descriptor_wide): Likewise.
6870 (scompare_loc_descriptor_narrow): Likewise.
6871 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
6873 * except.c (sjlj_emit_dispatch_table): Likewise.
6874 (expand_builtin_eh_copy_values): Likewise.
6875 * explow.c (convert_memory_address_addr_space_1): Likewise.
6876 Take a scalar_int_mode rather than a machine_mode.
6877 (convert_memory_address_addr_space): Take a scalar_int_mode rather
6878 than a machine_mode.
6879 (memory_address_addr_space): Use scalar_int_mode for local variables.
6880 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
6881 rather than a machine_mode.
6882 * expmed.c (mask_rtx): Likewise.
6883 (init_expmed_one_conv): Likewise.
6884 (expand_mult_highpart_adjust): Likewise.
6885 (extract_high_half): Likewise.
6886 (expmed_mult_highpart_optab): Likewise.
6887 (expmed_mult_highpart): Likewise.
6888 (expand_smod_pow2): Likewise.
6889 (expand_sdiv_pow2): Likewise.
6890 (emit_store_flag_int): Likewise.
6891 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
6893 (extract_low_bits): Likewise.
6894 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
6896 * expr.c (pieces_addr::adjust): Likewise.
6897 (can_store_by_pieces): Likewise.
6898 (store_by_pieces): Likewise.
6899 (clear_by_pieces_1): Likewise.
6900 (expand_expr_addr_expr_1): Likewise.
6901 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
6902 (expand_expr_real_1): Likewise.
6903 (try_casesi): Likewise.
6904 * final.c (shorten_branches): Likewise.
6905 * fold-const.c (fold_convert_const_int_from_fixed): Change the
6906 type of "mode" to machine_mode.
6907 * internal-fn.c (expand_arith_overflow_result_store): Take a
6908 scalar_int_mode rather than a machine_mode.
6909 (expand_mul_overflow): Use scalar_int_mode for local variables.
6910 * loop-doloop.c (doloop_modify): Likewise.
6911 (doloop_optimize): Likewise.
6912 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
6913 than a machine_mode.
6914 (expand_doubleword_shift_condmove): Likewise.
6915 (expand_doubleword_shift): Likewise.
6916 (expand_doubleword_clz): Likewise.
6917 (expand_doubleword_popcount): Likewise.
6918 (expand_doubleword_parity): Likewise.
6919 (expand_absneg_bit): Use scalar_int_mode for local variables.
6920 (prepare_float_lib_cmp): Likewise.
6921 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
6922 rather than a machine_mode.
6923 (convert_memory_address_addr_space): Likewise.
6924 (get_mode_bounds): Likewise.
6925 (get_address_mode): Return a scalar_int_mode rather than a
6927 * rtlanal.c (get_address_mode): Likewise.
6928 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
6929 than a machine_mode.
6930 * targhooks.c (default_mode_rep_extended): Likewise.
6931 (default_valid_pointer_mode): Likewise.
6932 (default_addr_space_valid_pointer_mode): Likewise.
6933 (default_eh_return_filter_mode): Return a scalar_int_mode rather
6934 than a machine_mode.
6935 (default_libgcc_cmp_return_mode): Likewise.
6936 (default_libgcc_shift_count_mode): Likewise.
6937 (default_unwind_word_mode): Likewise.
6938 (default_addr_space_pointer_mode): Likewise.
6939 (default_addr_space_address_mode): Likewise.
6940 * targhooks.h (default_eh_return_filter_mode): Likewise.
6941 (default_libgcc_cmp_return_mode): Likewise.
6942 (default_libgcc_shift_count_mode): Likewise.
6943 (default_unwind_word_mode): Likewise.
6944 (default_addr_space_pointer_mode): Likewise.
6945 (default_addr_space_address_mode): Likewise.
6946 (default_mode_rep_extended): Take a scalar_int_mode rather than
6948 (default_valid_pointer_mode): Likewise.
6949 (default_addr_space_valid_pointer_mode): Likewise.
6950 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
6952 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
6953 rather than a machine_mode.
6954 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
6955 for local variables.
6956 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6957 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
6958 than a machine_mode.
6960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6961 Alan Hayward <alan.hayward@arm.com>
6962 David Sherwood <david.sherwood@arm.com>
6964 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
6965 the mode argument to scalar_int_mode.
6966 (do_jump_by_parts_zero_rtx): Likewise.
6967 (do_jump_by_parts_equality_rtx): Likewise.
6968 (do_jump_by_parts_greater): Take a mode argument.
6969 (do_jump_by_parts_equality): Likewise.
6970 (do_jump_1): Update calls accordingly.
6972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6973 Alan Hayward <alan.hayward@arm.com>
6974 David Sherwood <david.sherwood@arm.com>
6976 * is-a.h (safe_dyn_cast): New function.
6977 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
6978 (jump_table_for_label): Likewise.
6979 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
6980 instead of an rtx_insn *.
6981 (shorten_branches): Use dyn_cast instead of LABEL_P and
6982 JUMP_TABLE_DATA_P. Use jump_table_for_label and
6983 rtx_jump_table_data::get_data_mode.
6984 (final_scan_insn): Likewise.
6986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6987 Alan Hayward <alan.hayward@arm.com>
6988 David Sherwood <david.sherwood@arm.com>
6990 * combine.c (try_combine): Use is_a <scalar_int_mode> when
6991 trying to combine a full-register integer set with a subreg
6994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6995 Alan Hayward <alan.hayward@arm.com>
6996 David Sherwood <david.sherwood@arm.com>
6998 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
6999 that is always either address_mode or pointer_mode.
7001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7002 Alan Hayward <alan.hayward@arm.com>
7003 David Sherwood <david.sherwood@arm.com>
7005 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
7006 when the two are known to be equal.
7008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7009 Alan Hayward <alan.hayward@arm.com>
7010 David Sherwood <david.sherwood@arm.com>
7012 * simplify-rtx.c (simplify_const_unary_operation): Use
7013 is_a <scalar_int_mode> instead of checking for a nonzero
7014 precision. Forcibly convert op_mode to a scalar_int_mode
7015 in that case. More clearly differentiate the operand and
7016 result modes and use the former when deciding what the value
7017 of a count-bits operation should be. Use is_int_mode instead
7018 of checking for a MODE_INT. Remove redundant check for whether
7019 this mode has a zero precision.
7021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7022 Alan Hayward <alan.hayward@arm.com>
7023 David Sherwood <david.sherwood@arm.com>
7025 * optabs.c (widen_leading): Change the type of the mode argument
7026 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
7027 (widen_bswap): Likewise.
7028 (expand_parity): Likewise.
7029 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
7030 (expand_ffs): Likewise.
7031 (epand_unop): Check for scalar integer modes before calling the
7034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7035 Alan Hayward <alan.hayward@arm.com>
7036 David Sherwood <david.sherwood@arm.com>
7038 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
7040 (expand_expr_real_1): Update call accordingly.
7042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7043 Alan Hayward <alan.hayward@arm.com>
7044 David Sherwood <david.sherwood@arm.com>
7046 * expmed.c (store_bit_field_using_insv): Add op0_mode and
7047 value_mode arguments. Use scalar_int_mode internally.
7048 (store_bit_field_1): Rename the new integer mode from imode
7049 to op0_mode and use it instead of GET_MODE (op0). Update calls
7050 to store_split_bit_field, store_bit_field_using_insv and
7051 store_fixed_bit_field.
7052 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
7053 Use scalar_int_mode internally. Use a bit count rather than a mode
7054 when calculating the largest bit size for get_best_mode.
7055 Update calls to store_split_bit_field and store_fixed_bit_field_1.
7056 (store_fixed_bit_field_1): Add mode and value_mode arguments.
7057 Remove assertion that OP0 has a scalar integer mode.
7058 (store_split_bit_field): Add op0_mode and value_mode arguments.
7059 Update calls to extract_fixed_bit_field.
7060 (extract_bit_field_using_extv): Add an op0_mode argument.
7061 Use scalar_int_mode internally.
7062 (extract_bit_field_1): Rename the new integer mode from imode to
7063 op0_mode and use it instead of GET_MODE (op0). Update calls to
7064 extract_split_bit_field, extract_bit_field_using_extv and
7065 extract_fixed_bit_field.
7066 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
7067 to extract_split_bit_field and extract_fixed_bit_field_1.
7068 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
7069 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
7071 (extract_split_bit_field): Add an op0_mode argument. Update call
7072 to extract_fixed_bit_field.
7074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7075 Alan Hayward <alan.hayward@arm.com>
7076 David Sherwood <david.sherwood@arm.com>
7078 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
7079 * explow.c (hard_function_value): Likewise.
7080 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
7081 convert_to_mode call outside the loop.
7082 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
7083 for the mode iterator. Require the mode specified by max_pieces
7085 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
7087 (copy_blkmode_to_reg): Likewise.
7088 (set_storage_via_setmem): Likewise.
7089 * optabs.c (prepare_cmp_insn): Likewise.
7090 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7091 * stor-layout.c (finish_bitfield_representative): Likewise.
7093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7094 Alan Hayward <alan.hayward@arm.com>
7095 David Sherwood <david.sherwood@arm.com>
7097 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
7098 * expr.c (convert_move): Use them.
7099 (convert_modes): Likewise.
7100 (store_expr_with_bounds): Likewise.
7102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7103 Alan Hayward <alan.hayward@arm.com>
7104 David Sherwood <david.sherwood@arm.com>
7106 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
7107 parameter for the mode of "x". Remove the "known_x", "known_mode"
7108 and "known_ret" arguments. Change the type of the mode argument
7110 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
7111 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
7112 (reg_num_sign_bit_copies_for_combine): Likewise.
7113 * rtlanal.c (nonzero_bits1): Likewise.
7114 (num_sign_bit_copies1): Likewise.
7115 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
7116 (reg_num_sign_bit_copies_general): Likewise.
7117 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
7118 (reg_nonzero_bits_general): Likewise.
7120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7121 Alan Hayward <alan.hayward@arm.com>
7122 David Sherwood <david.sherwood@arm.com>
7124 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
7125 than in subroutines. Return 1 for non-integer modes.
7126 (cached_num_sign_bit_copies): Change the type of the mode parameter
7128 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
7129 classes. Handle CONST_INT_P first and then check whether X also
7130 has a scalar integer mode. Check the same thing for inner registers
7131 of a SUBREG and for values that are being extended or truncated.
7133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7134 Alan Hayward <alan.hayward@arm.com>
7135 David Sherwood <david.sherwood@arm.com>
7137 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
7138 in subroutines. Return the mode mask for non-integer modes.
7139 (cached_nonzero_bits): Change the type of the mode parameter
7141 (nonzero_bits1): Likewise. Remove early exit for other mode
7142 classes. Handle CONST_INT_P first and then check whether X
7143 also has a scalar integer mode.
7145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7146 Alan Hayward <alan.hayward@arm.com>
7147 David Sherwood <david.sherwood@arm.com>
7149 * expr.c (widest_int_mode_for_size): Make the comment match the code.
7150 Return a scalar_int_mode and assert that the size is greater than
7152 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
7153 (op_by_pieces_d::op_by_pieces_d): Likewise.
7154 (op_by_pieces_d::run): Likewise.
7155 (can_store_by_pieces): Likewise.
7157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7158 Alan Hayward <alan.hayward@arm.com>
7159 David Sherwood <david.sherwood@arm.com>
7161 * combine.c (extract_left_shift): Add a mode argument and update
7163 (make_compound_operation_int): Change the type of the mode parameter
7164 to scalar_int_mode and update the call to extract_left_shift.
7166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7167 Alan Hayward <alan.hayward@arm.com>
7168 David Sherwood <david.sherwood@arm.com>
7170 * combine.c (simplify_and_const_int): Change the type of the mode
7171 parameter to scalar_int_mode.
7172 (simplify_and_const_int_1): Likewise. Update recursive call.
7174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7175 Alan Hayward <alan.hayward@arm.com>
7176 David Sherwood <david.sherwood@arm.com>
7178 * combine.c (simplify_compare_const): Check that the mode is a
7179 scalar_int_mode (rather than VOIDmode) before testing its
7181 (simplify_comparison): Move COMPARISON_P handling out of the
7182 loop and restrict the latter part of the loop to scalar_int_modes.
7183 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
7184 and when considering SUBREG_REGs. Use is_int_mode instead of
7185 checking GET_MODE_CLASS against MODE_INT.
7187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7188 Alan Hayward <alan.hayward@arm.com>
7189 David Sherwood <david.sherwood@arm.com>
7191 * combine.c (try_widen_shift_mode): Move check for equal modes to...
7192 (simplify_shift_const_1): ...here. Use scalar_int_mode for
7193 shift_unit_mode and for modes involved in scalar shifts.
7195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7196 Alan Hayward <alan.hayward@arm.com>
7197 David Sherwood <david.sherwood@arm.com>
7199 * combine.c (force_int_to_mode): New function, split out from...
7200 (force_to_mode): ...here. Keep xmode up-to-date and use it
7201 instead of GET_MODE (x).
7203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7204 Alan Hayward <alan.hayward@arm.com>
7205 David Sherwood <david.sherwood@arm.com>
7207 * optabs-query.h (extraction_insn::struct_mode): Change type to
7208 opt_scalar_int_mode and update comment.
7209 (extraction_insn::field_mode): Change type to scalar_int_mode.
7210 (extraction_insn::pos_mode): Likewise.
7211 * combine.c (make_extraction): Update accordingly.
7212 * optabs-query.c (get_traditional_extraction_insn): Likewise.
7213 (get_optab_extraction_insn): Likewise.
7214 * recog.c (simplify_while_replacing): Likewise.
7215 * expmed.c (narrow_bit_field_mem): Change the type of the mode
7216 parameter to opt_scalar_int_mode.
7218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7219 Alan Hayward <alan.hayward@arm.com>
7220 David Sherwood <david.sherwood@arm.com>
7222 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
7223 to a scalar_int_mode instead of a machine_mode.
7224 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
7225 (get_best_mode): Return a boolean and use a pointer argument to store
7226 the selected mode. Replace the limit mode parameter with a bit limit.
7227 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
7228 for the values returned by bit_field_mode_iterator::next_mode.
7229 (store_bit_field): Update call to get_best_mode.
7230 (store_fixed_bit_field): Likewise.
7231 (extract_fixed_bit_field): Likewise.
7232 * expr.c (optimize_bitfield_assignment_op): Likewise.
7233 * fold-const.c (optimize_bit_field_compare): Likewise.
7234 (fold_truth_andor_1): Likewise.
7235 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
7236 Update for new type of m_mode.
7237 (get_best_mode): As above.
7239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7240 Alan Hayward <alan.hayward@arm.com>
7241 David Sherwood <david.sherwood@arm.com>
7243 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
7244 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
7245 (store_bit_field): Check is_a <scalar_int_mode> before calling
7246 strict_volatile_bitfield_p.
7247 (extract_bit_field): Likewise.
7249 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7250 Alan Hayward <alan.hayward@arm.com>
7251 David Sherwood <david.sherwood@arm.com>
7253 * target.def (cstore_mode): Return a scalar_int_mode.
7254 * doc/tm.texi: Regenerate.
7255 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
7256 * targhooks.h (default_cstore_mode): Likewise.
7257 * targhooks.c (default_cstore_mode): Likewise, using a forced
7259 * expmed.c (emit_cstore): Expect the target of the cstore to be
7262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7263 Alan Hayward <alan.hayward@arm.com>
7264 David Sherwood <david.sherwood@arm.com>
7266 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
7268 (niter_desc): Likewise mode.
7269 (iv_analyze): Add a mode parameter.
7271 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
7272 and change its type to scalar_int_mode.
7273 * loop-iv.c: Update commentary at head of file.
7274 (iv_constant): Pass the mode paraeter before the rtx it describes
7275 and change its type to scalar_int_mode. Remove VOIDmode handling.
7276 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
7277 (iv_extend): Likewise.
7278 (shorten_into_mode): Likewise.
7279 (iv_add): Use scalar_int_mode.
7280 (iv_mult): Likewise.
7281 (iv_shift): Likewise.
7282 (canonicalize_iv_subregs): Likewise.
7283 (get_biv_step_1): Pass the outer_mode parameter before the rtx
7284 it describes and change its mode to scalar_int_mode. Also change
7285 the type of the returned inner_mode to scalar_int_mode.
7286 (get_biv_step): Likewise, turning outer_mode from a pointer
7287 into a direct parameter. Update call to get_biv_step_1.
7288 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
7289 iv_constant and get_biv_step.
7290 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
7291 and change its type to scalar_int_mode. Don't initialise iv->mode
7292 to VOIDmode and remove later checks for its still being VOIDmode.
7293 Update calls to iv_analyze_op and iv_analyze_expr. Check
7294 is_a <scalar_int_mode> when changing the mode under consideration.
7295 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
7296 Update call to iv_analyze_expr.
7297 (iv_analyze_op): Add a mode parameter. Reject subregs whose
7298 inner register is not also a scalar_int_mode. Update call to
7300 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
7301 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
7302 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
7303 separate mode class checks. Update calls to iv_analyze. Remove
7304 fix-up of VOIDmodes after iv_analyze_biv.
7305 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
7306 don't have a scalar_int_mode. Update call to biv_p.
7308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7309 Alan Hayward <alan.hayward@arm.com>
7310 David Sherwood <david.sherwood@arm.com>
7312 * cfgexpand.c (convert_debug_memory_address): Use
7313 as_a <scalar_int_mode>.
7314 * combine.c (expand_compound_operation): Likewise.
7315 (make_extraction): Likewise.
7316 (change_zero_ext): Likewise.
7317 (simplify_comparison): Likewise.
7318 * cse.c (cse_insn): Likewise.
7319 * dwarf2out.c (minmax_loc_descriptor): Likewise.
7320 (mem_loc_descriptor): Likewise.
7321 (loc_descriptor): Likewise.
7322 * expmed.c (init_expmed_one_mode): Likewise.
7323 (synth_mult): Likewise.
7324 (emit_store_flag_1): Likewise.
7325 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
7326 of a comparison with size.
7327 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
7328 (reduce_to_bit_field_precision): Likewise.
7329 * function.c (expand_function_end): Likewise.
7330 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
7331 * loop-doloop.c (doloop_modify): Likewise.
7332 * optabs.c (expand_binop): Likewise.
7333 (expand_unop): Likewise.
7334 (expand_copysign_absneg): Likewise.
7335 (prepare_cmp_insn): Likewise.
7336 (maybe_legitimize_operand): Likewise.
7337 * recog.c (const_scalar_int_operand): Likewise.
7338 * rtlanal.c (get_address_mode): Likewise.
7339 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7340 (simplify_cond_clz_ctz): Likewise.
7341 * tree-nested.c (get_nl_goto_field): Likewise.
7342 * tree.c (build_vector_type_for_mode): Likewise.
7343 * var-tracking.c (use_narrower_mode): Likewise.
7345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7346 Alan Hayward <alan.hayward@arm.com>
7347 David Sherwood <david.sherwood@arm.com>
7349 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
7350 * builtins.c (expand_builtin_signbit): Use it.
7351 * cfgexpand.c (expand_debug_expr): Likewise.
7352 * dojump.c (do_jump): Likewise.
7353 (do_compare_and_jump): Likewise.
7354 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
7355 * expmed.c (make_tree): Likewise.
7356 * expr.c (expand_expr_real_2): Likewise.
7357 (expand_expr_real_1): Likewise.
7358 (try_casesi): Likewise.
7359 * fold-const-call.c (fold_const_call_ss): Likewise.
7360 * fold-const.c (unextend): Likewise.
7361 (extract_muldiv_1): Likewise.
7362 (fold_single_bit_test): Likewise.
7363 (native_encode_int): Likewise.
7364 (native_encode_string): Likewise.
7365 (native_interpret_int): Likewise.
7366 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
7367 * internal-fn.c (expand_addsub_overflow): Likewise.
7368 (expand_neg_overflow): Likewise.
7369 (expand_mul_overflow): Likewise.
7370 (expand_arith_overflow): Likewise.
7371 * match.pd: Likewise.
7372 * stor-layout.c (layout_type): Likewise.
7373 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7374 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
7375 * tree-ssanames.c (get_range_info): Likewise.
7376 * tree-switch-conversion.c (array_value_type) Likewise.
7377 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
7378 (vect_recog_divmod_pattern): Likewise.
7379 (vect_recog_mixed_size_cond_pattern): Likewise.
7380 * tree-vrp.c (extract_range_basic): Likewise.
7381 (simplify_float_conversion_using_ranges): Likewise.
7382 * tree.c (int_fits_type_p): Likewise.
7383 * ubsan.c (instrument_bool_enum_load): Likewise.
7384 * varasm.c (mergeable_string_section): Likewise.
7385 (narrowing_initializer_constant_valid_p): Likewise.
7386 (output_constant): Likewise.
7388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7389 Alan Hayward <alan.hayward@arm.com>
7390 David Sherwood <david.sherwood@arm.com>
7392 * machmode.h (NARROWEST_INT_MODE): New macro.
7393 * expr.c (alignment_for_piecewise_move): Use it instead of
7394 GET_CLASS_NARROWEST_MODE (MODE_INT).
7395 (push_block): Likewise.
7396 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
7398 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7401 Alan Hayward <alan.hayward@arm.com>
7402 David Sherwood <david.sherwood@arm.com>
7404 * postreload.c (move2add_valid_value_p): Change the type of the
7405 mode parameter to scalar_int_mode.
7406 (move2add_use_add2_insn): Add a mode parameter and use it instead
7408 (move2add_use_add3_insn): Likewise.
7409 (reload_cse_move2add): Update accordingly.
7411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7412 Alan Hayward <alan.hayward@arm.com>
7413 David Sherwood <david.sherwood@arm.com>
7415 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
7417 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
7418 * optabs.c (expand_unop): Likewise.
7420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7421 Alan Hayward <alan.hayward@arm.com>
7422 David Sherwood <david.sherwood@arm.com>
7424 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
7425 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
7426 (popcount_loc_descriptor): Likewise.
7427 (bswap_loc_descriptor): Likewise.
7428 (rotate_loc_descriptor): Likewise.
7429 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
7430 calling the functions above.
7432 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7433 Alan Hayward <alan.hayward@arm.com>
7434 David Sherwood <david.sherwood@arm.com>
7436 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
7438 (try_combine): Likewise.
7439 (simplify_if_then_else): Likewise.
7440 * cse.c (cse_insn): Likewise.
7441 * dwarf2out.c (mem_loc_descriptor): Likewise.
7442 * emit-rtl.c (gen_lowpart_common): Likewise.
7443 * simplify-rtx.c (simplify_truncation): Likewise.
7444 (simplify_binary_operation_1): Likewise.
7445 (simplify_const_relational_operation): Likewise.
7446 (simplify_ternary_operation): Likewise.
7447 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
7449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7450 Alan Hayward <alan.hayward@arm.com>
7451 David Sherwood <david.sherwood@arm.com>
7453 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
7454 * reload.c (push_reload): Likewise.
7455 (find_reloads): Likewise.
7457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7458 Alan Hayward <alan.hayward@arm.com>
7459 David Sherwood <david.sherwood@arm.com>
7461 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
7462 (make_compound_operation_int): Likewise.
7463 (change_zero_ext): Likewise.
7464 * expr.c (convert_move): Likewise.
7465 (convert_modes): Likewise.
7466 * fwprop.c (forward_propagate_subreg): Likewise.
7467 * loop-iv.c (get_biv_step_1): Likewise.
7468 * optabs.c (widen_operand): Likewise.
7469 * postreload.c (move2add_valid_value_p): Likewise.
7470 * recog.c (simplify_while_replacing): Likewise.
7471 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7472 (simplify_binary_operation_1): Likewise. Remove redundant
7473 mode equality check.
7475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7476 Alan Hayward <alan.hayward@arm.com>
7477 David Sherwood <david.sherwood@arm.com>
7479 * combine.c (combine_simplify_rtx): Add checks for
7480 is_a <scalar_int_mode>.
7481 (simplify_if_then_else): Likewise.
7482 (make_field_assignment): Likewise.
7483 (simplify_comparison): Likewise.
7484 * ifcvt.c (noce_try_bitop): Likewise.
7485 * loop-invariant.c (canonicalize_address_mult): Likewise.
7486 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7489 Alan Hayward <alan.hayward@arm.com>
7490 David Sherwood <david.sherwood@arm.com>
7492 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
7493 is_a <scalar_int_mode> instead of != BLKmode.
7495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7496 Alan Hayward <alan.hayward@arm.com>
7497 David Sherwood <david.sherwood@arm.com>
7499 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
7500 instead of != VOIDmode.
7501 * combine.c (if_then_else_cond): Likewise.
7502 (change_zero_ext): Likewise.
7503 * dwarf2out.c (mem_loc_descriptor): Likewise.
7504 (loc_descriptor): Likewise.
7505 * rtlanal.c (canonicalize_condition): Likewise.
7506 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
7508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7509 Alan Hayward <alan.hayward@arm.com>
7510 David Sherwood <david.sherwood@arm.com>
7512 * simplify-rtx.c (simplify_binary_operation_1): Use
7513 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
7515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7516 Alan Hayward <alan.hayward@arm.com>
7517 David Sherwood <david.sherwood@arm.com>
7519 * wide-int.h (int_traits<unsigned char>) New class.
7520 (int_traits<unsigned short>) Likewise.
7521 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
7522 Use GET_MODE_UNIT_PRECISION and remove redundant test for
7524 * combine.c (set_nonzero_bits_and_sign_copies): Use
7525 is_a <scalar_int_mode>.
7526 (find_split_point): Likewise.
7527 (combine_simplify_rtx): Likewise.
7528 (simplify_logical): Likewise.
7529 (expand_compound_operation): Likewise.
7530 (expand_field_assignment): Likewise.
7531 (make_compound_operation): Likewise.
7532 (extended_count): Likewise.
7533 (change_zero_ext): Likewise.
7534 (simplify_comparison): Likewise.
7535 * dwarf2out.c (scompare_loc_descriptor): Likewise.
7536 (ucompare_loc_descriptor): Likewise.
7537 (minmax_loc_descriptor): Likewise.
7538 (mem_loc_descriptor): Likewise.
7539 (loc_descriptor): Likewise.
7540 * expmed.c (init_expmed_one_mode): Likewise.
7541 * lra-constraints.c (lra_constraint_offset): Likewise.
7542 * optabs.c (prepare_libcall_arg): Likewise.
7543 * postreload.c (move2add_note_store): Likewise.
7544 * reload.c (operands_match_p): Likewise.
7545 * rtl.h (load_extend_op): Likewise.
7546 * rtlhooks.c (gen_lowpart_general): Likewise.
7547 * simplify-rtx.c (simplify_truncation): Likewise.
7548 (simplify_unary_operation_1): Likewise.
7549 (simplify_binary_operation_1): Likewise.
7550 (simplify_const_binary_operation): Likewise.
7551 (simplify_const_relational_operation): Likewise.
7552 (simplify_subreg): Likewise.
7553 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7554 * var-tracking.c (adjust_mems): Likewise.
7555 (prepare_call_arguments): Likewise.
7557 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7558 Alan Hayward <alan.hayward@arm.com>
7559 David Sherwood <david.sherwood@arm.com>
7561 * machmode.h (is_int_mode): New fuction.
7562 * combine.c (find_split_point): Use it.
7563 (combine_simplify_rtx): Likewise.
7564 (simplify_if_then_else): Likewise.
7565 (simplify_set): Likewise.
7566 (simplify_shift_const_1): Likewise.
7567 (simplify_comparison): Likewise.
7568 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
7569 * cse.c (notreg_cost): Likewise.
7570 (cse_insn): Likewise.
7571 * cselib.c (cselib_lookup_1): Likewise.
7572 * dojump.c (do_jump_1): Likewise.
7573 (do_compare_rtx_and_jump): Likewise.
7574 * dse.c (get_call_args): Likewise.
7575 * dwarf2out.c (rtl_for_decl_init): Likewise.
7576 (native_encode_initializer): Likewise.
7577 * expmed.c (emit_store_flag_1): Likewise.
7578 (emit_store_flag): Likewise.
7579 * expr.c (convert_modes): Likewise.
7580 (store_field): Likewise.
7581 (expand_expr_real_1): Likewise.
7582 * fold-const.c (fold_read_from_constant_string): Likewise.
7583 * gimple-ssa-sprintf.c (get_format_string): Likewise.
7584 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
7585 * optabs.c (expand_binop): Likewise.
7586 (expand_unop): Likewise.
7587 (expand_abs_nojump): Likewise.
7588 (expand_one_cmpl_abs_nojump): Likewise.
7589 * simplify-rtx.c (mode_signbit_p): Likewise.
7590 (val_signbit_p): Likewise.
7591 (val_signbit_known_set_p): Likewise.
7592 (val_signbit_known_clear_p): Likewise.
7593 (simplify_relational_operation_1): Likewise.
7594 * tree.c (vector_type_mode): Likewise.
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 * machmode.h (smallest_mode_for_size): Fix formatting.
7601 (smallest_int_mode_for_size): New function.
7602 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
7603 instead of smallest_mode_for_size.
7604 * combine.c (make_extraction): Likewise.
7605 * config/arc/arc.c (arc_expand_movmem): Likewise.
7606 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
7607 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7608 * config/s390/s390.c (s390_expand_insv): Likewise.
7609 * config/sparc/sparc.c (assign_int_registers): Likewise.
7610 * config/spu/spu.c (spu_function_value): Likewise.
7611 (spu_function_arg): Likewise.
7612 * coverage.c (get_gcov_type): Likewise.
7613 (get_gcov_unsigned_t): Likewise.
7614 * dse.c (find_shift_sequence): Likewise.
7615 * expmed.c (store_bit_field_1): Likewise.
7616 * expr.c (convert_move): Likewise.
7617 (store_field): Likewise.
7618 * internal-fn.c (expand_arith_overflow): Likewise.
7619 * optabs-query.c (get_best_extraction_insn): Likewise.
7620 * optabs.c (expand_twoval_binop_libfunc): Likewise.
7621 * stor-layout.c (layout_type): Likewise.
7622 (initialize_sizetypes): Likewise.
7623 * targhooks.c (default_get_mask_mode): Likewise.
7624 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
7626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7627 Alan Hayward <alan.hayward@arm.com>
7628 David Sherwood <david.sherwood@arm.com>
7630 * machmode.h (opt_mode::else_blk): New function.
7631 (int_mode_for_mode): Declare.
7632 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
7633 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
7635 * cfgexpand.c (expand_debug_expr): Likewise.
7636 * combine.c (gen_lowpart_or_truncate): Likewise.
7637 (gen_lowpart_for_combine): Likewise.
7638 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
7639 * config/avr/avr.c (avr_to_int_mode): Likewise.
7640 (avr_out_plus_1): Likewise.
7641 (avr_out_plus): Likewise.
7642 (avr_out_round): Likewise.
7643 * config/i386/i386.c (ix86_split_to_parts): Likewise.
7644 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
7645 (s390_expand_vcond): Likewise.
7646 * config/spu/spu.c (spu_split_immediate): Likewise.
7647 (spu_expand_mov): Likewise.
7648 * dse.c (get_stored_val): Likewise.
7649 * expmed.c (store_bit_field_1): Likewise.
7650 (convert_extracted_bit_field): Use int_mode_for_mode instead of
7652 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
7653 (extract_low_bits): Likewise.
7654 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
7655 handling rather than repeating the check.
7656 (emit_group_store): Likewise.
7657 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
7658 * optabs.c (expand_absneg_bit): Likewise.
7659 (expand_copysign_absneg): Likewise.
7660 (expand_copysign_bit): Likewise.
7661 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
7662 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
7663 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
7664 * var-tracking.c (prepare_call_arguments): Likewise.
7665 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
7666 int_mode_for_mode instead of mode_for_size.
7667 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
7669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7670 Alan Hayward <alan.hayward@arm.com>
7671 David Sherwood <david.sherwood@arm.com>
7673 * machmode.h (int_mode_for_size): New function.
7674 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
7675 instead of mode_for_size.
7676 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
7678 * combine.c (expand_field_assignment): Use int_mode_for_size
7679 instead of mode_for_size.
7680 (make_extraction): Likewise.
7681 (simplify_shift_const_1): Likewise.
7682 (simplify_comparison): Likewise.
7683 * dojump.c (do_jump): Likewise.
7684 * dwarf2out.c (mem_loc_descriptor): Likewise.
7685 * emit-rtl.c (init_derived_machine_modes): Likewise.
7686 * expmed.c (flip_storage_order): Likewise.
7687 (convert_extracted_bit_field): Likewise.
7688 * expr.c (copy_blkmode_from_reg): Likewise.
7689 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
7690 * internal-fn.c (expand_mul_overflow): Likewise.
7691 * lower-subreg.c (simple_move): Likewise.
7692 * optabs-libfuncs.c (init_optabs): Likewise.
7693 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7694 * tree.c (vector_type_mode): Likewise.
7695 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
7696 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
7697 * tree-vect-generic.c (expand_vector_parallel): Likewise.
7698 * tree-vect-stmts.c (vectorizable_load): Likewise.
7699 (vectorizable_store): Likewise.
7701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7702 Alan Hayward <alan.hayward@arm.com>
7703 David Sherwood <david.sherwood@arm.com>
7705 * coretypes.h (pod_mode): New type.
7706 (scalar_int_mode_pod): New typedef.
7707 * machmode.h (pod_mode): New class.
7708 (int_n_data_t::m): Change type to scalar_int_mode_pod.
7709 * genmodes.c (emit_mode_int_n): Update accordingly.
7710 * lower-subreg.h (target_lower_subreg): Change type to
7711 scalar_int_mode_pod.
7712 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
7713 scalar_int_mode_pod.
7715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7716 Alan Hayward <alan.hayward@arm.com>
7717 David Sherwood <david.sherwood@arm.com>
7719 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
7720 machine_mode to scalar_int_mode.
7721 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
7722 (rs6000_option_override_internal): Remove cast to int.
7723 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
7724 machine_mode to scalar_int_mode.
7725 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
7726 (rs6000_option_override_internal): Remove cast to int.
7727 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
7728 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
7730 * config/s390/s390.c (s390_expand_builtin): Likewise.
7731 * coretypes.h (scalar_int_mode): New type.
7732 (opt_scalar_int_mode): New typedef.
7733 * machmode.h (scalar_int_mode): New class.
7734 (scalar_int_mode::includes_p): New function.
7735 (byte_mode): Change type to scalar_int_mode.
7736 (word_mode): Likewise.
7737 (ptr_mode): Likewise.
7738 * emit-rtl.c (byte_mode): Likewise.
7739 (word_mode): Likewise.
7740 (ptr_mode): Likewise.
7741 (init_derived_machine_modes): Update accordingly.
7742 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
7743 and MODE_PARTIAL_INT.
7744 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
7745 opt_scalar_int_mode.
7747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7748 Alan Hayward <alan.hayward@arm.com>
7749 David Sherwood <david.sherwood@arm.com>
7751 * target.def (libgcc_floating_mode_supported_p): Take a
7753 * doc/tm.texi: Regenerate.
7754 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
7756 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7757 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
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 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
7765 * doc/tm.texi: Regenerate.
7766 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
7767 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
7768 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
7769 * targhooks.h (default_floatn_mode): Likewise.
7770 * targhooks.c (default_floatn_mode): Likewise.
7771 * tree.c (build_common_tree_nodes): Update accordingly.
7773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7774 Alan Hayward <alan.hayward@arm.com>
7775 David Sherwood <david.sherwood@arm.com>
7777 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
7778 (mode_iterator::iterate_p): Likewise.
7779 (mode_iterator::get_wider): Likewise.
7780 * expr.c (init_expr_target): Use opt_scalar_float_mode.
7782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7783 Alan Hayward <alan.hayward@arm.com>
7784 David Sherwood <david.sherwood@arm.com>
7786 * coretypes.h (opt_scalar_float_mode): New typedef.
7787 * machmode.h (float_mode_for_size): New function.
7788 * emit-rtl.c (double_mode): Delete.
7789 (init_emit_once): Use float_mode_for_size.
7790 * stor-layout.c (layout_type): Likewise.
7791 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
7793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7794 Alan Hayward <alan.hayward@arm.com>
7795 David Sherwood <david.sherwood@arm.com>
7797 * output.h (assemble_real): Take a scalar_float_mode.
7798 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
7799 * config/arm/arm.md (consttable_4): Likewise.
7800 (consttable_8): Likewise.
7801 (consttable_16): Likewise.
7802 * config/mips/mips.md (consttable_float): Likewise.
7803 * config/s390/s390.c (s390_output_pool_entry): Likewise.
7804 * varasm.c (assemble_real): Take a scalar_float_mode.
7805 (output_constant_pool_2): Update accordingly.
7806 (output_constant): Likewise.
7808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7809 Alan Hayward <alan.hayward@arm.com>
7810 David Sherwood <david.sherwood@arm.com>
7812 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
7813 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
7814 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
7815 (native_encode_real): Likewise.
7816 (native_interpret_real): Likewise.
7817 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
7818 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7821 Alan Hayward <alan.hayward@arm.com>
7822 David Sherwood <david.sherwood@arm.com>
7824 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
7825 <scalar_float_mode>. Simplify.
7826 (gen_extend_conv_libfunc): Likewise.
7828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7829 Alan Hayward <alan.hayward@arm.com>
7830 David Sherwood <david.sherwood@arm.com>
7832 * coretypes.h (scalar_float_mode): New type.
7833 * machmode.h (mode_traits::from_int): Use machine_mode if
7834 USE_ENUM_MODES is defined.
7835 (is_a): New function.
7837 (dyn_cast): Likewise.
7838 (scalar_float_mode): New class.
7839 (scalar_float_mode::includes_p): New function.
7840 (is_float_mode): Likewise.
7841 * gdbhooks.py (MachineModePrinter): New class.
7842 (build_pretty_printer): Use it for scalar_float_mode.
7843 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
7844 (format_helper::format_helper): Turn into a template.
7845 * genmodes.c (get_mode_class): New function.
7846 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
7847 or machine_mode if none.
7848 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
7849 as_a <scalar_float_mode>.
7850 * dwarf2out.c (mem_loc_descriptor): Likewise.
7851 (insert_float): Likewise.
7852 (add_const_value_attribute): Likewise.
7853 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7854 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
7855 (expand_unop): Update accordingly.
7856 (expand_abs_nojump): Likewise.
7857 (expand_copysign_absneg): Take a scalar_float_mode.
7858 (expand_copysign_bit): Likewise.
7859 (expand_copysign): Update accordingly.
7861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7862 Alan Hayward <alan.hayward@arm.com>
7863 David Sherwood <david.sherwood@arm.com>
7865 * coretypes.h (opt_mode): New class.
7866 * machmode.h (opt_mode): Likewise.
7867 (opt_mode::else_void): New function.
7868 (opt_mode::require): Likewise.
7869 (opt_mode::exists): Likewise.
7870 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
7871 (GET_MODE_2XWIDER_MODE): Likewise.
7872 (mode_iterator::get_wider): Update accordingly.
7873 (mode_iterator::get_2xwider): Likewise.
7874 (mode_iterator::get_known_wider): Likewise, turning into a template.
7875 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
7876 forcing a wider mode to exist.
7877 * config/cr16/cr16.h (LONG_REG_P): Likewise.
7878 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7879 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
7880 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
7881 * lower-subreg.c (init_lower_subreg): Likewise.
7882 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
7883 on the final iteration.
7884 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
7885 a wider mode exists before asking for a move pattern.
7886 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
7887 forcing a wider mode to exist.
7888 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
7889 returning false if no such mode exists.
7890 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
7891 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
7892 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
7893 Avoid checking for a MODE_INT if we already know the mode is not a
7895 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
7896 forcing a wider mode to exist.
7897 (expmed_mult_highpart_optab): Likewise.
7898 (expmed_mult_highpart): Likewise.
7899 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
7901 * lto-streamer-in.c (lto_input_mode_table): Likewise.
7902 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
7903 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7904 * internal-fn.c (expand_mul_overflow): Update use of
7905 GET_MODE_2XWIDER_MODE.
7906 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7907 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
7908 GET_MODE_WIDER_MODE.
7909 (convert_plusminus_to_widen): Likewise.
7910 * tree-switch-conversion.c (array_value_type): Likewise.
7911 * var-tracking.c (emit_note_insn_var_location): Likewise.
7912 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7913 Return false inside rather than outside the loop if no wider mode
7915 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
7916 and GET_MODE_2XWIDER_MODE
7917 (can_compare_p): Use else_void.
7918 * gdbhooks.py (OptMachineModePrinter): New class.
7919 (build_pretty_printer): Use it for opt_mode.
7921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7922 Alan Hayward <alan.hayward@arm.com>
7923 David Sherwood <david.sherwood@arm.com>
7925 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
7926 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
7928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7929 Alan Hayward <alan.hayward@arm.com>
7930 David Sherwood <david.sherwood@arm.com>
7932 * machmode.h (mode_traits): New structure.
7933 (get_narrowest_mode): New function.
7934 (mode_iterator::start): Likewise.
7935 (mode_iterator::iterate_p): Likewise.
7936 (mode_iterator::get_wider): Likewise.
7937 (mode_iterator::get_known_wider): Likewise.
7938 (mode_iterator::get_2xwider): Likewise.
7939 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
7940 (FOR_EACH_MODE): Likewise.
7941 (FOR_EACH_MODE_FROM): Likewise.
7942 (FOR_EACH_MODE_UNTIL): Likewise.
7943 (FOR_EACH_WIDER_MODE): Likewise.
7944 (FOR_EACH_2XWIDER_MODE): Likewise.
7945 * builtins.c (expand_builtin_strlen): Use new mode iterators.
7946 * combine.c (simplify_comparison): Likewise
7947 * config/i386/i386.c (type_natural_mode): Likewise.
7948 * cse.c (cse_insn): Likewise.
7949 * dse.c (find_shift_sequence): Likewise.
7950 * emit-rtl.c (init_derived_machine_modes): Likewise.
7951 (init_emit_once): Likewise.
7952 * explow.c (hard_function_value): Likewise.
7953 * expmed.c (extract_fixed_bit_field_1): Likewise.
7954 (extract_bit_field_1): Likewise.
7955 (expand_divmod): Likewise.
7956 (emit_store_flag_1): Likewise.
7957 * expr.c (init_expr_target): Likewise.
7958 (convert_move): Likewise.
7959 (alignment_for_piecewise_move): Likewise.
7960 (widest_int_mode_for_size): Likewise.
7961 (emit_block_move_via_movmem): Likewise.
7962 (copy_blkmode_to_reg): Likewise.
7963 (set_storage_via_setmem): Likewise.
7964 (compress_float_constant): Likewise.
7965 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7966 * optabs-query.c (get_best_extraction_insn): Likewise.
7967 * optabs.c (expand_binop): Likewise.
7968 (expand_twoval_unop): Likewise.
7969 (expand_twoval_binop): Likewise.
7970 (widen_leading): Likewise.
7971 (widen_bswap): Likewise.
7972 (expand_parity): Likewise.
7973 (expand_unop): Likewise.
7974 (prepare_cmp_insn): Likewise.
7975 (prepare_float_lib_cmp): Likewise.
7976 (expand_float): Likewise.
7977 (expand_fix): Likewise.
7978 (expand_sfix_optab): Likewise.
7979 * postreload.c (move2add_use_add2_insn): Likewise.
7980 * reg-stack.c (reg_to_stack): Likewise.
7981 * reginfo.c (choose_hard_reg_mode): Likewise.
7982 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7983 * stor-layout.c (mode_for_size): Likewise.
7984 (smallest_mode_for_size): Likewise.
7985 (mode_for_vector): Likewise.
7986 (finish_bitfield_representative): Likewise.
7987 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
7988 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
7989 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7990 * var-tracking.c (prepare_call_arguments): 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 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
7997 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
7998 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
7999 * machmode.h (mode_size): Move earlier in file.
8000 (mode_precision): Likewise.
8001 (mode_inner): Likewise.
8002 (mode_nunits): Likewise.
8003 (mode_unit_size): Likewise.
8004 (unit_unit_precision): Likewise.
8005 (mode_wider): Likewise.
8006 (mode_2xwider): Likewise.
8007 (machine_mode): New class.
8008 (mode_to_bytes): New function.
8009 (mode_to_bits): Likewise.
8010 (mode_to_precision): Likewise.
8011 (mode_to_inner): Likewise.
8012 (mode_to_unit_size): Likewise.
8013 (mode_to_unit_precision): Likewise.
8014 (mode_to_nunits): Likewise.
8015 (GET_MODE_SIZE): Use mode_to_bytes.
8016 (GET_MODE_BITSIZE): Use mode_to_bits.
8017 (GET_MODE_PRECISION): Use mode_to_precision.
8018 (GET_MODE_INNER): Use mode_to_inner.
8019 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
8020 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
8021 (GET_MODE_NUNITS): Use mode_to_nunits.
8022 * system.h (ALWAYS_INLINE): New macro.
8023 * config/powerpcspe/powerpcspe-c.c
8024 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
8025 int for arg1_mode and arg2_mode.
8027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8028 Alan Hayward <alan.hayward@arm.com>
8029 David Sherwood <david.sherwood@arm.com>
8031 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8032 Prefix mode names with E_ in case statements.
8033 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8034 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
8035 (aarch64_split_simd_move): Likewise.
8036 (aarch64_gen_storewb_pair): Likewise.
8037 (aarch64_gen_loadwb_pair): Likewise.
8038 (aarch64_gen_store_pair): Likewise.
8039 (aarch64_gen_load_pair): Likewise.
8040 (aarch64_get_condition_code_1): Likewise.
8041 (aarch64_constant_pool_reload_icode): Likewise.
8042 (get_rsqrte_type): Likewise.
8043 (get_rsqrts_type): Likewise.
8044 (get_recpe_type): Likewise.
8045 (get_recps_type): Likewise.
8046 (aarch64_gimplify_va_arg_expr): Likewise.
8047 (aarch64_simd_container_mode): Likewise.
8048 (aarch64_emit_load_exclusive): Likewise.
8049 (aarch64_emit_store_exclusive): Likewise.
8050 (aarch64_expand_compare_and_swap): Likewise.
8051 (aarch64_gen_atomic_cas): Likewise.
8052 (aarch64_emit_bic): Likewise.
8053 (aarch64_emit_atomic_swap): Likewise.
8054 (aarch64_emit_atomic_load_op): Likewise.
8055 (aarch64_evpc_trn): Likewise.
8056 (aarch64_evpc_uzp): Likewise.
8057 (aarch64_evpc_zip): Likewise.
8058 (aarch64_evpc_ext): Likewise.
8059 (aarch64_evpc_rev): Likewise.
8060 (aarch64_evpc_dup): Likewise.
8061 (aarch64_gen_ccmp_first): Likewise.
8062 (aarch64_gen_ccmp_next): Likewise.
8063 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8064 (alpha_emit_xfloating_libcall): Likewise.
8065 (emit_insxl): Likewise.
8066 (alpha_arg_type): Likewise.
8067 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
8068 (arc_preferred_simd_mode): Likewise.
8069 (arc_secondary_reload): Likewise.
8070 (get_arc_condition_code): Likewise.
8071 (arc_print_operand): Likewise.
8072 (arc_legitimate_constant_p): Likewise.
8073 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8074 * config/arc/arc.md (casesi_load): Likewise.
8075 (casesi_compact_jump): Likewise.
8076 * config/arc/predicates.md (proper_comparison_operator): Likewise.
8077 (cc_use_register): Likewise.
8078 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8079 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8080 (arm_init_iwmmxt_builtins): Likewise.
8081 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
8082 (neon_expand_vector_init): Likewise.
8083 (arm_attr_length_move_neon): Likewise.
8084 (maybe_get_arm_condition_code): Likewise.
8085 (arm_emit_vector_const): Likewise.
8086 (arm_preferred_simd_mode): Likewise.
8087 (arm_output_iwmmxt_tinsr): Likewise.
8088 (thumb1_output_casesi): Likewise.
8089 (thumb2_output_casesi): Likewise.
8090 (arm_emit_load_exclusive): Likewise.
8091 (arm_emit_store_exclusive): Likewise.
8092 (arm_expand_compare_and_swap): Likewise.
8093 (arm_evpc_neon_vuzp): Likewise.
8094 (arm_evpc_neon_vzip): Likewise.
8095 (arm_evpc_neon_vrev): Likewise.
8096 (arm_evpc_neon_vtrn): Likewise.
8097 (arm_evpc_neon_vext): Likewise.
8098 (arm_validize_comparison): Likewise.
8099 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
8100 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
8101 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
8102 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
8103 (c6x_preferred_simd_mode): Likewise.
8104 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
8105 (epiphany_rtx_costs): Likewise.
8106 * config/epiphany/predicates.md (proper_comparison_operator):
8108 * config/frv/frv.c (condexec_memory_operand): Likewise.
8109 (frv_emit_move): Likewise.
8110 (output_move_single): Likewise.
8111 (output_condmove_single): Likewise.
8112 (frv_hard_regno_mode_ok): Likewise.
8113 (frv_matching_accg_mode): Likewise.
8114 * config/h8300/h8300.c (split_adds_subs): Likewise.
8115 (h8300_rtx_costs): Likewise.
8116 (h8300_print_operand): Likewise.
8117 (compute_mov_length): Likewise.
8118 (output_logical_op): Likewise.
8119 (compute_logical_op_length): Likewise.
8120 (compute_logical_op_cc): Likewise.
8121 (h8300_shift_needs_scratch_p): Likewise.
8122 (output_a_shift): Likewise.
8123 (compute_a_shift_length): Likewise.
8124 (compute_a_shift_cc): Likewise.
8125 (expand_a_rotate): Likewise.
8126 (output_a_rotate): Likewise.
8127 * config/i386/i386.c (classify_argument): Likewise.
8128 (function_arg_advance_32): Likewise.
8129 (function_arg_32): Likewise.
8130 (function_arg_64): Likewise.
8131 (function_value_64): Likewise.
8132 (ix86_gimplify_va_arg): Likewise.
8133 (ix86_legitimate_constant_p): Likewise.
8134 (put_condition_code): Likewise.
8135 (split_double_mode): Likewise.
8136 (ix86_avx256_split_vector_move_misalign): Likewise.
8137 (ix86_expand_vector_logical_operator): Likewise.
8138 (ix86_split_idivmod): Likewise.
8139 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
8140 (ix86_build_const_vector): Likewise.
8141 (ix86_build_signbit_mask): Likewise.
8142 (ix86_match_ccmode): Likewise.
8143 (ix86_cc_modes_compatible): Likewise.
8144 (ix86_expand_branch): Likewise.
8145 (ix86_expand_sse_cmp): Likewise.
8146 (ix86_expand_sse_movcc): Likewise.
8147 (ix86_expand_int_sse_cmp): Likewise.
8148 (ix86_expand_vec_perm_vpermi2): Likewise.
8149 (ix86_expand_vec_perm): Likewise.
8150 (ix86_expand_sse_unpack): Likewise.
8151 (ix86_expand_int_addcc): Likewise.
8152 (ix86_split_to_parts): Likewise.
8153 (ix86_vectorize_builtin_gather): Likewise.
8154 (ix86_vectorize_builtin_scatter): Likewise.
8155 (avx_vpermilp_parallel): Likewise.
8156 (inline_memory_move_cost): Likewise.
8157 (ix86_tieable_integer_mode_p): Likewise.
8158 (x86_maybe_negate_const_int): Likewise.
8159 (ix86_expand_vector_init_duplicate): Likewise.
8160 (ix86_expand_vector_init_one_nonzero): Likewise.
8161 (ix86_expand_vector_init_one_var): Likewise.
8162 (ix86_expand_vector_init_concat): Likewise.
8163 (ix86_expand_vector_init_interleave): Likewise.
8164 (ix86_expand_vector_init_general): Likewise.
8165 (ix86_expand_vector_set): Likewise.
8166 (ix86_expand_vector_extract): Likewise.
8167 (emit_reduc_half): Likewise.
8168 (ix86_emit_i387_round): Likewise.
8169 (ix86_mangle_type): Likewise.
8170 (ix86_expand_round_sse4): Likewise.
8171 (expand_vec_perm_blend): Likewise.
8172 (canonicalize_vector_int_perm): Likewise.
8173 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
8174 (expand_vec_perm_1): Likewise.
8175 (expand_vec_perm_interleave3): Likewise.
8176 (expand_vec_perm_even_odd_pack): Likewise.
8177 (expand_vec_perm_even_odd_1): Likewise.
8178 (expand_vec_perm_broadcast_1): Likewise.
8179 (ix86_vectorize_vec_perm_const_ok): Likewise.
8180 (ix86_expand_vecop_qihi): Likewise.
8181 (ix86_expand_mul_widen_hilo): Likewise.
8182 (ix86_expand_sse2_abs): Likewise.
8183 (ix86_expand_pextr): Likewise.
8184 (ix86_expand_pinsr): Likewise.
8185 (ix86_preferred_simd_mode): Likewise.
8186 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
8187 * config/i386/sse.md (*andnot<mode>3): Likewise.
8188 (<mask_codefor><code><mode>3<mask_name>): Likewise.
8189 (*<code><mode>3): Likewise.
8190 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
8191 (ia64_expand_atomic_op): Likewise.
8192 (ia64_arg_type): Likewise.
8193 (ia64_mode_to_int): Likewise.
8194 (ia64_scalar_mode_supported_p): Likewise.
8195 (ia64_vector_mode_supported_p): Likewise.
8196 (expand_vec_perm_broadcast): Likewise.
8197 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
8198 (iq2000_function_arg_advance): Likewise.
8199 (iq2000_function_arg): Likewise.
8200 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
8201 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
8202 (m68k_libcall_value): Likewise.
8203 (m68k_function_value): Likewise.
8204 (sched_attr_op_type): Likewise.
8205 * config/mcore/mcore.c (mcore_output_move): Likewise.
8206 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
8208 (microblaze_function_arg): Likewise.
8209 * config/mips/mips.c (mips16_build_call_stub): Likewise.
8210 (mips_print_operand): Likewise.
8211 (mips_mode_ok_for_mov_fmt_p): Likewise.
8212 (mips_vector_mode_supported_p): Likewise.
8213 (mips_preferred_simd_mode): Likewise.
8214 (mips_expand_vpc_loongson_even_odd): Likewise.
8215 (mips_expand_vec_unpack): Likewise.
8216 (mips_expand_vi_broadcast): Likewise.
8217 (mips_expand_vector_init): Likewise.
8218 (mips_expand_vec_reduc): Likewise.
8219 (mips_expand_msa_cmp): Likewise.
8220 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
8221 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
8222 (cc_flags_for_mode): Likewise.
8223 * config/msp430/msp430.c (msp430_print_operand): Likewise.
8224 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
8225 (nds32_output_casesi_pc_relative): Likewise.
8226 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8227 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
8228 (nvptx_gen_unpack): Likewise.
8229 (nvptx_gen_pack): Likewise.
8230 (nvptx_gen_shuffle): Likewise.
8231 (nvptx_gen_wcast): Likewise.
8232 (nvptx_preferred_simd_mode): Likewise.
8233 * config/pa/pa.c (pa_secondary_reload): Likewise.
8234 * config/pa/predicates.md (base14_operand): Likewise.
8235 * config/powerpcspe/powerpcspe-c.c
8236 (altivec_resolve_overloaded_builtin): Likewise.
8237 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
8239 (rs6000_preferred_simd_mode): Likewise.
8240 (output_vec_const_move): Likewise.
8241 (rs6000_expand_vector_extract): Likewise.
8242 (rs6000_split_vec_extract_var): Likewise.
8243 (reg_offset_addressing_ok_p): Likewise.
8244 (rs6000_legitimate_offset_address_p): Likewise.
8245 (rs6000_legitimize_address): Likewise.
8246 (rs6000_emit_set_const): Likewise.
8247 (rs6000_const_vec): Likewise.
8248 (rs6000_emit_move): Likewise.
8249 (spe_build_register_parallel): Likewise.
8250 (rs6000_darwin64_record_arg_recurse): Likewise.
8251 (swap_selector_for_mode): Likewise.
8252 (spe_init_builtins): Likewise.
8253 (paired_init_builtins): Likewise.
8254 (altivec_init_builtins): Likewise.
8255 (do_load_for_compare): Likewise.
8256 (rs6000_generate_compare): Likewise.
8257 (rs6000_expand_float128_convert): Likewise.
8258 (emit_load_locked): Likewise.
8259 (emit_store_conditional): Likewise.
8260 (rs6000_output_function_epilogue): Likewise.
8261 (rs6000_handle_altivec_attribute): Likewise.
8262 (rs6000_function_value): Likewise.
8263 (emit_fusion_gpr_load): Likewise.
8264 (emit_fusion_p9_load): Likewise.
8265 (emit_fusion_p9_store): Likewise.
8266 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
8267 (fusion_gpr_mem_load): Likewise.
8268 (fusion_addis_mem_combo_load): Likewise.
8269 (fusion_addis_mem_combo_store): Likewise.
8270 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
8271 (fusion_gpr_mem_load): Likewise.
8272 (fusion_addis_mem_combo_load): Likewise.
8273 (fusion_addis_mem_combo_store): Likewise.
8274 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8276 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
8277 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8278 (rs6000_preferred_simd_mode): Likewise.
8279 (output_vec_const_move): Likewise.
8280 (rs6000_expand_vector_extract): Likewise.
8281 (rs6000_split_vec_extract_var): Likewise.
8282 (reg_offset_addressing_ok_p): Likewise.
8283 (rs6000_legitimate_offset_address_p): Likewise.
8284 (rs6000_legitimize_address): Likewise.
8285 (rs6000_emit_set_const): Likewise.
8286 (rs6000_const_vec): Likewise.
8287 (rs6000_emit_move): Likewise.
8288 (rs6000_darwin64_record_arg_recurse): Likewise.
8289 (swap_selector_for_mode): Likewise.
8290 (paired_init_builtins): Likewise.
8291 (altivec_init_builtins): Likewise.
8292 (rs6000_expand_float128_convert): Likewise.
8293 (emit_load_locked): Likewise.
8294 (emit_store_conditional): Likewise.
8295 (rs6000_output_function_epilogue): Likewise.
8296 (rs6000_handle_altivec_attribute): Likewise.
8297 (rs6000_function_value): Likewise.
8298 (emit_fusion_gpr_load): Likewise.
8299 (emit_fusion_p9_load): Likewise.
8300 (emit_fusion_p9_store): Likewise.
8301 * config/rx/rx.c (rx_gen_move_template): Likewise.
8302 (flags_from_mode): Likewise.
8303 * config/s390/predicates.md (s390_alc_comparison): Likewise.
8304 (s390_slb_comparison): Likewise.
8305 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
8306 (s390_vector_mode_supported_p): Likewise.
8307 (s390_cc_modes_compatible): Likewise.
8308 (s390_match_ccmode_set): Likewise.
8309 (s390_canonicalize_comparison): Likewise.
8310 (s390_emit_compare_and_swap): Likewise.
8311 (s390_branch_condition_mask): Likewise.
8312 (s390_rtx_costs): Likewise.
8313 (s390_secondary_reload): Likewise.
8314 (__SECONDARY_RELOAD_CASE): Likewise.
8315 (s390_expand_cs): Likewise.
8316 (s390_preferred_simd_mode): Likewise.
8317 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
8318 * config/sh/sh.c (sh_print_operand): Likewise.
8319 (dump_table): Likewise.
8320 (sh_secondary_reload): Likewise.
8321 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8322 * config/sh/sh.md (casesi_worker_1): Likewise.
8323 (casesi_worker_2): Likewise.
8324 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
8325 (fcc_comparison_operator): Likewise.
8326 * config/sparc/sparc.c (sparc_expand_move): Likewise.
8327 (emit_soft_tfmode_cvt): Likewise.
8328 (sparc_preferred_simd_mode): Likewise.
8329 (output_cbranch): Likewise.
8330 (sparc_print_operand): Likewise.
8331 (sparc_expand_vec_perm_bmask): Likewise.
8332 (vector_init_bshuffle): Likewise.
8333 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8334 (spu_vector_mode_supported_p): Likewise.
8335 (spu_expand_insv): Likewise.
8336 (spu_emit_branch_or_set): Likewise.
8337 (spu_handle_vector_attribute): Likewise.
8338 (spu_builtin_splats): Likewise.
8339 (spu_builtin_extract): Likewise.
8340 (spu_builtin_promote): Likewise.
8341 (spu_expand_sign_extend): Likewise.
8342 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8343 (tilegx_simd_int): Likewise.
8344 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
8345 (tilepro_simd_int): Likewise.
8346 * config/v850/v850.c (const_double_split): Likewise.
8347 (v850_print_operand): Likewise.
8348 (ep_memory_offset): Likewise.
8349 * config/vax/vax.c (vax_rtx_costs): Likewise.
8350 (vax_output_int_move): Likewise.
8351 (vax_output_int_add): Likewise.
8352 (vax_output_int_subtract): Likewise.
8353 * config/visium/predicates.md (visium_branch_operator): Likewise.
8354 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
8355 (visium_print_operand_address): Likewise.
8356 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8357 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
8358 (xtensa_expand_conditional_branch): Likewise.
8359 (xtensa_copy_incoming_a7): Likewise.
8360 (xtensa_output_literal): Likewise.
8361 * dfp.c (decimal_real_maxval): Likewise.
8362 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8365 Alan Hayward <alan.hayward@arm.com>
8366 David Sherwood <david.sherwood@arm.com>
8368 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
8369 (mode_nunits_inline): Likewise.
8370 (mode_inner_inline): Likewise.
8371 (mode_unit_size_inline): Likewise.
8372 (mode_unit_precision_inline): Likewise.
8373 (emit_insn_modes_h): Likewise. Also emit a #define of the
8375 (emit_mode_wider): Add an E_ prefix to mode names.
8376 (emit_mode_complex): Likewise.
8377 (emit_mode_inner): Likewise.
8378 (emit_mode_adjustments): Likewise.
8379 (emit_mode_int_n): Likewise.
8380 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
8381 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
8382 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
8383 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
8384 (CRC32_BUILTIN, ENTRY): Likewise.
8385 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
8386 (aarch64_pop_regs): Likewise.
8387 (aarch64_process_components): Likewise.
8388 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
8389 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
8390 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
8391 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
8392 * config/arm/arm.c (arm_init_libfuncs): Likewise.
8393 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
8395 * config/i386/i386-builtin.def (pcmpestr): Likewise.
8396 (pcmpistr): Likewise.
8397 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
8398 * config/mmix/mmix.c (mmix_output_condition): Likewise.
8399 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8401 * config/rl78/rl78.c (mduc_regs): Likewise.
8402 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8403 (htm_expand_builtin): Likewise.
8404 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
8405 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
8406 * config/xtensa/xtensa.c (print_operand): Likewise.
8407 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
8408 (NUM_MODE_VECTOR_INT): Likewise.
8409 * genoutput.c (null_operand): Likewise.
8410 (output_operand_data): Likewise.
8411 * genrecog.c (print_parameter_value): Likewise.
8412 * lra.c (debug_operand_data): Likewise.
8414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8416 * dwarf2out.c (scompare_loc_descriptor_wide)
8417 (scompare_loc_descriptor_narrow): New functions, split out from...
8418 (scompare_loc_descriptor): ...here.
8419 * expmed.c (emit_store_flag_int): New function, split out from...
8420 (emit_store_flag): ...here.
8422 2017-08-30 Richard Biener <rguenther@suse.de>
8424 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
8425 (dwarf2out_early_finish): Move setting of AT_pubnames from
8426 early debug output to early finish.
8428 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8430 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
8431 and -mdata-region to the assembler.
8433 2017-08-30 Richard Biener <rguenther@suse.de>
8435 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
8437 (gen_subprogram_die): Add DW_AT_object_pointer only early.
8438 (dwarf2out_early_global_decl): Only generate a DIE for the
8439 abstract origin if it doesn't already exist or is a declaration DIE.
8440 (resolve_addr): Do not add the linkage name twice when
8441 generating a stub DIE for the DW_TAG_GNU_call_site target.
8443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8445 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8446 Use machine_mode rather than int for arg1_mode.
8448 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8451 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
8452 that the second argument of the built-in functions to unpack
8453 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
8454 switch statement instead a lot of if statements.
8455 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
8456 Allow 64-bit values to be in Altivec registers as well as
8457 traditional floating point registers.
8458 (pack<mode>, FMOVE128_VSX iterator): Likewise.
8460 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
8462 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
8463 MAX_REGS_PER_ADDRESS == 1.
8465 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
8467 * config/i386/i386.opt (flag_fentry): Do not init to -1.
8468 * config/i386/i386.c (ix86_option_override_internal): Simplify
8469 setting of opts->x_flag_entry.
8471 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8472 Jakub Jelinek <jakub@redhat.com>
8473 Richard Biener <rguenther@suse.de>
8475 PR tree-optimization/81503
8476 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
8477 folded constant fits in the target type; reorder tests for clarity.
8479 2017-08-29 Martin Liska <mliska@suse.cz>
8481 * passes.def: Include pass_lower_switch.
8482 * stmt.c (dump_case_nodes): Remove and move to
8483 tree-switch-conversion.
8484 (case_values_threshold): Likewise.
8485 (expand_switch_as_decision_tree_p): Likewise.
8486 (emit_case_decision_tree): Likewise.
8487 (expand_case): Likewise.
8488 (balance_case_nodes): Likewise.
8489 (node_has_low_bound): Likewise.
8490 (node_has_high_bound): Likewise.
8491 (node_is_bounded): Likewise.
8492 (emit_case_nodes): Likewise.
8493 (struct simple_case_node): New struct.
8494 (add_case_node): Remove.
8495 (emit_case_dispatch_table): Use vector instead of case_list.
8496 (reset_out_edges_aux): Remove.
8497 (compute_cases_per_edge): Likewise.
8498 (expand_case): Build list of simple_case_node.
8499 (expand_sjlj_dispatch_table): Use it.
8500 * tree-switch-conversion.c (struct case_node): Moved from
8501 stmt.c and adjusted.
8502 (emit_case_nodes): Likewise.
8503 (node_has_low_bound): Likewise.
8504 (node_has_high_bound): Likewise.
8505 (node_is_bounded): Likewise.
8506 (case_values_threshold): Likewise.
8507 (reset_out_edges_aux): Likewise.
8508 (compute_cases_per_edge): Likewise.
8509 (add_case_node): Likewise.
8510 (dump_case_nodes): Likewise.
8511 (balance_case_nodes): Likewise.
8512 (expand_switch_as_decision_tree_p): Likewise.
8513 (emit_jump): Likewise.
8514 (emit_case_decision_tree): Likewise.
8515 (try_switch_expansion): Likewise.
8516 (do_jump_if_equal): Likewise.
8517 (emit_cmp_and_jump_insns): Likewise.
8518 (fix_phi_operands_for_edge): New function.
8519 (record_phi_operand_mapping): Likewise.
8520 (class pass_lower_switch): New pass.
8521 (pass_lower_switch::execute): New function.
8522 (make_pass_lower_switch): Likewise.
8523 (conditional_probability):
8524 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
8525 * tree-pass.h: Add make_pass_lower_switch.
8527 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8530 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
8533 2017-08-29 Richard Biener <rguenther@suse.de>
8535 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
8536 we do not add a DW_AT_inline attribute twice.
8537 (gen_subprogram_die): Remove code setting DW_AT_inline on
8538 DECL_ABSTRACT_P nodes.
8540 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
8542 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
8543 calls to internal functions.
8544 (gimplify_modify_expr): Likewise.
8545 * tree-call-cdce.c (use_internal_fn): Likewise.
8546 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
8547 (convert_to_divmod): Set the nothrow flag.
8548 * tree-if-conv.c (predicate_mem_writes): Likewise.
8549 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8550 (vectorizable_call): Likewise.
8551 (vectorizable_store): Likewise.
8552 (vectorizable_load): Likewise.
8553 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
8554 (vect_recog_mask_conversion_pattern): Likewise.
8556 2017-08-29 Martin Liska <mliska@suse.cz>
8559 * gcc.c (driver_handle_option): Add new argument.
8560 * opts-common.c (handle_option): Pass
8561 target_option_override_hook.
8562 * opts-global.c (lang_handle_option): Add new option.
8563 (set_default_handlers): Add new argument.
8564 (decode_options): Likewise.
8565 * opts.c (target_handle_option): Likewise.
8566 (common_handle_option): Call target_option_override_hook.
8567 * opts.h (struct cl_option_handler_func): Add hook for
8568 target option override.
8569 (struct cl_option_handlers): Likewise.
8570 (set_default_handlers): Add new argument.
8571 (decode_options): Likewise.
8572 (common_handle_option): Likewise.
8573 (target_handle_option): Likewise.
8574 * toplev.c (toplev::main): Pass targetm.target_option.override
8577 2017-08-29 Richard Biener <rguenther@suse.de>
8578 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8580 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
8581 life to the active subtree.
8583 2017-08-28 Jeff Law <law@redhat.com>
8585 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
8586 derive_equivalences.
8587 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
8589 (edge_info::derive_equivalences): New private member function
8591 * tree-ssa-dom.c (class edge_info): Changed from a struct
8592 to a class. Add ctor/dtor, methods and data members.
8593 (edge_info::edge_info): Renamed from allocate_edge_info.
8594 Initialize additional members.
8595 (edge_info::~edge_info): New.
8596 (free_dom_edge_info): Delete the edge info.
8597 (record_edge_info): Use new class & associated member functions.
8598 Tighten forms for testing for edge equivalences.
8599 (record_temporary_equivalences): Iterate over the simple
8600 equivalences rather than assuming there's only one per edge.
8601 (cprop_into_successor_phis): Iterate over the simple
8602 equivalences rather than assuming there's only one per edge.
8603 (optimize_stmt): Use operand_equal_p rather than pointer
8604 equality for mini-DSE code.
8606 2017-08-28 Nathan Sidwell <nathan@acm.org>
8608 * gcc.c (execute): Fold SIGPIPE handling into switch
8609 statement. Adjust internal error message.
8611 2017-08-28 Richard Biener <rguenther@suse.de>
8614 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
8615 Do nothing for removed DIEs.
8617 2017-08-28 Richard Biener <rguenther@suse.de>
8619 PR tree-optimization/81977
8620 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
8623 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
8626 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
8627 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
8628 using targetm.gen_mem_thread_fence.
8630 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
8633 * config/i386/i386.md (*<btsc><mode>): Change operand 2
8634 predicate to register_operand. Reorder operands.
8635 (*btr<mode>): Ditto.
8636 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
8637 (*btr<mode>_mask): Ditto.
8639 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
8641 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
8642 * config/rs6000/xmmintrin.h: New file.
8643 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
8645 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8648 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
8649 parameter and_insn and return it.
8650 (recombine_lvx_pattern): Insert a copy to ensure availability of
8651 the base register of the copied masking operation at the point of
8652 the instruction replacement.
8653 (recombine_stvx_pattern): Likewise.
8655 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
8657 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
8658 undocumented switches.
8659 (-mpower9-dform-vector): Likewise.
8660 (-mpower9-dform): Likewise.
8661 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
8662 comments to delete references to -mpower9-dform* switches.
8663 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
8664 Delete reference to -mpower9-dform* switches, test for
8665 -mpower9-vector instead.
8666 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
8667 (OTHER_P9_VECTOR_MASKS): Likewise.
8668 (POWERPC_MASKS): Likewise.
8669 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
8670 tests against -mpower9-dform* to -mpower9-vector. Delete code
8671 that checked for -mpower9-dform* consistancy with other options.
8672 Add test for -mpower9-misc to enable other power9 switches.
8673 (rs6000_init_hard_regno_mode_ok): Likewise.
8674 (rs6000_option_override_internal): Likewise.
8675 (rs6000_emit_prologue): Likewise.
8676 (rs6000_emit_epilogue): Likewise.
8677 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
8678 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
8679 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
8681 (emit_fusion_p9_store): Likewise.
8682 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
8683 resetting these macros if the assembler does not support ISA 3.0
8685 (TARGET_P9_DFORM_VECTOR): Likewise.
8686 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
8687 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
8689 2017-08-25 Alan Modra <amodra@gmail.com>
8691 PR rtl-optimization/81747
8692 * cse.c (cse_extended_basic_block): Don't attempt to record
8693 equivalences for degenerate conditional branches that branch
8694 to their fall-through.
8696 2017-08-24 Martin Sebor <msebor@redhat.com>
8699 * gimple-fold.c (size_must_be_zero_p): New function.
8700 (gimple_fold_builtin_memory_op): Call it.
8702 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
8704 * config/rs6000/mm_malloc.h: New file.
8706 2017-08-24 Bin Cheng <bin.cheng@arm.com>
8708 PR tree-optimization/81913
8709 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
8710 analysis when either IVs in condition can wrap.
8712 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
8714 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
8715 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
8717 2017-08-24 Richard Biener <rguenther@suse.de>
8720 * targhooks.c (default_target_can_inline_p): Properly
8721 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
8722 is present and always compare.
8723 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
8724 infer -mfpmath=sse from TARGET_SSE_P.
8725 (ix86_can_inline_p): Properly use target_option_default_node when
8726 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
8728 2017-08-24 Richard Biener <rguenther@suse.de>
8731 * dwarf2out.c (output_die): Handle flag_generate_offload like
8733 (output_comp_unit): Likewise.
8734 (gen_array_type_die): Likewise.
8735 (dwarf2out_early_finish): Likewise.
8736 (note_variable_value_in_expr): Likewise.
8737 (dwarf2out_finish): Likewise. Adjust assert.
8738 * cgraphunit.c (symbol_table::compile): Move setting of
8739 flag_generate_offload earlier ...
8740 (symbol_table::finalize_compilation_unit): ... here, before
8741 early debug finalization.
8743 2017-08-24 Richard Biener <rguenther@suse.de>
8745 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
8746 and ipa-fnsummary.h.
8747 (ix86_can_inline_p): When ix86_fpmath flags do not match
8748 check whether the callee uses FP math at all.
8750 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
8753 * tree-ssanames.c (get_nonzero_bits): Use element_precision
8754 instead of TYPE_PRECISION.
8756 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
8757 Alan Hayward <alan.hayward@arm.com>
8758 David Sherwood <david.sherwood@arm.com>
8760 * combine.c (make_extraction): Use subreg_offset_from_lsb.
8762 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
8764 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
8766 (ix86_frame::stack_realign_allocate): New field.
8767 (struct machine_frame_state): Modify comments.
8768 (machine_frame_state::sp_realigned_fp_end): New field.
8769 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
8771 (sp_valid_at): Add assertion to assure no attempt to access invalid
8772 offset of a realigned stack.
8773 (fp_valid_at): Likewise.
8774 (choose_baseaddr): Modify comments.
8775 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
8776 ix86_expand_prologue.
8777 (ix86_expand_prologue): Modify stack realignment and allocation.
8778 (ix86_expand_epilogue): Modify comments.
8779 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
8780 avx2_runtime, avx512f, and avx512f_runtime.
8782 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
8784 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
8785 (mstackrealign): Do not init to -1.
8786 * config/i386/i386.c (ix86_option_override_internal):
8787 Check opts_set, not opts when setting default value of
8788 opts->x_ix86_force_align_arg_pointer.
8790 2017-08-23 Richard Biener <rguenther@suse.de>
8792 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
8793 lang_hooks.decl_printable_name.
8794 * print-rtl-function.c (print_rtx_function): Likewise.
8795 * tree-pretty-print.c (dump_function_header): Likewise.
8797 2017-08-23 Richard Biener <rguenther@suse.de>
8800 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
8801 -g0 at compile-time.
8803 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8806 * doc/sourcebuild.texi (Other hardware attributes):
8809 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8812 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
8813 Fix single-use check.
8815 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
8817 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
8819 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8821 * doc/install.texi: Modify to add more details on running selected
8824 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8826 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
8827 is combined with -mabi=ms.
8828 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
8831 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8833 PR tree-optimization/81488
8834 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
8835 and cached_basis fields.
8836 (MAX_SPREAD): New constant.
8837 (alloc_cand_and_find_basis): Initialize new fields.
8838 (clear_visited): New function.
8839 (create_phi_basis_1): Rename from create_phi_basis, set visited
8840 and cached_basis fields.
8841 (create_phi_basis): New wrapper function.
8842 (phi_add_costs_1): Rename from phi_add_costs, add spread
8843 parameter, set visited field, short-circuit when limits reached.
8844 (phi_add_costs): New wrapper function.
8845 (record_phi_increments_1): Rename from record_phi_increments, set
8847 (record_phi_increments): New wrapper function.
8848 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
8849 (phi_incr_cost): New wrapper function.
8850 (all_phi_incrs_profitable_1): Rename from
8851 all_phi_incrs_profitable, set visited field.
8852 (all_phi_incrs_profitable): New wrapper function.
8854 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8855 Alan Hayward <alan.hayward@arm.com>
8856 David Sherwood <david.sherwood@arm.com>
8858 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
8859 that takes the outer and inner modes.
8860 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
8861 comparison as the canonical test for a paradoxical subreg.
8862 * combine.c (simplify_set): Use paradoxical_subreg_p.
8863 (make_extraction): Likewise.
8864 (force_to_mode): Likewise.
8865 (rtx_equal_for_field_assignment_p): Likewise.
8866 (gen_lowpart_for_combine): Likewise.
8867 (simplify_comparison): Likewise.
8868 * cse.c (equiv_constant): Likewise.
8869 * expmed.c (store_bit_field_1): Likewise.
8870 * final.c (alter_subreg): Likewise.
8871 * fwprop.c (propagate_rtx): Likewise.
8872 (forward_propagate_subreg): Likewise.
8873 * ira-conflicts.c (ira_build_conflicts): Likewise.
8874 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
8875 * lra-constraints.c (curr_insn_transform): Likewise.
8876 (split_reg): Likewise.
8877 * lra-eliminations.c (move_plus_up): Likewise.
8878 (lra_eliminate_regs_1): Likewise.
8879 * recog.c (general_operand): Likewise.
8880 * ree.c (combine_reaching_defs): Likewise.
8881 * reload.c (push_reload): Likewise.
8882 (find_reloads): Likewise.
8883 * reload1.c (elimination_effects): Likewise.
8884 (compute_reload_subreg_offset): Likewise.
8885 (choose_reload_regs): Likewise.
8886 * rtlanal.c (subreg_lsb_1): Likewise.
8887 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8888 (simplify_subreg): Likewise.
8889 * var-tracking.c (track_loc_p): Likewise.
8890 * emit-rtl.c (byte_lowpart_offset): Likewise.
8891 (paradoxical_subreg_p): Delete out-of-line definition.
8893 2017-08-22 Jeff Law <law@redhat.com>
8895 PR tree-optimization/81741
8896 PR tree-optimization/71947
8897 * tree-ssa-dom.c: Include tree-inline.h.
8898 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
8899 equivalences if one is more expensive to compute than the other.
8900 * tree-ssa-scopedtables.h (class const_or_copies): Make
8901 record_const_or_copy_raw method private.
8902 (class avail_exprs_stack): New method simplify_binary_operation.
8903 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
8904 avail_exprs_stack::simplify_binary_operation as needed.
8905 (avail_exprs_stack::simplify_binary_operation): New function.
8907 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
8909 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
8910 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
8911 (DOT_SYMBOLS): Likewise.
8912 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
8913 (RELOCATABLE_NEEDS_FIXUP): Likewise.
8914 (RS6000_ABI_NAME): Likewise.
8915 (TARGET_CMODEL): Likewise.
8916 (TOC_SECTION_ASM_OP): Likewise.
8917 (SET_CMODEL): New macro.
8918 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
8920 2017-08-22 Richard Biener <rguenther@suse.de>
8922 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
8923 to allow for free-lang-data replacements similar to verify_type_variant.
8925 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8926 Alan Hayward <alan.hayward@arm.com>
8927 David Sherwood <david.sherwood@arm.com>
8929 * config/aarch64/aarch64.md (casesi): Use DImode rather than
8930 VOIDmode for the LABEL_REF.
8932 2017-08-22 Richard Biener <rguenther@suse.de>
8934 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
8936 2017-08-22 Richard Biener <rguenther@suse.de>
8938 * common.opt (feliminate-dwarf2-dups): Ignore.
8939 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
8940 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
8941 same_die_p_wrap, compute_section_prefix,
8942 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
8943 (comdat_symbol_id, comdat_symbol_number): Likewise.
8944 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
8946 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
8947 (output_die): Mark unreachable path unreachable.
8948 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
8949 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
8950 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
8951 (dwarf2out_early_finish): Likewise.
8953 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
8955 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
8957 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
8960 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
8961 not VAR_P. Filter attribute warnings with OPT_Wattributes.
8962 (avr_attribute_table) <io, io_low, address>: Initialize
8963 .decl_required with true.
8965 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
8967 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
8968 undocumented debugging options.
8969 (-mvsx-scalar-double): Likewise.
8970 (-mallow-df-permute): Likewise.
8971 (-mvectorize-builtins): Likewise.
8972 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8973 (rs6000_builtin_vectorized_function): Likewise.
8974 (rs6000_builtin_md_vectorized_function): Likewise.
8975 (rs6000_opt_vars): Likewise.
8977 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
8980 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
8981 (*btrq_imm): Rename from *btrq.
8982 (*btcq_imm): Rename from *btcq.
8983 (btsc): New code attribute.
8984 (*<btsc><mode>): New insn pattern.
8985 (*btr<mode>): Ditto.
8986 (*<btsc><mode>_mask): New insn_and_split pattern.
8987 (*btr<mode>_mask): Ditto.
8989 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8990 Alan Hayward <alan.hayward@arm.com>
8991 David Sherwood <david.sherwood@arm.com>
8993 * function.c (pad_below): Simplify padding calculation.
8995 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8996 Alan Hayward <alan.hayward@arm.com>
8997 David Sherwood <david.sherwood@arm.com>
8999 * target.def (function_prologue): Remove frame size argument.
9000 (function_epilogue): Likewise.
9001 * doc/tm.texi: Regenerate.
9002 * final.c (final_start_function): Update call to function_prologue.
9003 (final_end_function): Update call to function_epilogue.
9004 (default_function_pro_epilogue): Remove frame size argument.
9005 * output.h (default_function_pro_epilogue): Likewise.
9006 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
9007 (arm_output_function_prologue): Likewise.
9008 * config/frv/frv.c (frv_function_prologue): Likewise.
9009 (frv_function_epilogue): Likewise.
9010 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
9011 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
9012 (ia64_output_function_epilogue): Likewise.
9013 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
9014 (m32r_output_function_epilogue): Likewise.
9015 * config/microblaze/microblaze.c (microblaze_function_prologue)
9016 (microblaze_function_epilogue): Likewise.
9017 * config/mips/mips.c (mips_output_function_prologue): Likewise.
9018 (mips_output_function_epilogue): Likewise.
9019 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
9020 (mmix_target_asm_function_epilogue): Likewise.
9021 * config/msp430/msp430.c (msp430_start_function): Likewise.
9022 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
9023 (nds32_asm_function_epilogue): Likewise.
9024 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
9025 * config/pa/pa.c (pa_output_function_prologue): Likewise.
9026 (pa_output_function_epilogue): Likewise.
9027 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
9028 (rs6000_output_function_epilogue): Likewise.
9029 * config/rl78/rl78.c (rl78_start_function): Likewise.
9030 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
9031 (rs6000_output_function_epilogue): Likewise.
9032 * config/rx/rx.c (rx_output_function_prologue): Likewise.
9033 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
9034 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
9035 (sparc_asm_function_epilogue): Likewise.
9037 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9039 * tree.h (type_has_mode_precision_p): New function.
9040 * convert.c (convert_to_integer_1): Use it.
9041 * expr.c (expand_expr_real_2): Likewise.
9042 (expand_expr_real_1): Likewise.
9043 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
9044 * match.pd: Likewise.
9045 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
9046 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
9047 * tree-tailcall.c (process_assignment): Likewise.
9048 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9049 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
9050 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
9051 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9052 (vectorizable_assignment): Likewise.
9053 (vectorizable_shift): Likewise.
9054 (vectorizable_operation): Likewise.
9055 * tree-vrp.c (register_edge_assert_for_2): Likewise.
9057 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
9059 * match.pd: Add pow (C, x) simplification.
9061 2017-08-21 Richard Biener <rguenther@suse.de>
9063 PR tree-optimization/81900
9064 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
9065 for blocks with abnormal predecessors.
9066 (compute_antic): Do not set visited flag prematurely.
9068 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
9071 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
9073 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9075 * stor-layout.h (vector_type_mode): Move to...
9076 * tree.h (vector_type_mode): ...here.
9077 * stor-layout.c (vector_type_mode): Move to...
9078 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
9080 2017-08-21 Richard Biener <rguenther@suse.de>
9082 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
9083 register_external_die hooks.
9084 (debug_false_tree_charstarstar_uhwistar): Declare.
9085 (debug_nothing_tree_charstar_uhwi): Likewise.
9086 * debug.c (do_nothing_debug_hooks): Adjust.
9087 (debug_false_tree_charstarstar_uhwistar): New do nothing.
9088 (debug_nothing_tree_charstar_uhwi): Likewise.
9089 * dbxout.c (dbx_debug_hooks): Adjust.
9090 (xcoff_debug_hooks): Likewise.
9091 * sdbout.c (sdb_debug_hooks): Likewise.
9092 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9093 * dwarf2out.c (macinfo_label_base): New global.
9094 (dwarf2out_register_external_die): New function for the
9095 register_external_die hook.
9096 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
9097 (dwarf2_debug_hooks): Use them.
9098 (dwarf2_lineno_debug_hooks): Adjust.
9099 (struct die_struct): Add with_offset flag.
9100 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
9101 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
9102 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
9103 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
9104 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
9105 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
9106 defining section names for the early LTO debug variants.
9107 (reset_indirect_string): New helper.
9108 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
9109 (print_dw_val): Add support for offsetted symbol references.
9110 (get_ultimate_context): Split out from is_cxx.
9111 (is_cxx): Use get_ultimate_context.
9112 (is_fortran): Add decl overload.
9113 (compute_comp_unit_symbol): Split out worker from
9114 compute_section_prefix.
9115 (compute_section_prefix): Call compute_comp_unit_symbol and
9116 set comdat_type_p here.
9117 (output_die): Skip DIE symbol output for the LTO added one.
9118 Handle DIE symbol references with offset.
9119 (output_comp_unit): Guard section name mangling properly.
9120 For LTO debug sections emit a symbol at the section beginning
9121 which we use to refer to its DIEs.
9122 (add_abstract_origin_attribute): For DIEs registered via
9123 dwarf2out_register_external_die directly refer to the early
9124 DIE rather than indirectly through the shadow one we created.
9125 Remove obsolete call to dwarf2out_abstract_function for
9126 non-function/block origins.
9127 (gen_array_type_die): When generating early LTO debug do
9128 not emit DW_AT_string_length.
9129 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
9130 late when in LTO. As suggested place a gcc_unreachable for
9131 the DECL_ABSTRACT_P case.
9132 (gen_subprogram_die): Avoid another specification DIE
9133 for early built declarations/definitions for the late LTO case.
9134 (gen_variable_die): Add type references for late duplicated VLA dies
9136 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
9137 we have the abstract instance already.
9138 (process_scope_var): Adjust decl DIE contexts in LTO which
9139 first puts them in limbo.
9140 (gen_decl_die): Do not generate type DIEs late apart from
9141 types for VLAs or for decls we do not yet have a DIE. Do not
9142 call dwarf2out_abstract_function late.
9143 (dwarf2out_early_global_decl): Make sure to create DIEs
9144 for abstract instances of a decl first.
9145 (dwarf2out_late_global_decl): Adjust comment.
9146 (output_macinfo_op): With multiple macro sections use
9147 macinfo_label_base to distinguish labels.
9148 (output_macinfo): Likewise. Update macinfo_label_base.
9149 Pass in the line info label.
9150 (note_variable_value_in_expr): When generating LTO resolve
9151 all variable values here by generating DIEs as needed.
9152 (init_sections_and_labels): Add early LTO debug flag parameter
9153 and generate different sections and names if set. Add generation
9154 counter for the labels so we can have multiple of them.
9155 (reset_dies): Helper to allow DIEs to be output multiple times.
9156 (dwarf2out_finish): When outputting DIEs to the fat part of an
9157 LTO object first reset DIEs.
9158 (dwarf2out_early_finish): Output early DIEs when generating LTO.
9159 (modified_type_die): Check for decl_ultimate_origin being self
9161 (gen_type_die_with_usage): Likewise.
9162 (gen_typedef_die): Allow decl_ultimate_origin being self.
9163 (set_decl_abstract_flags): Remove.
9164 (set_block_abstract_flags): Likewise.
9165 (dwarf2out_abstract_function): Treat the early generated DIEs
9166 as the abstract copy and only add DW_AT_inline and
9167 DW_AT_artificial here and call set_decl_origin_self.
9168 If the DIE has an abstract origin don't do anything.
9169 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
9170 if we have none yet (Go fails to build one, PR78628).
9171 (variably_modified_type_p): Prevent endless recursion for Ada
9172 cyclic pointer types.
9173 * lto-streamer-in.c: Include debug.h.
9174 (dref_queue): New global.
9175 (lto_read_tree_1): Stream in DIE references.
9176 (lto_input_tree): Register DIE references.
9177 (input_function): Stream DECL_DEBUG_ARGS.
9178 * lto-streamer-out.c: Include debug.h.
9179 (lto_write_tree_1): Output DIE references.
9180 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
9181 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
9182 (output_function): Stream DECL_DEBUG_ARGS.
9183 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
9184 Stream DECL_ABSTRACT_ORIGIN.
9185 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
9186 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
9187 DECL_CONTEXT for file-scope decls.
9188 * lto-streamer.h (struct dref_entry): Declare.
9189 (dref_queue): Likewise.
9190 * cfgexpand.c (pass_expand::execute): Do not call the
9191 outlining_inline_function hook here.
9192 * lto-wrapper.c (debug_obj): New global.
9193 (tool_cleanup): Unlink it if required.
9194 (debug_objcopy): New function.
9195 (run_gcc): Handle early debug sections in the IL files by
9196 extracting them to separate files, partially linkin them and
9197 feeding the result back as result to the linker.
9198 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
9199 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
9200 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
9201 sections into a separate segment.
9202 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
9204 (darwin_asm_dwarf_section): Likewise.
9205 (darwin_asm_output_dwarf_offset): Likewise.
9206 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
9208 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9209 Alan Hayward <alan.hayward@arm.com>
9210 David Sherwood <david.sherwood@arm.com>
9212 * read-md.h (md_reader::record_potential_iterator_use): Replace
9213 pointer argument with an rtx and an index.
9214 * read-rtl.c (iterator_group::apply_iterator): Likewise.
9215 (apply_mode_iterator): Likewise.
9216 (apply_code_iterator): Likewise.
9217 (apply_int_iterator): Likewise.
9218 (apply_subst_iterator): Likewise.
9219 (record_iterator_use): Likewise.
9220 (record_attribute_use): Likewise.
9221 (md_reader::record_potential_iterator_use): Likewise. Update calls
9222 to record_iterator_use and apply_iterator.
9223 (iterator_use): Replace ptr with x and index.
9224 (attribute_use): Likewise.
9225 (apply_attribute_uses): Update calls to apply_iterator.
9226 (apply_iterators): Likewise. Update initialization of iterator_use.
9227 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
9228 and record_potential_iterator_use.
9229 (rtx_reader::read_rtx_operand): Likewise.
9231 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9232 Alan Hayward <alan.hayward@arm.com>
9233 David Sherwood <david.sherwood@arm.com>
9235 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
9238 2017-08-21 Richard Biener <rguenther@suse.de>
9241 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
9242 at struct end conservatively when comparing common bases.
9244 2017-08-21 Richard Biener <rguenther@suse.de>
9246 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
9247 (mem_ref_in_stmt): Remove.
9248 (determine_max_movement): Use ref index to get at the reference.
9249 (invariantness_dom_walker::before_dom_children): Deal with
9250 lim data already initialized.
9251 (gather_mem_refs_stmt): Initialize lim data and record ref index.
9253 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9255 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
9256 (TARGET_ISA_ROUND): Ditto.
9257 (TARGET_ROUND): Ditto.
9258 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
9259 * config/i386/i386.md: Ditto.
9260 * config/i386/sse.md: Ditto.
9261 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
9262 with OPTION_MASK_ISA_SSE4_1.
9264 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9267 * doc/extend.texi (x86 Built-in Functions): Correct the name of
9268 __builtin_ia32_lzcnt_u16.
9270 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9273 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
9274 (rs6000_set_current_function): Rewrite function to use it.
9276 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
9279 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
9280 and TYPE_WARN_IF_NOT_ALIGN.
9281 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
9282 (handle_warn_if_not_align): New.
9283 (place_union_field): Call handle_warn_if_not_align.
9284 (place_field): Call handle_warn_if_not_align.
9285 Copy TYPE_WARN_IF_NOT_ALIGN.
9286 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
9287 (layout_type): Likewise.
9288 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
9290 (tree_decl_common): Add warn_if_not_align.
9291 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
9292 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
9293 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
9294 (DECL_WARN_IF_NOT_ALIGN): Likewise.
9295 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
9296 * doc/extend.texi: Document warn_if_not_aligned attribute.
9297 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
9299 2017-08-17 Martin Liska <mliska@suse.cz>
9302 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
9303 (get_data_dependence): Use it as pointer type.
9304 (distribute_loop): Likewise.
9306 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9308 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
9309 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
9310 (p8_vmrgow_<mode>_direct): New define_insn.
9311 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
9312 handle endianness for vmrgew and vmrgow permute patterns.
9314 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9316 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
9317 * config/rs6000/rs6000-cpus.def: Remove comment.
9318 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
9319 (POWERPC_MASKS): Likewise.
9320 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
9321 use of TARGET_VSX_TIMODE.
9322 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
9323 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
9324 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
9325 (rs6000_option_override_internal): Remove dead code.
9326 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
9327 (rs6000_legitimize_reload_address): Likewise.
9328 (rs6000_legitimate_address_p): Likewise.
9329 (rs6000_opt_masks): Delete "vsx-timode".
9330 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
9331 from function comment.
9332 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
9333 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
9334 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
9336 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
9337 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
9338 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
9341 (define_peephole2): Likewise.
9343 2017-08-17 Martin Sebor <msebor@redhat.com>
9346 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
9347 past the end of an array.
9348 (test_pp_format): Add test cases.
9350 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
9352 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
9353 missing ECF_NOTHROW flags.
9355 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9358 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
9359 operands residing in integer registers.
9360 (*vsx_le_perm_load_<mode>): Likewise.
9361 (*vsx_le_perm_store_<mode>): Likewise.
9362 (define_peephole2): Add peepholes to optimize the above.
9364 2017-08-17 Marek Polacek <polacek@redhat.com>
9367 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
9368 to mimic what shorten_compare did. Change the return type to bool.
9369 (fold_cond_expr_with_comparison): Update call to
9370 operand_equal_for_comparison_p.
9371 (fold_ternary_loc): Likewise.
9373 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
9375 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
9377 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
9379 2017-08-17 Richard Biener <rguenther@suse.de>
9381 * tree-ssa-structalias.c (solve_graph): When propagating
9382 to successors update the graphs succ edges and avoid duplicate work.
9384 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
9387 * config/i386/i386.c (ix86_option_override_internal): Save target
9388 specific options after ix86_stack_protector_guard_reg was changed.
9390 2017-08-17 Richard Biener <rguenther@suse.de>
9392 PR tree-optimization/81827
9393 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
9394 (new_var_info): Initialize it conservatively.
9395 (get_call_vi): Mark register vars.
9396 (new_scalar_tmp_constraint_exp): Likewise.
9397 (handle_rhs_call): Likewise.
9398 (handle_const_call): Likewise.
9399 (create_function_info_for): Likewise.
9400 (solve_constraints): Sort varinfos to separate register from
9401 non-register vars to pack points-to solution bitmaps during
9404 2017-08-17 Marek Polacek <polacek@redhat.com>
9406 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
9408 2017-08-17 Richard Biener <rguenther@suse.de>
9410 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
9411 to true when overflow is undefined and we saturated the result.
9413 2017-08-17 Alan Modra <amodra@gmail.com>
9416 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
9417 Don't use store multiple if only one reg needs saving.
9418 (interesting_frame_related_regno): New function.
9419 (rs6000_frame_related): Don't emit frame info for regs that
9421 (rs6000_emit_epilogue): Likewise.
9423 2017-08-16 Nathan Sidwell <nathan@acm.org>
9425 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
9426 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
9427 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
9428 * tree.c (free_lang_data_in_type): Use else-if chain. Always
9429 clear TYPE_LANG_1. Remove obsolete member-function stripping.
9430 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
9431 (verify_type): Adjust for TYPE_BINFO move.
9432 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
9433 process TYPE_BINFO directly.
9434 (hash_tree): Likewise.
9435 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9437 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9440 2017-08-16 David Malcolm <dmalcolm@redhat.com>
9442 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
9444 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
9447 * config/i386/i386.md (*anddi_1_btr): Change predicates of
9448 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
9449 Add ix86_binary_operator_ok to insn constraint.
9450 (*iordi_1_bts): Ditto.
9451 (*xordi_1_btc): Ditto.
9452 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
9453 Update corresponding peephole2 pattern.
9457 2017-08-16 Bin Cheng <bin.cheng@arm.com>
9459 PR tree-optimization/81832
9460 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
9461 copy loop header which has IFN_LOOP_DIST_ALIAS call.
9463 2017-08-16 Marek Polacek <polacek@redhat.com>
9466 * fold-const.c (fold_indirect_ref_1): Restore original behavior
9467 regarding size_zero_node.
9469 2017-08-16 Martin Liska <mliska@suse.cz>
9472 * config.gcc: Respect previously set extra_objs in case
9475 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
9477 PR tree-optimization/81835
9478 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
9479 the phi SSA_NAME. Check that the condition in a COND_EXPR does
9480 not depend on the phi.
9482 2017-08-16 Alan Modra <amodra@gmail.com>
9484 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
9487 2017-08-16 Alan Modra <amodra@gmail.com>
9489 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
9490 (save_reg_p): ..into this. Update all callers.
9491 (first_reg_to_save): Simplify.
9493 2017-08-16 Alan Modra <amodra@gmail.com>
9495 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
9498 2017-08-15 Joseph Myers <joseph@codesourcery.com>
9502 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
9503 constant count if that count is less than 32.
9505 2017-08-15 Nathan Sidwell <nathan@acm.org>
9507 * gcc.c (execute): Emit friendlier message if inferior is killed
9508 by an external cause.
9510 2017-08-15 Richard Biener <rguenther@suse.de>
9512 PR tree-optimization/81790
9513 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
9514 CONSTRUCTORs from simplifying and VN.
9516 2017-08-14 Martin Sebor <msebor@redhat.com>
9518 * builtin-attrs.def: Add comments.
9520 2017-08-14 Martin Sebor <msebor@redhat.com>
9523 * doc/extend.texi (attribute nonstring): Document new attribute.
9525 2017-08-14 Martin Sebor <msebor@redhat.com>
9528 * tree-diagnostic.c (default_tree_printer): Handle %G.
9529 * gimple-pretty-print.h (percent_G_format): Declare new function.
9530 * gimple-pretty-print.c (percent_G_format): Define.
9531 * tree-pretty-print.c (percent_K_format): Add argument.
9533 2017-08-14 Martin Sebor <msebor@redhat.com>
9535 PR translation/79998
9536 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
9537 Remove a stray space.
9539 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
9542 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
9543 (*iordi_1_bts): Ditto.
9544 (*xordi_1_btc): Ditto.
9546 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9549 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
9550 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9552 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
9553 quoted strings, and make more translator-friendly.
9554 (darwin_rs6000_override_options): Likewise.
9555 (rs6000_option_override_internal): Likewise.
9556 (rs6000_return_in_memory): Fix overlong line.
9557 (init_cmulative_args): Use quoted strings, and make more
9558 translator-friendly.
9559 (rs6000_pass_by_reference): Fix overlong line.
9560 (def_builtin): Use quoted strings.
9561 (altivec_expand_predicate_builtin): Use quoted strings, and make
9562 more translator-friendly.
9563 (htm_expand_builtin): Use quoted strings.
9564 (cpu_expand_builtin): Use quoted strings, and make more
9565 translator-friendly.
9566 (altivec_expand_builtin): Likewise.
9567 (paired_expand_predicate_builtin): Likewise.
9568 (rs6000_invalid_builtin): Likewise.
9569 (builtin_function_type): Use quoted strings.
9570 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
9571 more translator-friendly.
9572 (rs6000_trampoline_init): Likewise.
9573 (rs6000_handle_altivec_attribute): Likewise.
9574 (rs6000_inner_target_options): Use quoted strings.
9575 (rs6000_disable_incompatible_switches): Likewise.
9576 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
9577 strings, and make more translator-friendly.
9578 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
9580 2017-08-14 Bin Cheng <bin.cheng@arm.com>
9582 PR tree-optimization/81799
9583 * tree-loop-distribution.c (version_loop_by_alias_check): Force
9584 cond_expr to simple gimple operand.
9586 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
9589 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
9591 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
9595 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
9596 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
9597 TARGET_GASISR_PROLOGUES.
9598 * config/avr/avr.c (avr_option_override): Same.
9599 (avr_pass_pre_proep::execute): Same.
9601 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
9604 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
9605 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
9607 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
9609 * config/i386/i386.md (*load_tp_<mode>): Redefine as
9610 define_insn_and_split. Split to a memory load from 0 in
9611 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
9612 using PTR mode iterator.
9613 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
9614 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
9615 (*add_tp_<mode>): Redefine as define_insn_and_split.
9616 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
9617 address space. Merge with *add_tp_x32 using PTR mode iterator.
9618 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
9619 Split to an add with a memory load from 0 in
9620 DEFAULT_TLS_SEG_REG address space.
9622 2017-08-12 Andrew Pinski <apinski@cavium.com>
9624 * config/aarch64/aarch64-option-extensions.def (rdma):
9625 Fix feature string to what Linux prints out in /proc/cpuinfo.
9627 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
9630 * dwarf2out.c (modified_type_die): For C typedef types that have
9631 an ultimate origin, process the ultimate origin instead of the
9633 (gen_typedef_die): Assert that input DECLs have no ultimate
9635 (gen_type_die_with_usage): For typedef variants that have an
9636 ultimate origin, just call gen_decl_die on the original DECL.
9637 (process_scope_var): Avoid creating DIEs for local typedefs and
9638 concrete static variables.
9640 2017-08-12 Alan Modra <amodra@gmail.com>
9644 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
9645 match gnu-user.h startfile.
9646 (ENDFILE_LINUX_SPEC): Similarly.
9648 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
9651 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
9653 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
9655 2017-08-11 Tamar Christina <tamar.christina@arm.com>
9656 * config/aarch64/aarch64.md (mov<mode>): Change.
9657 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9658 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
9659 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
9661 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
9663 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
9664 for storage order barriers.
9666 2017-08-11 Martin Liska <mliska@suse.cz>
9669 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
9670 variables of void type.
9672 2017-08-11 Martin Liska <mliska@suse.cz>
9674 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
9675 TARGET_SUPPORTS_ALIASES.
9676 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
9677 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
9678 (optimize_weakref): Likewise.
9679 * symtab.c (symtab_node::noninterposable_alias): Likewise.
9680 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
9681 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
9683 2017-08-11 Martin Liska <mliska@suse.cz>
9686 * config/i386/i386.c (make_resolver_func): Do complete
9687 refactoring of the function.
9689 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9692 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
9693 * config/i386/i386.c (ix86_stack_protect_guard): Use
9694 ix86_stack_protect_guard_symbol_str to generate varible declaration.
9695 * doc/invoke.texi (x86 Options): Document
9696 -mstack-protector-guard-symbol= option.
9698 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9700 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
9701 * config/i386/i386.c (ix86_split_stack_guard): New function.
9702 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
9703 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
9704 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
9705 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
9706 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
9707 (split_stack_space_check): Call ix86_split_stack_guard.
9709 2017-08-10 Martin Sebor <msebor@redhat.com>
9711 * print-tree.c (print_node): Print location using the established
9713 Replace spaces with colons.
9714 (debug_raw, debug): Ditto.
9716 2017-08-10 Martin Sebor <msebor@redhat.com>
9719 * pretty-print.c (pp_format): Correct the handling of %s precision.
9721 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
9724 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9726 (ix86_finalize_stack_frame_flags): This. Also clear
9727 frame_pointer_needed if -fno-omit-frame-pointer is used without
9729 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9730 with ix86_finalize_stack_frame_flags.
9731 (ix86_expand_epilogue): Likewise.
9732 (ix86_expand_split_stack_prologue): Likewise.
9733 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
9735 2017-08-10 Martin Liska <mliska@suse.cz>
9738 * c-attribs.c (handle_target_attribute):
9739 Report warning for an empty string argument of target attribute.
9741 2017-08-09 Jakub Jelinek <jakub@redhat.com>
9744 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
9746 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
9747 or DECL_NONLOCAL labels.
9748 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
9749 or DECL_NONLOCAL labels here.
9751 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
9753 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
9754 to indicate when early gimple folding has been disabled.
9755 (rs6000_gimple_fold_builtin): Add debug content.
9756 (rs6000_invalid_builtin): Fix whitespace.
9757 (rs6000_expand_builtin): Fix whitespace.
9758 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
9760 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
9763 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
9764 SAVE_MULTIPLE if not all the registers that saves, should be saved.
9766 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
9768 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
9769 (qdf24xx): Likewise.
9770 * config/aarch64/aarch64.md: Include falkor.md.
9771 * config/aarch64/falkor.md: New.
9773 2017-08-09 Marek Polacek <polacek@redhat.com>
9776 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
9777 * diagnostic.c (emit_diagnostic): Add a comment.
9778 (emit_diagnostic_valist): New function.
9780 2017-08-09 Marek Polacek <polacek@redhat.com>
9783 * input.c (make_location): New overload.
9784 * input.h (make_location): Declare.
9786 2017-08-08 Alan Modra <amodra@gmail.com>
9787 H.J. Lu <hongjiu.lu@intel.com>
9790 * gcc.c (NO_PIE_SPEC): Delete.
9791 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
9793 (LINK_PIE_SPEC): ..to here.
9794 (LINK_COMMAND_SPEC): Support -no-pie.
9795 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
9796 chain of crtbegin*.o selection, update for PIE_SPEC changes and
9798 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
9799 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
9800 (ENDFILE_CRTEND_SPEC): Similarly.
9802 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
9805 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
9806 (mstack-protector-guard-offset=): Ditto.
9807 * config/i386/i386.c (ix86_option_override): Handle
9808 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
9810 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
9811 ix86_stack_protect_guard_offset variables.
9812 (TARGET_STACK_PROTECT_GUARD): Always define.
9813 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
9814 and -mstack-protector-guard-offset= options.
9816 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9818 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
9819 boundary case for the last candidate.
9821 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9823 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
9824 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
9826 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9829 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
9830 * config/aarch64/aarch64-builtins.c
9831 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
9832 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
9833 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
9835 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9836 Andrew Pinski <pinskia@gmail.com>
9839 * internal-fn.def (XORSIGN): New.
9840 * optabs.def (xorsign_optab): New.
9841 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
9842 (convert_expand_mult_copysign): New.
9843 (pass_optimize_widening_mul::execute): Call
9844 convert_expand_mult_copysign.
9846 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9848 PR tree-optimization/81354
9849 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
9850 Insert on edges rather than explicitly creating landing pads.
9851 (analyze_candidates_and_replace): Commit edge inserts.
9853 2017-08-08 Richard Biener <rguenther@suse.de>
9856 * tree-ssa-loop-niter.c: Include tree-dfa.h.
9857 (expand_simple_operations): Also look through ADDR_EXPRs with
9858 MEM_REF bases treating them as POINTER_PLUS_EXPR.
9860 2017-08-08 Richard Biener <rguenther@suse.de>
9862 PR tree-optimization/81723
9863 * tree-vect-slp.c (struct bst_traits): New hash traits.
9864 (bst_fail): New global.
9865 (vect_build_slp_tree_2): New worker, split out from ...
9866 (vect_build_slp_tree): ... this now wrapping it with using
9867 bst_fail set to cache SLP tree build fails. Properly handle
9869 (vect_analyze_slp_instance): Allocate and free bst_fail.
9871 2017-08-08 Martin Liska <mliska@suse.cz>
9874 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
9875 LABEL_DECLs that can be from a different function.
9877 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9879 PR tree-optimization/81744
9880 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
9881 loop's number of iterations.
9883 2017-08-08 Martin Liska <mliska@suse.cz>
9885 * asan.c: Include header files.
9886 * attribs.c (build_decl_attribute_variant): New function moved
9888 (build_type_attribute_qual_variant): Likewise.
9889 (cmp_attrib_identifiers): Likewise.
9890 (simple_cst_list_equal): Likewise.
9891 (omp_declare_simd_clauses_equal): Likewise.
9892 (attribute_value_equal): Likewise.
9893 (comp_type_attributes): Likewise.
9894 (build_type_attribute_variant): Likewise.
9895 (lookup_ident_attribute): Likewise.
9896 (remove_attribute): Likewise.
9897 (merge_attributes): Likewise.
9898 (merge_type_attributes): Likewise.
9899 (merge_decl_attributes): Likewise.
9900 (merge_dllimport_decl_attributes): Likewise.
9901 (handle_dll_attribute): Likewise.
9902 (attribute_list_equal): Likewise.
9903 (attribute_list_contained): Likewise.
9904 * attribs.h (lookup_attribute): New function moved from tree.[ch].
9905 (lookup_attribute_by_prefix): Likewise.
9906 * bb-reorder.c: Include header files.
9907 * builtins.c: Likewise.
9908 * calls.c: Likewise.
9909 * cfgexpand.c: Likewise.
9910 * cgraph.c: Likewise.
9911 * cgraphunit.c: Likewise.
9912 * convert.c: Likewise.
9913 * dwarf2out.c: Likewise.
9914 * final.c: Likewise.
9915 * fold-const.c: Likewise.
9916 * function.c: Likewise.
9917 * gimple-expr.c: Likewise.
9918 * gimple-fold.c: Likewise.
9919 * gimple-pretty-print.c: Likewise.
9920 * gimple.c: Likewise.
9921 * gimplify.c: Likewise.
9922 * hsa-common.c: Likewise.
9923 * hsa-gen.c: Likewise.
9924 * internal-fn.c: Likewise.
9925 * ipa-chkp.c: Likewise.
9926 * ipa-cp.c: Likewise.
9927 * ipa-devirt.c: Likewise.
9928 * ipa-fnsummary.c: Likewise.
9929 * ipa-inline.c: Likewise.
9930 * ipa-visibility.c: Likewise.
9932 * lto-cgraph.c: Likewise.
9933 * omp-expand.c: Likewise.
9934 * omp-general.c: Likewise.
9935 * omp-low.c: Likewise.
9936 * omp-offload.c: Likewise.
9937 * omp-simd-clone.c: Likewise.
9938 * opts-global.c: Likewise.
9939 * passes.c: Likewise.
9940 * predict.c: Likewise.
9941 * sancov.c: Likewise.
9942 * sanopt.c: Likewise.
9943 * symtab.c: Likewise.
9944 * toplev.c: Likewise.
9945 * trans-mem.c: Likewise.
9946 * tree-chkp.c: Likewise.
9947 * tree-eh.c: Likewise.
9948 * tree-into-ssa.c: Likewise.
9949 * tree-object-size.c: Likewise.
9950 * tree-parloops.c: Likewise.
9951 * tree-profile.c: Likewise.
9952 * tree-ssa-ccp.c: Likewise.
9953 * tree-ssa-live.c: Likewise.
9954 * tree-ssa-loop.c: Likewise.
9955 * tree-ssa-sccvn.c: Likewise.
9956 * tree-ssa-structalias.c: Likewise.
9957 * tree-ssa.c: Likewise.
9958 * tree-streamer-in.c: Likewise.
9959 * tree-vectorizer.c: Likewise.
9960 * tree-vrp.c: Likewise.
9962 * ubsan.c: Likewise.
9963 * varasm.c: Likewise.
9964 * varpool.c: Likewise.
9965 * tree.c: Remove functions moved to attribs.[ch].
9967 * config/aarch64/aarch64.c: Add attrs.h header file.
9968 * config/alpha/alpha.c: Likewise.
9969 * config/arc/arc.c: Likewise.
9970 * config/arm/arm.c: Likewise.
9971 * config/avr/avr.c: Likewise.
9972 * config/bfin/bfin.c: Likewise.
9973 * config/c6x/c6x.c: Likewise.
9974 * config/cr16/cr16.c: Likewise.
9975 * config/cris/cris.c: Likewise.
9976 * config/darwin.c: Likewise.
9977 * config/epiphany/epiphany.c: Likewise.
9978 * config/fr30/fr30.c: Likewise.
9979 * config/frv/frv.c: Likewise.
9980 * config/ft32/ft32.c: Likewise.
9981 * config/h8300/h8300.c: Likewise.
9982 * config/i386/winnt.c: Likewise.
9983 * config/ia64/ia64.c: Likewise.
9984 * config/iq2000/iq2000.c: Likewise.
9985 * config/lm32/lm32.c: Likewise.
9986 * config/m32c/m32c.c: Likewise.
9987 * config/m32r/m32r.c: Likewise.
9988 * config/m68k/m68k.c: Likewise.
9989 * config/mcore/mcore.c: Likewise.
9990 * config/microblaze/microblaze.c: Likewise.
9991 * config/mips/mips.c: Likewise.
9992 * config/mmix/mmix.c: Likewise.
9993 * config/mn10300/mn10300.c: Likewise.
9994 * config/moxie/moxie.c: Likewise.
9995 * config/msp430/msp430.c: Likewise.
9996 * config/nds32/nds32-isr.c: Likewise.
9997 * config/nds32/nds32.c: Likewise.
9998 * config/nios2/nios2.c: Likewise.
9999 * config/nvptx/nvptx.c: Likewise.
10000 * config/pa/pa.c: Likewise.
10001 * config/pdp11/pdp11.c: Likewise.
10002 * config/powerpcspe/powerpcspe.c: Likewise.
10003 * config/riscv/riscv.c: Likewise.
10004 * config/rl78/rl78.c: Likewise.
10005 * config/rx/rx.c: Likewise.
10006 * config/s390/s390.c: Likewise.
10007 * config/sh/sh.c: Likewise.
10008 * config/sol2.c: Likewise.
10009 * config/sparc/sparc.c: Likewise.
10010 * config/spu/spu.c: Likewise.
10011 * config/stormy16/stormy16.c: Likewise.
10012 * config/tilegx/tilegx.c: Likewise.
10013 * config/tilepro/tilepro.c: Likewise.
10014 * config/v850/v850.c: Likewise.
10015 * config/vax/vax.c: Likewise.
10016 * config/visium/visium.c: Likewise.
10017 * config/xtensa/xtensa.c: Likewise.
10019 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10022 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
10023 constraints since the -mupper-regs-* switches have been
10025 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
10026 into a vector from a double word element that was extracted from
10027 another vector, and eliminate extra XXPERMDI instructions.
10028 (vsx_concat_<mode>_2): Likewise.
10029 (vsx_concat_<mode>_3): Likewise.
10030 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
10031 concat to allow optimizing inserts from previous extracts.
10033 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
10035 * config/i386/i386.c (ix86_stack_protect_guard): Generate
10036 memory reference to a SSP offset in TLS address space.
10037 (ix86_print_operand) <case '@'>: Remove.
10038 (ix86_print_operand_punct_valid_p): Remove '@' code.
10039 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
10040 UNSPEC_SP_TLS_TEST.
10041 (stack_tls_protect_set_<mode>): Remove.
10042 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
10043 (stack_tls_protect_test_<mode>): Remove.
10044 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
10046 2017-08-07 Olivier Hainque <hainque@adacore.com>
10049 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
10051 2017-08-07 Douglas Rupp <rupp@adacore.com>
10053 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
10054 variable was referenced as multidir in command.
10056 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10059 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
10062 2017-08-07 Martin Liska <mliska@suse.cz>
10064 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
10065 * config/rl78/rl78.c: Add include of attribs.h.
10066 * config/sh/sh.c: Likewise.
10067 * config/v850/v850.c: Likewise.
10069 2017-08-07 Tom de Vries <tom@codesourcery.com>
10071 PR middle-end/78266
10072 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
10074 2017-08-07 Martin Liska <mliska@suse.cz>
10076 * config/mips/mips.c: Include attribs.h.
10078 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
10081 * doc/invoke.texi: Document change in behvaior for -Ofast for
10084 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
10086 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
10088 (aarch64_pop_regs): Likewise.
10089 (aarch64_gen_load_pair): Likewise.
10090 (aarch64_save_callee_saves): Likewise.
10091 (aarch64_restore_callee_saves): Likewise.
10093 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10095 * config/i386/i386.c: Revert the last change.
10097 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10100 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10102 (ix86_finalize_stack_frame_flags): This. Also clear
10103 frame_pointer_needed if -fno-omit-frame-pointer is used without
10105 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10106 with ix86_finalize_stack_frame_flags.
10107 (ix86_expand_epilogue): Likewise.
10108 (ix86_expand_split_stack_prologue): Likewise.
10110 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10113 * config/i386/i386.c (get_builtin_code_for_version): Set priority
10114 to P_AES for Westmere.
10116 2017-08-07 Jonathan Yong <10walls@gmail.com>
10118 * config/i386/mingw.opt (fset-stack-executable): Removed.
10119 * config/i386/cygming.opt (fset-stack-executable): Moved
10121 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
10123 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
10125 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
10127 2017-08-07 Marek Polacek <polacek@redhat.com>
10129 PR middle-end/81737
10130 * fold-const.c (fold_indirect_ref_1): Check type_domain.
10132 2017-08-07 Martin Liska <mliska@suse.cz>
10134 * attribs.h (canonicalize_attr_name): New function.
10135 (cmp_attribs): Move from c-format.c and adjusted.
10136 (is_attribute_p): Moved from tree.h.
10137 * tree-inline.c: Add new includes.
10138 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
10139 (private_is_attribute_p): Remove.
10140 (private_lookup_attribute): Likewise.
10141 (private_lookup_attribute_by_prefix): Simplify.
10142 (remove_attribute): Use is_attribute_p.
10143 * tree.h: Remove removed declarations.
10145 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10147 PR middle-end/81698
10148 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
10149 instead of computing it in the function. Formatting fix.
10150 (expand_case): Don't rely on default_edge being the first edge,
10151 clear it if removing it, pass default_edge to
10152 emit_case_dispatch_table.
10153 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
10156 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
10158 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
10159 insn in the function, emit NOP after the insn.
10161 2017-08-06 Tom de Vries <tom@codesourcery.com>
10163 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
10166 2017-08-06 Tom de Vries <tom@codesourcery.com>
10168 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
10171 2017-08-04 Yury Gribov <tetra2005@gmail.com>
10173 PR tree-optimization/57371
10174 * match.pd: New pattern.
10176 2017-08-04 Marek Polacek <polacek@redhat.com>
10178 PR middle-end/81695
10179 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
10180 perform the computation in offset_int.
10182 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10184 PR tree-optimization/81136
10185 * tree-vectorizer.h: Include tree-hash-traits.h.
10186 (vec_base_alignments): New typedef.
10187 (vec_info): Add a base_alignments field.
10188 (vect_record_base_alignments): Declare.
10189 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
10191 (DR_IS_CONDITIONAL_IN_STMT): New macro.
10192 (create_data_ref): Add an is_conditional_in_stmt argument.
10193 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
10194 the is_conditional_in_stmt field.
10195 (data_ref_loc): Add an is_conditional_in_stmt field.
10196 (get_references_in_stmt): Set the is_conditional_in_stmt field.
10197 (find_data_references_in_stmt): Update call to create_data_ref.
10198 (graphite_find_data_references_in_stmt): Likewise.
10199 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
10200 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
10201 (vect_record_base_alignment): New function.
10202 (vect_record_base_alignments): Likewise.
10203 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
10204 for nested statements even if we fail to compute a misalignment.
10205 Use pooled base alignments for unconditional references.
10206 (vect_find_same_alignment_drs): Compare base addresses instead
10208 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
10209 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
10211 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10213 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
10214 Add an explicit name for the enum. Use auto_vec for slp_instances
10215 and grouped_stores.
10216 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
10218 (_bb_vec_info): Add a constructor and destructor.
10219 (vinfo_for_stmt): Return NULL for uids of -1 as well.
10220 (destroy_loop_vec_info): Delete.
10221 (vect_destroy_datarefs): Likewise.
10222 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
10223 (vec_info::vec_info): New function.
10224 (vec_info::~vec_info): Likewise.
10225 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
10226 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
10227 destroy_loop_vec_info.
10228 * tree-vect-loop.c (new_loop_vec_info): Replace with...
10229 (_loop_vec_info::_loop_vec_info): ...this.
10230 (destroy_loop_vec_info): Replace with...
10231 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
10232 the stmt_vec_infos. Leave handling of vec_info information to its
10233 destructor. Remove explicit vector releases.
10234 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
10235 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
10236 * tree-vect-slp.c (new_bb_vec_info): Replace with...
10237 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
10238 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
10239 (destroy_bb_vec_info): Replace with...
10240 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
10241 information to its destructor.
10242 (vect_slp_analyze_bb_1): Use new and delete instead of
10243 new_bb_vec_info and destroy_bb_vec_info.
10244 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
10247 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10249 * tree-data-ref.h (subscript): Add access_fn field.
10250 (data_dependence_relation): Add could_be_independent_p.
10251 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
10252 (same_access_functions): Move to tree-data-ref.c.
10253 * tree-data-ref.c (ref_contains_union_access_p): New function.
10254 (access_fn_component_p): Likewise.
10255 (access_fn_components_comparable_p): Likewise.
10256 (dr_analyze_indices): Add a reference to access_fn_component_p.
10257 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
10259 (constant_access_functions): Likewise.
10260 (add_other_self_distances): Likewise.
10261 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
10262 (initialize_data_dependence_relation): Use XCNEW and remove
10263 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
10264 of access functions that have the same type. Allow the
10265 subsequence to end with different bases in some circumstances.
10266 Record the chosen access functions in SUB_ACCESS_FN.
10267 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
10268 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
10269 (subscript_dependence_tester_1): Likewise dra and drb.
10270 (build_classic_dist_vector): Update calls accordingly.
10271 (subscript_dependence_tester): Likewise.
10272 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
10273 DDR_COULD_BE_INDEPENDENT_P.
10274 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
10275 comp_alias_ddrs instead of may_alias_ddrs.
10276 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
10278 (vect_analyze_data_ref_dependence): Use it if
10279 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
10280 distance vectors if that fails.
10281 (dependence_distance_ge_vf): New function.
10282 (vect_prune_runtime_alias_test_list): Use it. Don't clear
10283 LOOP_VINFO_MAY_ALIAS_DDRS.
10285 2017-08-04 Richard Biener <rguenther@suse.de>
10287 PR middle-end/81705
10288 * fold-const.c (fold_binary_loc): Properly restrict
10289 minus_var0 && minus_var1 case when associating undefined overflow
10292 2017-08-04 Tom de Vries <tom@codesourcery.com>
10294 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
10296 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10298 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10299 Don't start diagnostic messages with a capital letter.
10300 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10302 (rs6000_invalid_builtin): Likewise.
10303 (rs6000_trampoline_init): Likewise.
10305 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10308 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
10309 after setting changeable df flags.
10311 2017-08-03 Richard Biener <rguenther@suse.de>
10313 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
10314 up if the use is in USE - X.
10316 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
10318 * toplev.c (dumpfile.h): New include.
10319 (internal_error_reentered): New static function. Use it...
10320 (internal_error_function): ...here to handle reentered internal_error.
10322 2017-08-03 Richard Biener <rguenther@suse.de>
10324 PR middle-end/81148
10325 * fold-const.c (split_tree): Add minus_var and minus_con
10326 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
10327 here but always use minus_*.
10328 (associate_trees): Assert we never associate with MINUS_EXPR
10329 and NULL first operand. Do not recurse for PLUS_EXPR operands
10330 when associating as MINUS_EXPR either.
10331 (fold_binary_loc): Track minus_var and minus_con.
10333 2017-08-03 Tom de Vries <tom@codesourcery.com>
10336 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
10337 ACCEL_COMPILER, apply finish_options on
10338 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
10340 2017-08-03 Tom de Vries <tom@codesourcery.com>
10343 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
10344 function_entry_patch_area_size > 0.
10346 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10349 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
10350 instead of 10??LU, perform unit multiplication in wide_int,
10351 don't change alloc_object_size_limit if the limit is larger
10354 PR tree-optimization/81655
10355 PR tree-optimization/81588
10356 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
10357 the case when ranges[i].low and high are 1 for unsigned type with
10360 PR middle-end/81052
10361 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
10362 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
10364 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10366 * tree-vrp.h: Add include guard.
10368 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
10371 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
10372 (ud2): New insn pattern.
10373 * config/i386/i386.c (ix86_expand_epilogue):
10374 For naked functions, generate ud2 instead of trap insn.
10376 2017-08-02 Marek Polacek <polacek@redhat.com>
10379 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
10381 2017-08-02 Tom de Vries <tom@codesourcery.com>
10382 Cesar Philippidis <cesar@codesourcery.com>
10384 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
10385 Add missing edge probabilities.
10387 2017-08-02 Tamar Christina <tamar.christina@arm.com>
10389 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
10390 Correct endianness.
10392 2017-08-02 Jakub Jelinek <jakub@redhat.com>
10394 PR middle-end/79499
10395 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
10396 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
10397 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
10398 prologue_seq sequences - if any.
10400 2017-08-02 Richard Biener <rguenther@suse.de>
10402 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
10403 via vectors if supported, integer extracts via punning if supported
10404 or otherwise vector extracts.
10406 2017-08-02 Richard Biener <rguenther@suse.de>
10408 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
10410 (bitmap_insert_into_set): ... this.
10412 2017-08-02 Richard Biener <rguenther@suse.de>
10414 PR tree-optimization/81633
10416 2015-08-17 Alan Hayward <alan.hayward@arm.com>
10418 PR tree-optimization/71752
10419 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
10421 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
10423 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
10424 (machine_function::call_ms2sysv_pad_out): Remove field.
10425 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
10426 (ix86_compute_frame_layout): Likewise.
10428 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
10431 * config/i386/i386.c (ix86_set_func_type): Disallow naked
10432 attribute with interrupt attribute.
10434 2017-08-01 Andrew Pinski <apinski@cavium.com>
10436 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
10437 BIT_INSERT_EXPR's operand 1
10438 to see if the types precision matches.
10440 2017-08-01 Martin Liska <mliska@suse.cz>
10442 PR middle-end/70140
10443 * builtins.c (expand_builtin_memcpy_args): Remove.
10444 (expand_builtin_memcpy): Call newly added function
10445 expand_builtin_memory_copy_args.
10446 (expand_builtin_memcpy_with_bounds): Likewise.
10447 (expand_builtin_mempcpy): Remove last argument.
10448 (expand_builtin_mempcpy_with_bounds): Likewise.
10449 (expand_builtin_memory_copy_args): New function created from
10450 expand_builtin_mempcpy_args with small modifications.
10451 (expand_builtin_mempcpy_args): Remove.
10452 (expand_builtin_stpcpy): Remove unused argument.
10453 (expand_builtin): Likewise.
10454 (expand_builtin_with_bounds): Likewise.
10456 2017-08-01 Martin Liska <mliska@suse.cz>
10459 Make mempcpy more optimal (PR middle-end/70140).
10461 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10464 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10465 __builtin_vec_cmpne verify both arguments are compatible vectors
10466 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
10467 verify arg1_type is a pointer or array type. For __builtin_vec_st,
10468 move computation of aligned to after checking the argument types.
10472 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10473 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
10476 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
10477 Doug Rupp <rupp@adacore.com>
10478 Olivier Hainque <hainque@adacore.com>
10480 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
10481 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
10483 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
10484 for TARGET_OS_CPP_BUILTIN.
10485 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
10486 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
10488 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
10489 passing required abi options to the assembler for EABI configurations.
10490 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
10491 of .text.hot and .text.unlikely sections for kernel modules when
10492 using ARM style exceptions.
10493 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
10494 options. Add EXTRA_CC1_SPEC.
10495 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
10497 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
10499 (ARM_TARGET2_DWARF_FORMAT): Define.
10500 * config/arm/t-vxworks: Adjust multilib control to removal of the
10501 Diab command line options.
10503 2017-08-01 Martin Liska <mliska@suse.cz>
10505 PR gcov-profile/81561
10506 * gcov.c (unblock): Make unblocking safe as we need to preserve
10507 index correspondence of blocks and block_lists.
10509 2017-08-01 Richard Biener <rguenther@suse.de>
10511 PR tree-optimization/81181
10512 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10513 (compute_antic): ... end of iteration here.
10515 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
10517 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
10518 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
10519 (ftree-slp-vectorize): Likewise.
10520 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
10521 can no longer be set independent of flag_tree_loop_vectorize.
10522 * omp-general.c (emp_max_vf): Likewise.
10523 * opts.c (enable_fdo_optimizations): Remove references to
10524 flag_tree_vectorize, these are now implicit.
10525 (common_handle_option): Remove handling for OPT_ftree_vectorize,
10526 and leave it for the options machinery.
10528 2017-08-01 Martin Liska <mliska@suse.cz>
10530 PR middle-end/70140
10531 * builtins.c (expand_builtin_memcpy_args): Remove.
10532 (expand_builtin_memcpy): Call newly added function
10533 expand_builtin_memory_copy_args.
10534 (expand_builtin_memcpy_with_bounds): Likewise.
10535 (expand_builtin_mempcpy): Remove last argument.
10536 (expand_builtin_mempcpy_with_bounds): Likewise.
10537 (expand_builtin_memory_copy_args): New function created from
10538 expand_builtin_mempcpy_args with small modifications.
10539 (expand_builtin_mempcpy_args): Remove.
10540 (expand_builtin_stpcpy): Remove unused argument.
10541 (expand_builtin): Likewise.
10542 (expand_builtin_with_bounds): Likewise.
10544 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10547 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
10548 print "ds:" only for immediates in generic address space.
10550 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10553 * config/i386/i386.c (ix86_funciton_naked): New prototype.
10554 (ix86_function_ok_for_sibcall): Return false for naked functions.
10556 2017-08-01 Richard Biener <rguenther@suse.de>
10558 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
10559 (compute_antic): Seed worklist with exit block predecessors.
10560 * cfganal.c (dfs_find_deadend): For a cycle return the source
10561 of the edge closing it.
10563 2017-08-01 Tamar Christina <tamar.christina@arm.com>
10565 * config/aarch64/aarch64.c
10566 (aarch64_can_const_movi_rtx_p): Move 0 check.
10568 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10570 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
10571 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
10573 * match.pd: Ditto in address comparison pattern.
10575 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10577 PR tree-optimization/81627
10578 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
10579 closed ssa form for store-store chain.
10581 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10583 PR tree-optimization/81620
10584 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
10585 for store-store chain.
10587 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10589 PR tree-optimization/81588
10590 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
10591 ranges[i].in_p, invert comparison code ccode. For >/>=,
10592 swap rhs1 and rhs2 and comparison code unconditionally,
10593 for </<= don't do that. Don't swap rhs1/rhs2 again if
10594 ranges[i].in_p, instead invert comparison code ccode if
10595 opcode or oe->rank is BIT_IOR_EXPR.
10598 * optabs.def (vec_extract_optab, vec_init_optab): Change from
10599 a direct optab to conversion optab.
10600 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
10601 with GET_MODE_INNER as last argument instead of optab_handler.
10602 * expmed.c (extract_bit_field_1): Likewise. Use vector from
10603 vector extraction if possible and optab is available.
10604 * expr.c (store_constructor): Use convert_optab_handler instead
10605 of optab_handler. Use vector initialization from smaller
10606 vectors if possible and optab is available.
10607 * tree-vect-stmts.c (vectorizable_load): Likewise.
10608 * doc/md.texi (vec_extract, vec_init): Document that the optabs
10609 now have two modes.
10610 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
10611 of vec_init from half-sized vectors with the same element mode.
10612 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
10613 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
10614 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
10615 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
10616 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
10617 after mode in gen_vec_extract* calls.
10618 (vec_extract<mode>): Renamed to ...
10619 (vec_extract<mode><ssescalarmodelower>): ... this.
10620 (vec_extract<mode><ssehalfvecmodelower>): New expander.
10621 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
10622 element mode after mode in gen_vec_init* calls.
10623 (VEC_INIT_HALF_MODE): New mode iterator.
10624 (vec_init<mode>): Renamed to ...
10625 (vec_init<mode><ssescalarmodelower>): ... this.
10626 (vec_init<mode><ssehalfvecmodelower>): New expander.
10627 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
10628 (vec_extractv2sfsf): ... this.
10629 (vec_initv2sf): Renamed to ...
10630 (vec_initv2sfsf): ... this.
10631 (vec_extractv2si): Renamed to ...
10632 (vec_extractv2sisi): ... this.
10633 (vec_initv2si): Renamed to ...
10634 (vec_initv2sisi): ... this.
10635 (vec_extractv4hi): Renamed to ...
10636 (vec_extractv4hihi): ... this.
10637 (vec_initv4hi): Renamed to ...
10638 (vec_initv4hihi): ... this.
10639 (vec_extractv8qi): Renamed to ...
10640 (vec_extractv8qiqi): ... this.
10641 (vec_initv8qi): Renamed to ...
10642 (vec_initv8qiqi): ... this.
10643 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
10644 (vec_init<mode>): Renamed to ...
10645 (vec_init<mode><VEC_base_l>): ... this.
10646 (vec_extract<mode>): Renamed to ...
10647 (vec_extract<mode><VEC_base_l>): ... this.
10648 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
10649 (vec_initv2sfsf): ... this.
10650 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
10651 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10652 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
10653 element mode after mode in gen_vec_init* calls.
10654 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
10655 (vec_init<mode><Vel>): ... this.
10656 (vec_extract<mode>): Renamed to ...
10657 (vec_extract<mode><Vel>): ... this.
10658 * config/aarch64/iterators.md (Vel): New mode attribute.
10659 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
10660 Add element mode after mode in gen_vec_extract* calls.
10661 * config/s390/vector.md (non_vec_l): New mode attribute.
10662 (vec_extract<mode>): Renamed to ...
10663 (vec_extract<mode><non_vec_l>): ... this.
10664 (vec_init<mode>): Renamed to ...
10665 (vec_init<mode><non_vec_l>): ... this.
10666 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
10667 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
10669 * config/arm/iterators.md (V_elem_l): New mode attribute.
10670 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
10671 (vec_extract<mode><V_elem_l>): ... this.
10672 (vec_extractv2di): Renamed to ...
10673 (vec_extractv2didi): ... this.
10674 (vec_init<mode>): Renamed to ...
10675 (vec_init<mode><V_elem_l>): ... this.
10676 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
10677 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
10678 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
10679 Add element mode after gen_vec_extract* calls.
10680 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
10681 (vec_init<mode><unitmode>): ... this.
10682 (vec_extract<mode>): Renamed to ...
10683 (vec_extract<mode><unitmode>): ... this.
10684 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
10685 (vec_init<mode><unitmode>): ... this.
10686 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
10687 (vec_initv2sfsf): ... this.
10688 (vec_extractv2sf): Renamed to ...
10689 (vec_extractv2sfsf): ... this.
10690 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
10691 Add element mode after gen_vec_extract* calls.
10692 * config/mips/mips.md (unitmode): New mode iterator.
10693 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
10694 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
10695 * config/spu/spu.md (inner_l): New mode attribute.
10696 (vec_init<mode>): Renamed to ...
10697 (vec_init<mode><inner_l>): ... this.
10698 (vec_extract<mode>): Renamed to ...
10699 (vec_extract<mode><inner_l>): ... this.
10700 * config/sparc/sparc.md (veltmode): New mode iterator.
10701 (vec_init<VMALL:mode>): Renamed to ...
10702 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
10703 * config/ia64/vect.md (vec_initv2si): Renamed to ...
10704 (vec_initv2sisi): ... this.
10705 (vec_initv2sf): Renamed to ...
10706 (vec_initv2sfsf): ... this.
10707 (vec_extractv2sf): Renamed to ...
10708 (vec_extractv2sfsf): ... this.
10709 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
10710 (vec_init<mode>): Renamed to ...
10711 (vec_init<mode><VEC_base_l>): ... this.
10712 (vec_extract<mode>): Renamed to ...
10713 (vec_extract<mode><VEC_base_l>): ... this.
10714 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
10715 (vec_initv2sfsf): ... this.
10716 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
10717 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10718 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
10719 gen_vec_init* calls.
10721 2017-08-01 Richard Biener <rguenther@suse.de>
10723 PR tree-optimization/81297
10724 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
10725 TREE_OVERFLOW from INTEGER_CSTs.
10727 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
10729 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
10731 2017-07-31 Carl Love <cel@us.ibm.com>
10733 * config/rs6000/rs6000-c: Add support for built-in functions
10734 vector signed char vec_xl_be (signed long long, signed char *);
10735 vector unsigned char vec_xl_be (signed long long, unsigned char *);
10736 vector signed int vec_xl_be (signed long long, signed int *);
10737 vector unsigned int vec_xl_be (signed long long, unsigned int *);
10738 vector signed long long vec_xl_be (signed long long, signed long long *);
10739 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
10740 vector signed short vec_xl_be (signed long long, signed short *);
10741 vector unsigned short vec_xl_be (signed long long, unsigned short *);
10742 vector double vec_xl_be (signed long long, double *);
10743 vector float vec_xl_be (signed long long, float *);
10744 * config/rs6000/altivec.h (vec_xl_be): Add #define.
10745 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
10746 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
10748 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
10749 (altivec_expand_builtin): Add switch statement to call
10750 altivec_expand_xl_be for each builtin.
10751 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
10752 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
10753 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
10754 __builtin_vsx_le_be_v16qi.
10755 * doc/extend.texi: Update the built-in documentation file for the
10756 new built-in functions.
10758 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10761 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
10763 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10765 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10767 * config.gcc: Add z14.
10768 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
10769 CPU model numbers for z13s and z14.
10770 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
10772 * config/s390/s390-opts.h (enum processor_type): Rename
10773 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10774 * config/s390/s390.c (processor_table): Add field for CPU name to
10775 be passed to Binutils.
10776 (s390_asm_output_machine_for_arch): Use the new field in
10777 processor_table for Binutils.
10778 (s390_expand_builtin): Replace arch12 with z14.
10779 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10780 (s390_get_sched_attrmask): Likewise.
10781 (s390_get_unit_mask): Likewise.
10782 * config/s390/s390.opt: Add z14 to processor_type enum.
10784 2017-07-31 Martin Jambor <mjambor@suse.cz>
10787 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
10788 regardless of optimization level.
10790 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
10791 Martin Liska <mliska@suse.cz>
10793 * predict.def: Remove old comment and adjust probability.
10794 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
10795 PREDICT statements.
10797 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10800 * config/i386/i386.c (ix86_function_naked): New function.
10801 (ix86_can_use_return_insn_p): Return false for naked functions.
10802 (ix86_expand_prologue): Skip prologue for naked functions.
10803 (ix86_expand_epilogue): Skip epilogue for naked functions
10804 and emit trap instruction.
10805 (ix86_warn_func_return): New function.
10806 (ix86_attribute_table): Add "naked" attribute specification.
10807 (TARGET_WARN_FUNC_RETURN): Define.
10808 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
10810 2017-07-31 Martin Liska <mliska@suse.cz>
10812 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
10813 (dump_gimple_bb_header): Always dump BB info.
10814 (pp_cfg_jump): Do not append info about BB when dumping a jump.
10816 2017-07-31 Martin Liska <mliska@suse.cz>
10819 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
10820 also with current_function_decl non-null equality.
10822 2017-07-31 Jakub Jelinek <jakub@redhat.com>
10825 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
10826 change type to the element type, instead add eltype variable and
10827 use it where we are interested in the element type.
10829 PR tree-optimization/81603
10830 * ipa-polymorphic-call.c
10831 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
10832 offset arithmetic in offset_int, bail out if the resulting bit offset
10833 doesn't fit into shwi.
10835 2017-07-31 Martin Liska <mliska@suse.cz>
10837 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
10838 (gimplify_save_expr): Fix comment.
10840 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
10843 * config/i386/i386.c (ix86_function_arg): Update arguments for
10845 (ix86_compute_frame_layout): Set the initial stack offset to
10846 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
10847 INCOMING_FRAME_SP_OFFSET.
10848 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
10849 stack before exception handler returns.
10850 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
10851 the 'ERROR_CODE' for exception handler.
10853 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
10855 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
10856 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
10857 (ASM_OUTPUT_REG_POP): Ditto.
10858 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
10859 instead of asm_fprintf to output pure string.
10861 2017-07-29 Jakub Jelinek <jakub@redhat.com>
10863 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
10864 to imported_module_or_decl hook.
10865 (debug_nothing_tree_tree_tree_bool): Remove.
10866 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
10867 * debug.c (do_nothing_debug_hooks): Use
10868 debug_nothing_tree_tree_tree_bool_bool instead of
10869 debug_nothing_tree_tree_tree_bool.
10870 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10871 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
10872 * sdbout.c (sdb_debug_hooks): Likewise.
10873 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
10874 (gen_namespace_die): Add DW_AT_export_symbols attribute if
10876 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
10877 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
10878 attribute, don't add anything.
10880 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10882 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
10883 (fold_build2_stat_loc): Likewise.
10884 (fold_build3_stat_loc): Likewise.
10885 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
10886 (fold_build1_loc): Remove macro.
10887 (fold_build2_loc): Likewise.
10888 (fold_build3_loc): Likewise.
10890 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10892 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
10893 (gimple_build_debug_bind_source_stat): Likewise.
10894 * gimple.h (gimple_build_debug_bind): Remove macro.
10895 (gimple_build_debug_bind_source): Likewise.
10897 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10899 * bitmap.c (bitmap_alloc): Adjust.
10900 (bitmap_gc_alloc): Likewise.
10901 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
10903 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10905 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
10906 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
10907 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
10908 (bitmap_gc_alloc_stat): Likewise.
10909 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
10911 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10913 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
10914 * rtl.h (shallow_copy_rtx): Remove macro.
10916 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10918 * emit-rtl.c (gen_raw_REG): Adjust.
10919 * gengenrtl.c (gendef): Likewise.
10920 * rtl.c (rtx_alloc_stat): Remove _stat from name.
10921 * rtl.h (rtx_alloc): Remove macro.
10923 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10925 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
10926 (build_tree_list_stat): Likewise.
10927 * tree.h (build_tree_list): Remove macro.
10928 (build_tree_list_vec): Likewise.
10930 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10932 * tree.c (make_vector_stat): Remove _stat from name.
10933 (build_vector_stat): Likewise.
10934 * tree.h (make_vector_stat): Remove macro.
10935 (build_vector_stat): Likewise.
10937 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10939 * tree.h (build_var_debug_value): Remove prototype.
10941 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10943 * tree.c (tree_cons_stat): Remove _stat from name.
10944 * tree.h (tree_cons): Remove macro.
10946 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10948 * tree.c (build_vl_exp_stat): Remove _stat from name.
10949 * tree.h (build_vl_exp): Remove macro.
10951 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10953 * tree.c (build_decl_stat): Remove _stat from name.
10954 * tree.h (build_decl): Remove macro.
10956 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10958 * gimple.c (gimple_build_with_ops_stat): Adjust.
10959 (gimple_alloc_stat): Remove _stat from name.
10960 * gimple.h (gimple_alloc): Remove macro.
10962 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10964 * tree.c (make_tree_vec_stat): Remove _stat from name.
10965 (grow_tree_vec_stat): Likewise.
10966 * tree.h (make_tree_vec_stat): Adjust prototype.
10967 (grow_tree_vec_stat): Likewise.
10968 (make_tree_vec): Remove macro.
10969 (grow_tree_vec): Likewise.
10971 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10973 * fold-const.c (fold_build1_stat_loc): Adjust.
10974 (fold_build2_stat_loc): Likewise.
10975 (fold_build3_stat_loc): Likewise.
10976 * tree.c (build0_stat): Remove _stat from name.
10977 (build1_stat): Likewise.
10978 (build2_stat): Likewise.
10979 (build3_stat): Likewise.
10980 (build4_stat): Likewise.
10981 (build5_stat): Likewise.
10982 * tree.h (build1_loc): Remove macro, and rename _stat function
10984 (build2_loc): Likewise.
10985 (build3_loc): Likewise.
10986 (build4_loc): Likewise.
10987 (build5_loc): Likewise.
10989 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10991 * tree.c (make_int_cst_stat): Remove _stat from name.
10992 * tree.h (make_int_cst_stat): Adjust prototype.
10993 (make_int_cst): Remove macro.
10995 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10997 * tree.c (make_tre_binfo_stat): Remove _stat from name.
10998 * tree.h (make_tree_binfo_stat): Adjust prototype.
10999 (make_tree_binfo): Remove.
11001 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11003 * tree.c (copy_node_stat): Rename to copy_node.
11004 (build_distinct_type_copy): Adjust.
11005 * tree.h (copy_node_stat): Adjust prototype.
11006 (copy_node): Remove macro.
11008 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11010 * tree.c (make_node_stat): rename to make_node.
11011 (build_tree_list_stat): Adjust.
11012 (build0_stat): Likewise.
11013 (build2_stat): Likewise.
11014 (build3_stat): Likewise.
11015 (build4_stat): Likewise.
11016 (build5_stat): Likewise.
11017 (build_decl_stat): Likewise.
11018 * tree.h (make_node_stat): Adjust prototype.
11019 (make_node): remove macro.
11021 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
11023 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
11024 (PPC_FEATURE2_SCV): Likewise.
11025 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
11027 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11029 * config/aarch64/aarch64.c
11030 (aarch64_internal_mov_immediate): Add new special pattern.
11031 * config/aarch64/aarch64.md (*movdi_aarch64):
11032 Add reg/32bit const mov case.
11034 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11035 Richard Sandiford <richard.sandiford@linaro.org>
11037 * config/aarch64/aarch64.md (mov<mode>): Generalize.
11038 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11039 Add integer and movi cases.
11040 (movi-split-hf-df-sf split, fp16): New.
11041 (enabled): Added TARGET_FP_F16INST.
11042 * config/aarch64/iterators.md (GPF_HF): New.
11043 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
11045 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11047 * config/aarch64/aarch64.c
11048 (aarch64_simd_container_mode): Add prototype.
11049 (aarch64_expand_mov_immediate): Add HI support.
11050 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
11051 (aarch64_can_const_movi_rtx_p): New.
11052 (aarch64_preferred_reload_class):
11053 Remove restrictions of using FP registers for certain SIMD operations.
11054 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
11055 (aarch64_valid_floating_const): Add integer move validation.
11056 (aarch64_simd_imm_scalar_p): Remove.
11057 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
11058 (aarch64_legitimate_constant_p): Expand list of supported cases.
11059 * config/aarch64/aarch64-protos.h
11060 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
11061 (aarch64_reinterpret_float_as_int): New.
11062 (aarch64_simd_imm_scalar_p): Remove.
11063 * config/aarch64/constraints.md (Uvi): New.
11064 (Dd): Split into Ds and new Dd.
11065 * config/aarch64/aarch64.md (*movsi_aarch64):
11067 (*movdi_aarch64): Add SIMD mov case.
11069 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11071 * tree-predcom.c: (struct chain): Handle store-store chain in which
11072 stores for elimination only store loop invariant values.
11073 (execute_pred_commoning_chain): Ditto.
11074 (prepare_initializers_chain_store_elim): Ditto.
11075 (prepare_finalizers): Ditto.
11076 (is_inv_store_elimination_chain): New function.
11077 (initialize_root_vars_store_elim_1): New function.
11079 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11081 * tree-predcom.c: Revise general description of the pass.
11082 (enum chain_type): New enum type for store elimination.
11083 (struct chain): New field supporting store elimination.
11084 (struct component): Ditto.
11085 (dump_chain): Dump store-stores chain.
11086 (release_chain): Release resources.
11087 (split_data_refs_to_components): Compute and create component
11088 contains only stores for elimination.
11089 (get_chain_last_ref_at): New function.
11090 (make_invariant_chain): Initialization.
11091 (make_rooted_chain): Specify chain type in parameter and record it.
11092 (add_looparound_copies): Skip for store-stores chain.
11093 (determine_roots_comp): Compute type of chain and pass it to
11095 (initialize_root_vars_store_elim_2): New function.
11096 (finalize_eliminated_stores): New function.
11097 (remove_stmt): Handle store for elimination.
11098 (execute_pred_commoning_chain): Execute predictive commoning on
11099 store-store chains.
11100 (determine_unroll_factor): Skip unroll for store-stores chain.
11101 (prepare_initializers_chain_store_elim): New function.
11102 (prepare_initializers_chain): Hanlde store-store chain.
11103 (prepare_finalizers_chain, prepare_finalizers): New function.
11104 (tree_predictive_commoning_loop): Return integer value indicating
11105 if loop is unrolled or lcssa form is corrupted.
11106 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
11108 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11110 * tree-predcom.c (initialize_root): Delete.
11111 (execute_pred_commoning_chain): Initialize root vars and replace
11112 reference of non-combined chain directly, rather than call above
11115 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11117 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
11118 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
11120 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11122 * tree-predcom.c (struct chain): New field init_seq.
11123 (release_chain): Release init_seq.
11124 (prepare_initializers_chain): Record intialization stmts in above
11126 (insert_init_seqs): New function.
11127 (tree_predictive_commoning_loop): Call insert_init_seqs.
11129 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11131 * tree-predcom.c (determine_roots_comp): Skip trivial components.
11133 2017-07-28 Richard Biener <rguenther@suse.de>
11135 * match.pd: Remove superfluous :c.
11136 * genmatch.c (simplify::id): Add member.
11137 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
11139 (current_id): New global.
11140 (dt_node::parent): Move from ...
11141 (dt_operand::parent): ... here. Add for_id member.
11142 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
11143 (decision_tree::find_node): Relax order requirement when
11144 merging DT_TRUE nodes to ones inbetween the current simplify
11145 and the one we try to merge with. Add diagnostic whenever
11146 we need to enforce pattern order by not merging.
11147 (decision_tree::insert): Set current_id.
11148 (decision_tree::print_node): Dump parent node and for_id.
11149 (parser::last_id): Add member.
11150 (parser::push_simplify): Assign unique id.
11151 (parser::parser): Initialize last_id.
11153 2017-07-28 Martin Liska <mliska@suse.cz>
11156 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
11157 gimple_build_debug_bind.
11159 2017-07-28 Richard Biener <rguenther@suse.de>
11161 PR tree-optimization/81502
11162 * match.pd: Add pattern combining BIT_INSERT_EXPR with
11164 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
11166 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
11167 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
11168 for BIT_FIELD_REF args.
11169 * fold-const.c (make_bit_field_ref): Likewise.
11170 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
11172 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11175 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
11176 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
11177 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
11178 Or it into SANITIZER_UNDEFINED.
11179 * ubsan.c: Include gimple-fold.h and varasm.h.
11180 (ubsan_expand_ptr_ifn): New function.
11181 (instrument_pointer_overflow): New function.
11182 (maybe_instrument_pointer_overflow): New function.
11183 (instrument_object_size): Formatting fix.
11184 (pass_ubsan::execute): Call instrument_pointer_overflow
11185 and maybe_instrument_pointer_overflow.
11186 * internal-fn.c (expand_UBSAN_PTR): New function.
11187 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
11188 * sanitizer.def (__ubsan_handle_pointer_overflow,
11189 __ubsan_handle_pointer_overflow_abort): New builtins.
11190 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
11191 * internal-fn.def (UBSAN_PTR): New internal function.
11192 * opts.c (sanitizer_opts): Add pointer-overflow.
11193 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
11194 * fold-const.c (build_range_check): Compute pointer range check in
11195 integral type if pointer arithmetics would be needed. Formatting
11198 2017-07-28 Martin Liska <mliska@suse.cz>
11201 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
11202 parameters that are of a variable-length.
11204 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
11206 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
11208 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
11209 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
11210 (CRT_CALL_STATIC_FUNCTION): Likewise.
11211 (ASM_DEFAULT_SPEC): New define.
11212 (ASM_SPEC32): Likewise.
11213 (ASM_SPEC64): Likewise.
11214 (ASM_SPEC_COMMON): Likewise.
11215 (ASM_SPEC): Likewise.
11216 (INVALID_64BIT): Likewise.
11217 (LINK_OS_DEFAULT_SPEC): Likewise.
11218 (LINK_OS_SPEC32): Likewise.
11219 (LINK_OS_SPEC64): Likewise.
11220 (POWERPC_LINUX): Likewise.
11221 (PTRDIFF_TYPE): Likewise.
11222 (RESTORE_FP_PREFIX): Likewise.
11223 (RESTORE_FP_SUFFIX): Likewise.
11224 (SAVE_FP_PREFIX): Likewise.
11225 (SAVE_FP_SUFFIX): Likewise.
11226 (SIZE_TYPE): Likewise.
11227 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
11228 (TARGET_64BIT): Likewise.
11229 (TARGET_64BIT): Likewise.
11230 (TARGET_AIX): Likewise.
11231 (WCHAR_TYPE_SIZE): Likewise.
11232 (WCHAR_TYPE): Undefine.
11233 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
11234 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
11235 (CPP_OS_RTEMS_SPEC): Delete.
11236 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
11237 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
11239 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
11241 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11243 PR tree-optimization/81578
11244 * tree-parloops.c (build_new_reduction): Bail out if
11245 reduction_code isn't one of the standard OpenMP reductions.
11246 Move the details printing after that decision.
11248 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11250 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
11251 related to reload_in_progress.
11252 (splat_input_operand): Likewise.
11253 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
11255 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
11257 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
11258 (TARGET_INSTANTIATE_DECLS): Likewise.
11259 (legitimate_indexed_address_p): Delete reload_in_progress code.
11260 (rs6000_debug_legitimate_address_p): Likewise.
11261 (rs6000_eliminate_indexed_memrefs): Likewise.
11262 (rs6000_emit_le_vsx_store): Likewise.
11263 (rs6000_emit_move_si_sf_subreg): Likewise.
11264 (rs6000_emit_move): Likewise.
11265 (register_to_reg_type): Likewise.
11266 (rs6000_pre_atomic_barrier): Likewise.
11267 (rs6000_machopic_legitimize_pic_address): Likewise.
11268 (rs6000_allocate_stack_temp): Likewise.
11269 (rs6000_address_for_fpconvert): Likewise.
11270 (rs6000_address_for_altivec): Likewise.
11271 (rs6000_secondary_memory_needed_rtx): Delete function.
11272 (rs6000_check_sdmode): Likewise.
11273 (rs6000_alloc_sdmode_stack_slot): Likewise.
11274 (rs6000_instantiate_decls): Likewise.
11275 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
11276 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
11277 Delete reload_in_progress.
11278 (*vec_reload_and_plus_<mptrsize>): Likewise.
11279 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
11280 (vsx_div_v2di): Likewise.
11281 (vsx_udiv_v2di): Likewise.
11283 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11285 * config/rs6000/rs6000.opt (mlra): Replace with stub.
11286 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
11287 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
11288 (rs6000_debug_reg_global): Delete print of LRA status.
11289 (rs6000_option_override_internal): Delete dead LRA related code.
11290 (rs6000_lra_p): Delete function.
11291 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
11293 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
11295 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
11296 * config/riscv/rtems.h: New file.
11298 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11299 Sudakshina Das <sudi.das@arm.com>
11301 * config/aarch64/aarch64.md
11302 (define_split for and<mode>3nr_compare): Move
11303 non aarch64_logical_operand to a register.
11304 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
11305 register immediate operand to a register.
11306 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
11308 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11310 PR middle-end/81564
11311 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
11313 2017-07-27 Richard Biener <rguenther@suse.de>
11315 PR tree-optimization/81573
11316 PR tree-optimization/81494
11317 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
11318 multi defuse cycle case.
11320 2017-07-27 Richard Biener <rguenther@suse.de>
11322 PR tree-optimization/81571
11323 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
11326 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
11328 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
11329 earlier and only if MASK_FPU is set. Adjust formatting.
11331 2017-07-27 Martin Liska <mliska@suse.cz>
11333 * opt-functions.awk: Add validation of value of Init.
11334 * optc-gen.awk: Pass new argument.
11336 2017-07-27 Martin Liska <mliska@suse.cz>
11338 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
11339 Fix wrong condition.
11341 2017-07-27 Martin Liska <mliska@suse.cz>
11343 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
11344 BBs and edges seen by autoFDO.
11346 2017-07-27 Richard Biener <rguenther@suse.de>
11348 PR tree-optimization/81502
11349 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
11350 with incompatible but same sized type.
11351 (execute_update_addresses_taken): Likewise.
11353 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
11355 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
11356 flag_tree_loop_vectorize rather than flag_tree_vectorize.
11358 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11361 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
11362 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
11363 Change s_operand to memory_operand.
11365 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
11367 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
11368 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
11369 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
11370 Emit instructions rather than returning an expression. Handle TFmode
11371 and KFmode by casting to TImode.
11372 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
11373 (rs6000_emit_le_vsx_store): Likewise.
11374 * config/rs6000/vsx.md (VSX_TI): New iterator.
11375 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
11376 (*vsx_le_undo_permute_<mode>): Likewise.
11377 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
11378 emit the split sequence.
11379 (*vsx_le_perm_store_<mode>): Likewise.
11381 2017-07-27 Jakub Jelinek <jakub@redhat.com>
11383 PR tree-optimization/81555
11384 PR tree-optimization/81556
11385 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
11386 if true, force CHANGED for the recursive invocation.
11387 (reassociate_bb): Remember original length of ops array, pass
11388 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
11390 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
11391 attributes for noipa attribute. For naked attribute use
11392 lookup_attribute first before lookup_attribute_spec.
11393 * final.c (rest_of_handle_final): Disable IPA RA for functions with
11395 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
11396 for functions with noipa attribute.
11397 (cgraph_externally_visible_p): Return true for functions with noipa
11399 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
11400 for functions with noipa attribute.
11401 * doc/extend.texi: Document noipa function attribute.
11402 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
11403 also for functions with noipa attribute.
11404 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
11406 2017-07-26 Andrew Pinski <apinski@cavium.com>
11408 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
11409 vec_unalign_load_cost and vec_unalign_store_cost.
11411 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11413 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
11414 -mvsx-small-integer option.
11415 (ISA_3_0_MASKS_IEEE): Likewise.
11416 (OTHER_VSX_VECTOR_MASKS): Likewise.
11417 (POWERPC_MASKS): Likewise.
11418 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
11419 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
11420 code, only testing for DImode being allowed in non-VSX floating
11422 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
11423 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
11425 (rs6000_option_override_internal): Delete -mvsx-small-integer.
11426 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
11427 TARGET_P8_VECTOR test.
11428 (rs6000_secondary_reload_simple_move): Likewise.
11429 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
11430 since TARGET_P9_VECTOR was already tested.
11431 (rs6000_opt_masks): Remove -mvsx-small-integer.
11432 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
11433 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11435 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
11436 test for TARGET_VEXTRACTUB was used, and that uses
11438 (p9 extract splitter): Likewise.
11439 (vsx_extract_<mode>_di_p9): Likewise.
11440 (vsx_extract_<mode>_store_p9): Likewise.
11441 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
11442 for TARGET_P9_VECTOR was used. Delete code that is now dead with
11443 the elimination of TARGET_VSX_SMALL_INTEGER.
11444 (vsx_extract_<mode>_p8): Likewise.
11445 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
11446 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
11447 (vsx_set_<mode>_p9): Likewise.
11448 (vsx_set_v4sf_p9): Likewise.
11449 (vsx_set_v4sf_p9_zero): Likewise.
11450 (vsx_insert_extract_v4sf_p9): Likewise.
11451 (vsx_insert_extract_v4sf_p9_2): Likewise.
11452 * config/rs6000/rs6000.md (sign extend splitter): Change
11453 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
11454 (floatsi<mode>2_lfiwax_mem): Likewise.
11455 (floatunssi<mode>2_lfiwzx_mem): Likewise.
11456 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
11457 since a test for TARGET_P9_VECTOR was used.
11458 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11459 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
11460 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11461 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11462 TARGET_P8_VECTOR test.
11463 (fix_trunc<mode>si2_stfiwx): Likewise.
11464 (fix_trunc<mode>si2_internal): Likewise.
11465 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
11466 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11468 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11469 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11470 TARGET_P8_VECTOR test.
11471 (fixuns_trunc<mode>si2_stfiwx): Likewise.
11472 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
11473 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11475 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11476 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
11477 since a test for TARGET_P9_VECTOR was used.
11478 (splitter for loading small constants): Likewise.
11480 2017-07-26 Andrew Pinski <apinski@cavium.com>
11482 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
11485 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
11488 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
11489 (fp_valid_at): Likewise.
11491 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11493 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
11494 (qdf24xx_addrcost_table): Likewise.
11495 (cortexa57_tunings): Update to use generic_branch_cost.
11496 (cortexa72_tunings): Likewise.
11497 (cortexa73_tunings): Likewise.
11498 (qdf24xx_tunings): Likewise.
11500 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11502 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
11503 (thunderx2t99_branch_cost): Likewise.
11504 (cortexa35_tunings): Update to use generic_branch_cost.
11505 (cortexa53_tunings): Likewise.
11506 (cortexa57_tunings): Likewise.
11507 (cortexa72_tunings): Likewise.
11508 (cortexa73_tunings): Likewise.
11509 (thunderx2t99_tunings): Likewise.
11511 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11513 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
11514 (sparc_option_override): Honour MASK_FSMULD.
11515 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
11516 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
11517 * config/sparc/sparc.opt (mfsmuld): New option.
11518 * doc/invoke.texi (mfsmuld): Document option.
11520 2017-07-26 Marek Polacek <polacek@redhat.com>
11522 PR middle-end/70992
11523 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
11525 2017-07-26 Richard Biener <rguenther@suse.de>
11527 * gimple-match-head.c (do_valueize): Return OP if valueize
11529 (get_def): New helper to get at the def stmt of a SSA name
11530 if valueize allows.
11531 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
11532 do_valueize to get at the def stmt.
11533 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
11535 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
11537 PR middle-end/46932
11538 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
11540 2017-07-26 Martin Liska <mliska@suse.cz>
11543 * function.c (expand_function_start): Make expansion of
11544 nonlocal_goto_save_area after parm_birth_insn.
11546 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11548 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
11549 from all CPU target flags enable members.
11551 2017-07-26 Richard Biener <rguenther@suse.de>
11553 * genmatch.c (dt_simplify::gen): Make iterator vars const.
11554 (decision_tree::gen): Make 'type' const.
11555 (write_predicate): Likewise.
11557 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11559 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
11560 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
11561 (rs6000_option_override_internal): Likewise.
11562 (rs6000_expand_vector_set): Likewise.
11563 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
11564 (TARGET_UPPER_REGS_SF): Likewise.
11565 (TARGET_UPPER_REGS_DI): Likewise.
11566 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
11567 (TARGET_DIRECT_MOVE_64BIT): Likewise.
11568 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
11569 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11570 (Splitters for DI constants in Altivec registers): Likewise.
11571 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
11572 (vsx_set_v4sf_p9): Likewise.
11573 (vsx_set_v4sf_p9_zero): Likewise.
11574 (vsx_insert_extract_v4sf_p9): Likewise.
11575 (vsx_insert_extract_v4sf_p9_2): Likewise.
11577 2017-07-25 Carl Love <cel@us.ibm.com>
11579 * doc/extend.texi: Update the built-in documentation file for the
11580 existing built-in functions
11581 vector signed char vec_cnttz (vector signed char);
11582 vector unsigned char vec_cnttz (vector unsigned char);
11583 vector signed short vec_cnttz (vector signed short);
11584 vector unsigned short vec_cnttz (vector unsigned short);
11585 vector signed int vec_cnttz (vector signed int);
11586 vector unsigned int vec_cnttz (vector unsigned int);
11587 vector signed long long vec_cnttz (vector signed long long);
11588 vector unsigned long long vec_cnttz (vector unsigned long long);
11590 2017-07-25 Andrew Pinski <apinski@cavium.com>
11592 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
11593 accesses where the use is for the first operand of a BIT_INSERT.
11595 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11598 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11599 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11601 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11603 * config/i386/gstabs.h: Delete.
11604 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
11606 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
11608 * config/i386/i386.c (ix86_decompose_address): Do not check for
11609 register RTX when looking at index_reg or base_reg.
11610 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
11612 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
11614 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
11615 to update EH info here.
11617 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11619 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
11621 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11623 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
11625 2017-07-25 Torsten Duwe <duwe@suse.de>
11627 * common.opt: Introduce -fpatchable-function-entry
11628 command line option, and its variables function_entry_patch_area_size
11629 and function_entry_patch_area_start.
11630 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
11631 including a two-value parser.
11632 * target.def (print_patchable_function_entry): New target hook.
11633 * targhooks.h (default_print_patchable_function_entry): New function.
11634 * targhooks.c (default_print_patchable_function_entry): Likewise.
11635 * toplev.c (process_options): Switch off IPA-RA if
11636 patchable function entries are being generated.
11637 * varasm.c (assemble_start_function): Look at the
11638 patchable-function-entry command line switch and current
11639 function attributes and maybe generate NOP instructions by
11640 calling the print_patchable_function_entry hook.
11641 * doc/extend.texi: Document patchable_function_entry attribute.
11642 * doc/invoke.texi: Document -fpatchable_function_entry
11643 command line option.
11644 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
11646 * doc/tm.texi: Re-generate.
11648 2017-07-25 Jakub Jelinek <jakub@redhat.com>
11651 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
11652 TARGET_AVX512DQ rather than TARGET_AVX512BW.
11654 2017-07-25 Tamar Christina <tamar.christina@arm.com>
11656 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
11658 2017-07-25 Richard Biener <rguenther@suse.de>
11660 PR tree-optimization/81455
11661 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
11662 not walk in cycles when looking for guards.
11664 2017-07-25 Richard Biener <rguenther@suse.de>
11666 PR tree-optimization/81529
11667 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
11668 when optimizing backedge uses.
11670 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
11672 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
11674 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
11675 to dl_section_ref. On AIX, append an expression to subtract
11676 the size of the section length to dl_section_ref.
11678 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
11680 * configure.ac: If any of the config.* scripts fail, exit 1.
11681 * configure: Regenerate.
11683 2017-07-25 Richard Biener <rguenther@suse.de>
11685 PR middle-end/81546
11686 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
11687 of immediate uses, be more verbose on errors.
11689 2017-07-25 Richard Biener <rguenther@suse.de>
11691 PR tree-optimization/81510
11692 * tree-vect-loop.c (vect_is_simple_reduction): When the
11693 reduction stmt is not inside the loop bail out.
11695 2017-07-25 Richard Biener <rguenther@suse.de>
11697 PR tree-optimization/81303
11698 * tree-vect-loop-manip.c (vect_loop_versioning): Build
11699 profitability check against LOOP_VINFO_NITERSM1.
11701 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11703 * domwalk.c (cmp_bb_postorder): Simplify.
11704 (sort_bbs_postorder): New function. Use it...
11705 (dom_walker::walk): ...here to optimize common cases.
11707 2017-07-25 Martin Liska <mliska@suse.cz>
11710 * ipa-visibility.c (function_and_variable_visibility): Make the
11711 redirection just on target that supports aliasing.
11712 Fix GNU coding style.
11714 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
11717 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
11719 * config/arm/rtems.h: Likewise.
11720 * config/bfin/rtems.h: Likewise.
11721 * config/i386/rtemself.h: Likewise.
11722 * config/lm32/rtems.h: Likewise.
11723 * config/m32c/rtems.h: Likewise.
11724 * config/m68k/rtemself.h: Likewise.
11725 * config/microblaze/rtems.h: Likewise.
11726 * config/mips/rtems.h: Likewise.
11727 * config/moxie/rtems.h: Likewise.
11728 * config/nios2/rtems.h: Likewise.
11729 * config/powerpcspe/rtems.h: Likewise.
11730 * config/rs6000/rtems.h: Likewise.
11731 * config/rtems.h: Likewise.
11732 * config/sh/rtems.h: Likewise.
11733 * config/sh/rtemself.h: Likewise.
11734 * config/sparc/rtemself.h: Likewise.
11736 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11739 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
11740 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
11741 * tree-ssa-structalias.c (alias_get_name): Same.
11743 2017-07-25 Bin Cheng <bin.cheng@arm.com>
11746 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
11747 instructions if no du chain is found.
11749 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11751 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
11753 2017-07-25 Richard Biener <rguenther@suse.de>
11755 PR middle-end/81505
11756 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
11759 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11761 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
11762 upper-regs options.
11763 (ISA_2_7_MASKS_SERVER): Likewise.
11764 (ISA_3_0_MASKS_IEEE): Likewise.
11765 (OTHER_P8_VECTOR_MASKS): Likewise.
11766 (OTHER_VSX_VECTOR_MASKS): Likewise.
11767 (POWERPC_MASKS): Likewise.
11768 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
11769 duplicate list of options.
11770 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
11771 explicit -mupper-regs options.
11772 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
11773 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
11774 alias for -mupper-regs-df.
11775 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11776 (rs6000_init_hard_regno_mode_ok): Likewise.
11777 (rs6000_option_override_internal): Likewise.
11778 (rs6000_opt_masks): Likewise.
11779 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
11780 options in terms of whether -mvsx or -mpower8-vector was used.
11781 (TARGET_UPPER_REGS_DI): Likewise.
11782 (TARGET_UPPER_REGS_SF): Likewise.
11783 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
11784 -mupper-regs-* options.
11786 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11788 * passes.c (emergency_dump_function): Print some empty lines and a
11789 header before the RTL dump.
11791 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11793 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
11795 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
11798 * config/aarch64/aarch64.c (aarch64_classify_symbol):
11799 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
11801 2017-07-24 Carl Love <cel@us.ibm.com>
11803 * config/rs6000/rs6000-c.c: Add support for built-in functions
11804 vector float vec_extract_fp32_from_shorth (vector unsigned short);
11805 vector float vec_extract_fp32_from_shortl (vector unsigned short);
11806 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
11807 vec_extract_fp_from_shortl): Add defines for the two builtins.
11808 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
11809 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
11811 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
11812 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
11813 * doc/extend.texi: Update the built-in documentation file for the
11814 new built-in function.
11816 2017-07-24 Jakub Jelinek <jakub@redhat.com>
11819 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
11821 * doc/generic.texi: Likewise.
11822 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11823 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11825 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
11827 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
11828 (aarch64_mls_elt_merge<mode>): Likewise.
11830 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
11832 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11833 having __cxa_atexit.
11835 2017-07-23 Michael Collison <michael.collison@arm.com>
11837 * config/arm/arm.c (arm_option_override): Deprecate
11838 use of -mstructure-size-boundary.
11839 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
11840 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
11842 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11845 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
11846 Reduce cost estimate for direct moves.
11848 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
11851 * config/i386/i386.c (ix86_option_override_internal): Disable
11852 BMI, BMI2 and TBM instructions for -m16.
11854 2017-07-21 Carl Love <cel@us.ibm.com>
11856 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11857 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11858 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11859 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11860 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11861 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11862 VMULOSW): New enum "unspec" values.
11863 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11864 altivec_vmulosw): New patterns.
11865 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11866 VMULOSW): Add definitions.
11868 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
11870 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
11871 (qdf24xx): Likewise.
11872 * config/aarch64/aarch64-options-extensions.def (rdma); New.
11873 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
11874 (AARCH64_FL_V8_1): Renumber.
11875 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
11876 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
11877 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
11878 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
11879 rdma to feature modifiers list.
11881 2017-07-21 Yury Gribov <tetra2005@gmail.com>
11883 PR middle-end/56727
11884 * ipa-visibility (function_and_variable_visibility): Convert
11885 recursive PLT call to direct call if appropriate.
11887 2017-07-21 Andrew Pinski <apinski@cavium.com>
11889 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
11890 operand 1 to see if the types precision matches.
11891 * fold-const.c (operand_equal_p): Likewise.
11893 2017-07-21 Richard Biener <rguenther@suse.de>
11895 PR tree-optimization/81303
11896 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
11897 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
11898 (vect_peeling_hash_get_lowest_cost): Adjust.
11899 (vect_enhance_data_refs_alignment): Likewise. Use
11900 vect_get_peeling_costs_all_drs to compute the penalty for no
11901 peeling to match up costs.
11903 2017-07-21 Richard Biener <rguenther@suse.de>
11905 PR tree-optimization/81500
11906 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
11907 we didn't identify a reduction path.
11909 2017-07-21 Tom de Vries <tom@codesourcery.com>
11910 Cesar Philippidis <cesar@codesourcery.com>
11912 PR gcov-profile/81442
11913 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
11916 2017-07-21 Tom de Vries <tom@codesourcery.com>
11919 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
11921 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
11922 nvptx_override_options_after_change.
11924 2017-07-21 Ulrich Drepper <drepper@redhat.com>
11926 * dwarf2out.c (output_file_names): Avoid double testing for
11927 dwarf_version >= 5.
11929 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
11931 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
11933 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11935 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
11937 (try_crossjump_to_edge): Do not punt on partitioned functions.
11939 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11941 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
11942 Put all BBs reachable only via paths crossing cold region to cold
11944 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
11946 2016-07-21 Richard Biener <rguenther@suse.de>
11948 PR tree-optimization/81303
11949 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
11950 into account prologue and epilogue iterations when raising
11951 min_profitable_iters to sth at least covering one vector iteration.
11953 2017-07-21 Tamar Christina <tamar.christina@arm.com>
11955 * config/arm/arm.c (arm_test_cpu_arch_dat):
11958 2017-07-20 Nathan Sidwell <nathan@acm.org>
11960 Remove TYPE_METHODS.
11961 * tree.h (TYPE_METHODS): Delete.
11962 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
11963 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
11964 (dbxout_type_methods): Scan TYPE_FIELDS.
11965 (dbxout_type): Don't check TYPE_METHODS here.
11966 * function.c (use_register_for_decl): Always ignore register for
11967 class types when not optimizing.
11968 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
11969 * tree.c (free_lang_data_in_type): Stitch out member functions and
11970 templates from TYPE_FIELDS.
11971 (build_distinct_type_copy, verify_type_variant,
11972 verify_type): Member fns are on TYPE_FIELDS.
11973 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
11974 * tree-pretty-print.c (dump_generic_node): Likewise.
11976 2017-07-20 Jakub Jelinek <jakub@redhat.com>
11979 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
11980 V2TImode and V4TImode.
11981 (ix86_expand_vector_extract): Likewise.
11982 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
11983 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
11984 (ssescalarmode): Handle V4TImode and V2TImode.
11985 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
11986 (*vec_extractv2ti, *vec_extractv4ti): New insns.
11987 (VEXTRACTI128_MODE): New mode iterator.
11988 (splitter for *vec_extractv?ti first element): New.
11989 (VEC_INIT_MODE): New mode iterator.
11990 (vec_init<mode>): Consolidate 3 expanders into one using
11991 VEC_INIT_MODE mode iterator.
11993 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11995 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
11996 non_spilled_static_chain_regno_p.
11998 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12000 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
12002 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
12004 * bb-reorder.c (connect_traces): Allow copying of blocks within
12007 2017-07-20 Richard Biener <rguenther@suse.de>
12009 * gimple.h (gimple_phi_result): Add gphi * overload.
12010 (gimple_phi_result_ptr): Likewise.
12011 (gimple_phi_arg): Likewise. Adjust index assert to only
12012 allow actual argument accesses rather than all slots available
12014 (gimple_phi_arg_def): Add gphi * overload.
12015 * tree-phinodes.c (make_phi_node): Initialize only actual
12017 (resize_phi_node): Clear memory not covered by old node,
12018 do not initialize excess argument slots.
12019 (reserve_phi_args_for_new_edge): Initialize new argument slot
12022 2017-07-20 Bin Cheng <bin.cheng@arm.com>
12024 PR tree-optimization/81388
12026 2016-07-21 Bin Cheng <bin.cheng@arm.com>
12028 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
12029 by removing computation of may_be_zero.
12031 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12032 Tom de Vries <tom@codesourcery.com>
12034 PR middle-end/81030
12035 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
12036 when gimple level profile disagrees with what RTL expander did.
12038 2017-07-20 Richard Biener <rguenther@suse.de>
12040 PR tree-optimization/61171
12041 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
12042 (vect_analyze_stmt): Add slp instance parameter.
12043 (vectorizable_reduction): Likewise.
12044 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
12045 (vect_is_simple_reduction): Deal with chains not detected
12046 as SLP reduction chain, specifically not properly associated
12047 chains containing a mix of plus/minus.
12048 (get_reduction_op): Remove.
12049 (get_initial_defs_for_reduction): Simplify, pass in whether
12050 this is a reduction chain, pass in the SLP node for the PHIs.
12051 (vect_create_epilog_for_reduction): Get the SLP instance as
12053 (vectorizable_reduction): Get the SLP instance as arg.
12054 During analysis remember the SLP node with the PHIs in the
12055 instance. Simplify getting at the vectorized reduction PHIs.
12056 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
12057 through SLP instance.
12058 (vect_slp_analyze_operations): Likewise.
12059 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
12060 (vect_transform_stmt): Likewise.
12062 2017-07-20 Tom de Vries <tom@codesourcery.com>
12064 PR tree-optimization/81489
12065 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
12066 read of phi arg location to before loop that modifies phi.
12068 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12070 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
12073 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12075 PR middle-end/81331
12076 * except.c (execute): Fix ordering issue.
12078 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12080 PR rtl-optimization/81423
12081 * combine.c (make_compound_operation_int): Don't try to optimize
12082 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
12084 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12086 PR rtl-optimization/81423
12087 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
12088 with a constant that is -1 in the truncated to mode.
12090 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12092 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
12093 (determine_unlikely_bbs): ... here.
12094 * predict.h (propagate_unlikely_bbs_forward): Declare.
12095 * cfgexpand.c (pass_expand::execute): Use it.
12096 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
12098 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
12099 propagate_unlikely_bbs_forward.
12101 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12103 PR middle-end/81331
12104 * except.c (maybe_add_nop_after_section_switch): New function.
12107 2017-07-19 Tom de Vries <tom@codesourcery.com>
12109 * gimple.h (gimple_phi_set_arg): Make assert more strict.
12111 2017-07-19 Tom de Vries <tom@codesourcery.com>
12113 * gimple.h (gimple_phi_arg): Make assert more strict.
12115 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
12117 * config.gcc (powerpc*-*-*): Add mmintrin.h.
12118 * config/rs6000/mmintrin.h: New file.
12119 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
12121 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12123 PR tree-optimization/81346
12124 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
12126 2017-07-19 Tom de Vries <tom@codesourcery.com>
12128 * config/nvptx/nvptx.md (VECIM): Add V2DI.
12130 2017-07-19 Tom de Vries <tom@codesourcery.com>
12132 * config/nvptx/nvptx-modes.def: Add V2DImode.
12133 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
12134 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
12135 (nvptx_output_mov_insn): Handle lack of mov.b128.
12136 (nvptx_print_operand): Handle 'H' and 'L' codes.
12137 (nvptx_vector_mode_supported): Allow V2DImode.
12138 (nvptx_preferred_simd_mode): New function.
12139 (nvptx_data_alignment): New function.
12140 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
12141 nvptx_preferred_simd_mode.
12142 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
12144 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
12146 2017-07-19 Tom de Vries <tom@codesourcery.com>
12148 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
12149 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
12150 (nvptx_vector_mode_supported): New function. Allow V2SImode.
12151 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
12152 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
12153 (mov<VECIM>_insn): New define_insn.
12154 (define_expand "mov<VECIM>): New define_expand.
12156 2017-07-19 Tom de Vries <tom@codesourcery.com>
12158 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
12160 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12162 PR tree-optimization/81346
12163 * fold-const.h (fold_div_compare, range_check_type): Declare.
12164 * fold-const.c (range_check_type): New function.
12165 (build_range_check): Use range_check_type.
12166 (fold_div_compare): No longer static, rewritten into
12167 a match.pd helper function.
12168 (fold_comparison): Don't call fold_div_compare here.
12169 * match.pd (X / C1 op C2): New optimization using fold_div_compare
12170 as helper function.
12172 2017-07-19 Nathan Sidwell <nathan@acm.org>
12174 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
12175 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
12176 * tree.c (find_decls_types_r, verify_type): Use
12177 TYPE_{MIN,MAX}_VALUE_RAW.
12178 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
12179 (hash_tree): Likewise.
12180 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12182 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12185 2017-07-18 Tom de Vries <tom@codesourcery.com>
12187 PR middle-end/81464
12188 * omp-expand.c (expand_omp_for_static_chunk): Handle
12189 equal-argument loop exit phi.
12191 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
12194 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
12195 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
12196 operand 2 predicate.
12197 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
12198 operand 2 predicate.
12199 (ror,rol -> rorx splitters): Use const_int_operand as
12200 operand 2 predicate.
12202 2017-06-18 Richard Biener <rguenther@suse.de>
12204 PR tree-optimization/81410
12205 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
12206 the gap in the ! slp_perm SLP case after each group.
12208 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12210 PR middle-end/81463
12211 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
12214 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12216 PR middle-end/81462
12217 * predict.c (set_even_probabilities): Cleanup; do not affect
12218 probabilities that are already known.
12219 (combine_predictions_for_bb): Call even when count is set.
12221 2017-07-18 Nathan Sidwell <nathan@acm.org>
12223 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
12226 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12229 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
12230 optimization for loop niter analysis.
12232 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
12235 * config/avr/avr.c (avr_optimize_casesi): Don't use
12236 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
12238 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12240 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
12241 body_cost_vec from _vect_peel_extended_info.
12242 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
12243 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
12246 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12248 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
12250 2017-07-18 Richard Biener <rguenther@suse.de>
12252 PR tree-optimization/80620
12253 PR tree-optimization/81403
12254 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
12255 info when re-using a VN table entry.
12257 2017-07-18 Richard Biener <rguenther@suse.de>
12259 PR tree-optimization/81418
12260 * tree-vect-loop.c (vectorizable_reduction): Properly compute
12261 vectype_in. Verify that with lane-reducing reduction operations
12262 we have a single def-use cycle.
12264 2017-07-17 Carl Love <cel@us.ibm.com>
12266 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
12268 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12269 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12270 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12271 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12272 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12273 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12274 VMULOSW): New enum "unspec" values.
12275 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12276 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12277 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12278 altivec_vmulosw): New patterns.
12279 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12280 VMULOSW): Add definitions.
12281 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
12283 * config/alpha/alpha.c: Include predict.h.
12285 2017-07-17 Yury Gribov <tetra2005@gmail.com>
12287 * tree-vrp.c (compare_assert_loc): Fix comparison function
12288 to return predictable results.
12290 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12292 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
12294 (subdi3): Likewise.
12295 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
12296 * doc/invoke.texi (mexpand-adddi): Update text.
12298 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12300 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
12301 that also clobbers the CC register. The old expand code is moved
12303 (*arc_clzsi2): ... here.
12304 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
12305 the CC register. The old expand code is moved to ...
12306 (arc_ctzsi2): ... here.
12308 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12310 * config/arc/arc.opt (mindexed-loads): Use initial value
12311 TARGET_INDEXED_LOADS_DEFAULT.
12312 (mauto-modify-reg): Use initial value
12313 TARGET_AUTO_MODIFY_REG_DEFAULT.
12314 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12315 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12316 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12317 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12319 2017-07-17 Martin Liska <mliska@suse.cz>
12322 * opts.c (finish_options): Do not allow -fgnu-tm
12323 w/ -fsanitize={kernel-,}address. Say sorry.
12325 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12328 * tree-loop-distribution.c (classify_partition): Only assert on
12329 numer of iterations.
12330 (merge_dep_scc_partitions): Delete prameter. Update function call.
12331 (distribute_loop): Remove code handling loop with unknown niters.
12332 (pass_loop_distribution::execute): Skip loop with unknown niters.
12334 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12337 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
12338 function sort_partitions_by_post_order.
12340 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12342 PR tree-optimization/81374
12343 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
12344 the max index of basic blocks, rather than number of basic blocks.
12346 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12348 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
12350 (arc_legitimate_pic_operand_p): Likewise.
12351 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
12353 (arc_needs_pcl_p): Likewise.
12354 (arc_legitimate_pc_offset_p): Likewise.
12355 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
12356 function is also used in constrains.md.
12357 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
12358 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
12359 PLUS. Only return true/false in known cases, otherwise assert.
12360 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
12361 is already called in arc_legitimate_constant_p.
12362 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
12364 (LEGITIMATE_PIC_OPERAND_P): Use
12365 arc_raw_symbolic_reference_mentioned_p function.
12366 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
12371 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12372 Andrew Burgess <andrew.burgess@embecosm.com>
12374 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
12375 (arc_return_address_register): New function.
12376 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
12377 (arc_handle_fndecl_attribute): Add naked attribute.
12378 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12379 (TARGET_WARN_FUNC_RETURN): Likewise.
12380 (arc_allocate_stack_slots_for_args): New function.
12381 (arc_warn_func_return): Likewise.
12382 (machine_function): Change type fn_type.
12383 (arc_compute_function_type): Consider new naked function type,
12384 change function return type.
12385 (arc_must_save_register): Adapt to handle new
12386 arc_compute_function_type's return type.
12387 (arc_expand_prologue): Likewise.
12388 (arc_expand_epilogue): Likewise.
12389 (arc_return_address_regs): Delete.
12390 (arc_return_address_register): New function.
12391 (arc_epilogue_uses): Use above function.
12392 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
12393 (arc_function_type): Change encoding, add naked type.
12394 (ARC_INTERRUPT_P): Change to handle the new encoding.
12395 (ARC_FAST_INTERRUPT_P): Likewise.
12396 (ARC_NORMAL_P): Define.
12397 (ARC_NAKED_P): Likewise.
12398 (arc_compute_function_type): Delete prototype.
12399 * config/arc/arc.md (in_ret_delay_slot): Use
12400 arc_return_address_register function.
12401 (simple_return): Likewise.
12402 (p_return_i): Likewise.
12404 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12406 PR tree-optimization/81428
12407 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
12408 can't be built for those types.
12410 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12412 Remove stuff dead since r239246.
12414 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
12415 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
12416 (avr_inform_devices): Remove dead stuff.
12418 2017-07-17 Tamar Christina <tamar.christina@arm.com>
12420 * config/arm/arm_neon.h: Fix softp typo.
12422 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12424 PR tree-optimization/81365
12425 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
12426 aggregate moves onto bb predecessor edges, make sure there are no
12427 loads that could alias the lhs in between the start of bb and the
12430 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12433 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
12434 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
12435 [LSHIFTRT, outer_code = TRUNCATE]: Same.
12437 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12439 PR tree-optimization/81396
12440 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
12441 (init_symbolic_number): Initialize it to 1.
12442 (perform_symbolic_merge): Add n_ops from both operands into the new
12444 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
12445 without base_addr as useless if they need more than one operation.
12446 (bswap_replace): Handle !bswap case for NULL base_addr.
12448 2017-07-17 Tom de Vries <tom@codesourcery.com>
12451 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
12454 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
12456 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
12457 conditional builtin define __FIX_LEON3FT_B2BST.
12459 2017-07-17 Daniel Cederman <cederman@gaisler.com>
12461 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
12462 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
12465 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12467 PR rtl-optimization/81424
12468 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
12469 to remove potential trapping from operands if -fnon-call-exceptions.
12471 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12473 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
12474 profile_proability for scalling.
12475 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
12477 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12479 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
12481 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12483 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
12484 fixpoint arithmetics.
12486 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12488 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
12489 fixpoint arithmetics.
12491 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12493 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
12494 fixpoint arithmetics.
12496 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12498 * profile-count.h (profile_probability::from_reg_br_prob_note,
12499 profile_probability::to_reg_br_prob_note): New functions.
12500 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
12501 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
12502 * predict.c (probability_reliable_p): Update.
12503 (edge_probability_reliable_p): Update.
12504 (br_prob_note_reliable_p): Update.
12505 (invert_br_probabilities): Update.
12506 (add_reg_br_prob_note): New function.
12507 (combine_predictions_for_insn): Update.
12508 * asan.c (asan_clear_shadow): Update.
12509 * cfgbuild.c (compute_outgoing_frequencies): Update.
12510 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12511 (update_br_prob_note): Update.
12512 (rtl_verify_edges): Update.
12513 (purge_dead_edges): Update.
12514 (fixup_reorder_chain): Update.
12515 * emit-rtl.c (try_split): Update.
12516 * ifcvt.c (cond_exec_process_insns): Update.
12517 (cond_exec_process_if_block): Update.
12518 (dead_or_predicable): Update.
12519 * internal-fn.c (expand_addsub_overflow): Update.
12520 (expand_neg_overflow): Update.
12521 (expand_mul_overflow): Update.
12522 * loop-doloop.c (doloop_modify): Update.
12523 * loop-unroll.c (compare_and_jump_seq): Update.
12524 * optabs.c (emit_cmp_and_jump_insn_1): Update.
12525 * predict.h: Update.
12526 * reorg.c (mostly_true_jump): Update.
12528 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
12529 * config/alpha/alpha.c (emit_unlikely_jump): Update.
12530 * config/arc/arc.c: (emit_unlikely_jump): Update.
12531 * config/arm/arm.c: (emit_unlikely_jump): Update.
12532 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
12533 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
12534 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
12535 (ix86_print_operand): Update.
12536 (ix86_split_fp_branch): Update.
12537 (predict_jump): Update.
12538 * config/ia64/ia64.c (ia64_print_operand): Update.
12539 * config/mmix/mmix.c (mmix_print_operand): Update.
12540 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
12541 (rs6000_expand_split_stack_prologue): Update.
12542 * config/rs6000/rs6000.c: Update.
12543 * config/s390/s390.c (s390_expand_vec_strlen): Update.
12544 (s390_expand_vec_movstr): Update.
12545 (s390_expand_cs_tdsi): Update.
12546 (s390_expand_split_stack_prologue): Update.
12547 * config/sh/sh.c (sh_print_operand): Update.
12548 (expand_cbranchsi4): Update.
12549 (expand_cbranchdi4): Update.
12550 * config/sparc/sparc.c (output_v9branch): Update.
12551 * config/spu/spu.c (get_branch_target): Update.
12552 (ea_load_store_inline): Update.
12553 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
12554 * config/tilepro/tilepro.c: Update.
12556 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12558 * gimplify.c (mostly_copy_tree_r): Revert latest change.
12559 (gimplify_save_expr): Likewise.
12561 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12563 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
12565 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12567 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
12569 * timevar.def (TV_IPA_FNSUMMARY): Define.
12571 2017-07-16 Daniel Cederman <cederman@gaisler.com>
12573 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
12574 to back store errata sensitive sequence from being generated.
12575 (sqrtdf2_fix): Likewise.
12577 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12579 * tree-ssa-threadupdate.c (compute_path_counts,
12580 update_joiner_offpath_counts): Use profile_probability.
12582 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
12585 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12587 * config/arm/arm-c.c (arm_cpu_builtins): Define
12588 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12590 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
12592 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12593 array entries to represent __ieee128 versions of the
12594 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
12595 scalar_extract_sig, and scalar_insert_exp built-in functions.
12596 (altivec_resolve_overloaded_builtin): Add special case handling
12597 for the __builtin_scalar_insert_exp function, as represented by
12598 the P9V_BUILTIN_VEC_VSIEDP constant.
12599 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
12600 exponent support for __ieee128 argument.
12601 (VSESQP): Add scalar extract signature support for __ieee128
12603 (VSTDCNQP): Add scalar test negative support for __ieee128
12605 (VSIEQP): Add scalar insert exponent support for __int128 argument
12606 with __ieee128 result.
12607 (VSIEQPF): Add scalar insert exponent support for __ieee128
12608 argument with __ieee128 result.
12609 (VSTDCQP): Add scalar test data class support for __ieee128
12611 (VSTDCNQP): Add overload support for scalar test negative with
12612 __ieee128 argument.
12613 (VSTDCQP): Add overload support for scalar test data class
12614 __ieee128 argument.
12615 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
12616 UNSPEC_VSX_SXSIGDP.
12617 (UNSPEC_VSX_SIEXPQP): New constant.
12618 (xsxexpqp): New insn for VSX scalar extract exponent quad
12620 (xsxsigqp): New insn for VSX scalar extract significand quad
12622 (xsiexpqpf): New insn for VSX scalar insert exponent quad
12623 precision with floating point argument.
12624 (xststdcqp): New expand for VSX scalar test data class quad
12626 (xststdcnegqp): New expand for VSX scalar test negative quad
12628 (xststdcqp): New insn to match expansions for VSX scalar test data
12629 class quad precision and VSX scalar test negative quad precision.
12630 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
12631 special case operand checking to enforce that second operand of
12632 VSX scalar test data class with quad precision argument is a 7-bit
12634 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12635 prototypes and descriptions of __ieee128 versions of
12636 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
12637 scalar_test_data_class, and scalar_test_neg built-in functions.
12639 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12641 PR tree-optimization/81162
12642 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
12643 replace a negate with an add.
12645 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
12647 * doc/invoke.texi (arm/-mcpu): Document +crypto.
12649 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12651 * config/arm/arm-c.c (arm_cpu_builtins): Define
12652 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12654 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12656 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
12657 (armv8-r): Set ARM Cortex-R52 as default CPU.
12658 * config/arm/arm-tables.opt: Regenerate.
12659 * config/arm/arm-tune.md: Regenerate.
12660 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12662 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
12663 extension for -mcpu=cortex-r52.
12665 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12667 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
12668 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
12669 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
12670 (fp-armv8): Define it as FP_ARMv8 only.
12671 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
12672 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
12674 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
12675 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
12676 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
12677 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
12678 than TARGET_FPU_ARMV8.
12679 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
12680 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
12681 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
12683 * config/arm/neon.md (neon_vrint): Likewise.
12684 (neon_vcvt): Likewise.
12685 (neon_<fmaxmin_op><mode>): Likewise.
12686 (<fmaxmin><mode>3): Likewise.
12687 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
12688 * config/arm/predicates.md (arm_cond_move_operator): Check against
12689 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
12691 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
12693 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
12694 to top of function.
12696 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12698 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
12699 loop in comment with memset.
12701 2017-07-14 Martin Liska <mliska@suse.cz>
12703 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
12704 * dwarf2out.c (is_java): Remove the function.
12705 (output_pubname): Remove usage of the function.
12706 (lower_bound_default): Remove usage of DW_LANG_Java.
12707 (gen_compile_unit_die): Likewise.
12708 * gcc.c: Remove compiler defaults for .java and .zip files.
12709 * gimple-expr.c (remove_suffix): Change as there's no longer
12710 extension than 4-letter one.
12711 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
12712 (gimplify_save_expr): Likewise.
12713 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
12714 as it's possible even for other languages than Java.
12715 * langhooks.h (struct lang_hooks): Remove Java from a comment.
12716 * lto-opts.c (lto_write_options): Remove reference to Java.
12717 * opts.c (strip_off_ending): Update file extension handling.
12718 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
12719 * tree-eh.c (lower_resx): Likewise.
12720 * tree.c (free_lang_data_in_type): Remove dead code.
12721 (find_decls_types_r): Likewise.
12722 (build_common_builtin_nodes): Remove Java from a comment.
12723 (verify_type): Remove dead code.
12724 * varasm.c (assemble_external): Remove Java from a comment.
12726 2017-07-14 Martin Liska <mliska@suse.cz>
12728 * opts.c (finish_options): Add quotes.
12729 (common_handle_option): Likewise.
12731 2017-07-14 Martin Liska <mliska@suse.cz>
12733 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
12734 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
12735 Remove N_SO_PASCAL.
12736 * dwarf2out.c (lower_bound_default): Do not handle
12738 (gen_compile_unit_die): Likewise.
12739 * gcc.c: Remove default extension binding for GNU Pascal.
12740 * stmt.c: Remove Pascal language from a comment.
12741 * xcoffout.c: Likewise.
12743 2017-07-13 David Malcolm <dmalcolm@redhat.com>
12746 * diagnostic-show-locus.c (fixit_cmp): New function.
12747 (layout::layout): Sort m_fixit_hints.
12748 (column_range::column_range): Assert that the values are valid.
12749 (struct char_span): New struct.
12750 (correction::overwrite): New method.
12751 (struct source_line): New struct.
12752 (line_corrections::add_hint): Add assertions. Reimplement memcpy
12753 calls in terms of classes source_line and char_span, and
12754 correction::overwrite.
12755 (selftest::test_overlapped_fixit_printing_2): New function.
12756 (selftest::diagnostic_show_locus_c_tests): Call it.
12758 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
12760 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
12761 early if there is no lhs.
12763 2017-07-13 Martin Liska <mliska@suse.cz>
12765 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
12766 (gen_reference_type_die): Likewise.
12767 * stor-layout.c: Remove Pascal-related comment.
12769 2017-07-13 Martin Liska <mliska@suse.cz>
12771 * opts.c (finish_options): Add quotes to error messages.
12772 (parse_sanitizer_options): Likewise.
12774 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
12776 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
12778 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
12780 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
12782 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
12784 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
12786 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
12788 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12791 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
12792 provides the hardware capability bits, define the macro
12793 __BUILTIN_CPU_SUPPORTS__.
12794 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
12795 if GLIBC does not provide the hardware capability bits. Add a
12796 gcc_unreachable call if the built-in cpu function is neither
12797 __builtin_cpu_is nor __builtin_cpu_supports.
12798 (rs6000_get_function_versions_dispatcher): Change the warning
12799 that an old GLIBC is used which does not export the capability
12800 bits to be an error.
12801 * doc/extend.texi (target_clones attribute): Document the
12802 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
12803 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
12804 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
12805 the macros defined by GCC if the newer GLIBC is available.
12807 2017-07-12 Jeff Law <law@redhat.com>
12809 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
12810 remaining includes slightly.
12811 * config/riscv/riscv-builtins.c: Include profile-count.h.
12813 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12816 * config/avr/avr.c (avr_set_current_function): In diagnostic
12817 messages: Quote keywords and (parts of) identifiers.
12818 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
12821 2017-07-12 Carl Love <cel@us.ibm.com>
12823 * config/rs6000/rs6000-c.c: Add support for built-in functions
12824 vector bool char vec_revb (vector bool char);
12825 vector bool short vec_revb (vector short char);
12826 vector bool int vec_revb (vector bool int);
12827 vector bool long long vec_revb (vector bool long long);
12828 * doc/extend.texi: Update the built-in documentation file for the
12829 new built-in functions.
12831 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12833 * config/s390/s390.md: Remove movcc splitter.
12835 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12837 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
12838 load/store on condition.
12840 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12843 * config/avr/avr.c (avr_encode_section_info)
12844 [progmem && !TREE_READONLY]: Error if progmem object needs
12847 2017-07-11 Michael Collison <michael.collison@arm.com>
12849 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
12852 2017-07-11 Carl Love <cel@us.ibm.com>
12854 * config/rs6000/rs6000-c.c: Add support for builtins
12855 vector unsigned int vec_parity_lsbb (vector signed int);
12856 vector unsigned int vec_parity_lsbb (vector unsigned int);
12857 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
12858 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
12859 vector unsigned long long vec_parity_lsbb (vector signed long long);
12860 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
12861 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
12862 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
12863 * doc/extend.texi: Update the built-in documentation file for the
12864 new built-in functions.
12866 2017-07-11 David Malcolm <dmalcolm@redhat.com>
12868 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
12869 (layout::m_primary_loc): New field.
12870 (layout::layout): Initialize new field. Move location filtering
12871 logic from here to...
12872 (layout::maybe_add_location_range): ...this new method. Add
12873 support for filtering to just the lines already specified by other
12875 (layout::will_show_line_p): New method.
12876 (gcc_rich_location::add_location_if_nearby): New method.
12877 (selftest::test_add_location_if_nearby): New test function.
12878 (selftest::diagnostic_show_locus_c_tests): Call it.
12879 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
12882 2017-07-11 Tom de Vries <tom@codesourcery.com>
12884 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
12885 (bb_first_real_insn): New function.
12886 (nvptx_single): Add extra initialization of broadcasted condition
12889 2017-07-11 Nathan Sidwell <nathan@acm.org>
12891 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
12893 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
12895 * doc/extend.texi (AVR Function Attributes): Remove weblink to
12896 Binutils doc as TEXI will mess them up.
12897 * doc/invoke.texi (AVR Options): Same here.
12899 2017-07-11 Daniel Cederman <cederman@gaisler.com>
12901 * config/sparc/sparc.opt (mfix-ut700): New option.
12902 (mfix-gr712rc): Likewise.
12903 (sparc_fix_b2bst): New variable.
12904 * doc/invoke.texi (SPARC options): Document them.
12905 (ARM options): Fix warnings.
12906 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
12907 instructions to prevent sequences that can trigger the store-store
12908 errata for certain LEON3FT processors.
12909 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
12910 (sparc_option_override): Set sparc_fix_b2bst appropriately.
12911 * config/sparc/sparc.md (fix_b2bst): New attribute.
12912 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
12914 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
12917 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
12919 (*rsqrtsf2_sse): Ditto.
12921 (div<mode>3): Macroize insn from divdf3 and divsf3
12922 using MODEF mode iterator.
12924 2017-07-10 Martin Sebor <msebor@redhat.com>
12926 PR tree-optimization/80397
12927 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
12928 instead of testing for equality to INTEGER_TYPE.
12930 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
12932 * config.gcc: Remove uclibc from arc target spec.
12934 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
12936 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
12938 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12941 * lto-wrapper.c (remove_option): New function.
12942 (merge_and_complain): Merge PIC/PIE options more realistically.
12944 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12946 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
12950 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
12951 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
12952 * config.in: Regenerate.
12953 * configure: Regenerate.
12954 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
12955 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
12956 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
12957 (TARGET_GASISR_PROLOGUES): ...target mask.
12958 * common/config/avr/avr-common.c
12959 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
12960 Set -mgas-isr-prologues.
12961 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
12962 INSERT_PASS_BEFORE for it.
12963 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
12964 * config/avr/avr.c (avr_option_override)
12965 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
12966 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
12967 (avr_attribute_table) <no_gccisr>: Add new function attribute.
12968 (avr_set_current_function) <is_no_gccisr>: Init machine field.
12969 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
12971 (make_avr_pass_pre_proep): New function.
12972 (emit_push_sfr) <treg>: Add argument to function and use it
12973 instead of TMP_REG.
12974 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
12975 and set machine->gasisr.yes.
12976 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
12977 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
12978 __gcc_isr.n_pushed to .L__stack_usage.
12979 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
12980 (avr_asm_final_postscan_insn): ...this new static function.
12981 * config/avr/avr.h (machine_function)
12982 <is_no_gccisr, use_L__stack_usage>: New fields.
12983 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
12984 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
12985 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
12986 (gasisr, *gasisr): New expander and insn.
12987 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12988 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
12989 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
12991 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
12993 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
12996 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12998 Move jump-tables out of .text again.
13001 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
13002 (ASM_OUTPUT_ADDR_VEC): New function.
13003 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
13004 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
13005 INSN_ADDRESSes as asm comment.
13006 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
13007 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
13008 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
13009 * config/avr/avr.md (*tablejump): Adjust comment.
13010 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
13011 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
13013 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
13014 (avr_output_addr_vec): New proto.
13015 (avr_log_t) <insn_addresses>: New field.
13017 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
13020 * config/i386/i386.c (ix86_function_arg_advance): Set
13021 outgoing_args_on_stack to true if there are outgoing arguments
13023 (ix86_function_arg): Likewise.
13024 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
13025 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
13026 * config/i386/i386.h (machine_function): Add
13027 outgoing_args_on_stack.
13029 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
13031 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13032 supporting pthreds.
13033 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
13035 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
13037 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
13038 (REAL_H): Remove $(MACHMODE_H).
13039 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
13041 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
13042 $(MACHMODE_H) and double-int.h.
13043 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
13045 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
13048 2017-07-07 Andrew Pinski <apinski@cavium.com>
13050 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
13051 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
13053 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13055 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
13056 Add warning if GCC was not configured to link against a GLIBC that
13057 exports the hardware capability bits.
13058 (make_resolver_func): Make resolver function private and not a
13059 COMDAT function. Create the name with clone_function_name instead
13060 of make_unique_name.
13063 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
13064 correct operand in doing the split.
13066 2017-07-07 Carl Love <cel@us.ibm.com>
13068 * config/rs6000/rs6000-c: Add support for built-in function
13069 vector unsigned short vec_pack_to_short_fp32 (vector float,
13071 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
13072 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
13073 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
13074 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
13075 (convert_4f32_8i16): Add define_expand.
13076 * doc/extend.texi: Update the built-in documentation file for the
13077 new built-in function.
13079 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13081 * config/sparc/m8.md: New file.
13082 * config/sparc/sparc.md: Include m8.md.
13084 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13086 * config/sparc/sparc.opt: New option -mvis4b.
13087 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
13088 (sparc_option_override): Handle VIS4B.
13089 (enum sparc_builtins): Define
13090 SPARC_BUILTIN_DICTUNPACK{8,16,32},
13091 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
13092 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
13093 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
13094 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
13095 (check_constant_argument): New function.
13096 (sparc_vis_init_builtins): Define builtins
13097 __builtin_vis_dictunpack{8,16,32},
13098 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
13099 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
13100 __builtin_vis_fpcmpde{8,16,32}shl and
13101 __builtin_vis_fpcmpur{8,16,32}shl.
13102 (sparc_expand_builtin): Check that the constant operands to
13103 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
13104 constant and in range.
13105 * config/sparc/sparc-c.c (sparc_target_macros): Handle
13107 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
13108 (SPARC_IMM5_P): Likewise.
13109 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
13110 (enabled): Handle vis4b.
13111 (UNSPEC_DICTUNPACK): New unspec.
13112 (UNSPEC_FPCMPSHL): Likewise.
13113 (UNSPEC_FPUCMPSHL): Likewise.
13114 (UNSPEC_FPCMPDESHL): Likewise.
13115 (UNSPEC_FPCMPURSHL): Likewise.
13116 (cpu_feature): New CPU feature `vis4b'.
13117 (dictunpack{8,16,32}): New insns.
13118 (FPCSMODE): New mode iterator.
13119 (fpcscond): New code iterator.
13120 (fpcsucond): Likewise.
13121 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
13122 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
13123 (fpcmpde{8,16,32}{si,di}shl): Likewise.
13124 (fpcmpur{8,16,32}{si,di}shl): Likewise.
13125 * config/sparc/constraints.md: Define constraints `q' for unsigned
13126 2-bit integer constants and `t' for unsigned 5-bit integer
13128 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
13130 (imm5_operand_dictunpack16): Likewise.
13131 (imm5_operand_dictunpack32): Likewise.
13132 (imm2_operand): Likewise.
13133 * doc/invoke.texi (SPARC Options): Document -mvis4b.
13134 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
13135 ditunpack* and fpcmp*shl builtins.
13137 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13139 * config.gcc: Handle m8 in --with-{cpu,tune} options.
13140 * config.in: Add HAVE_AS_SPARC6 define.
13141 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
13143 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
13145 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
13146 (CPP_CPU_SPEC): Handle m8.
13147 (ASM_CPU_SPEC): Likewise.
13148 * config/sparc/sparc-opts.h (enum processor_type): Add
13150 * config/sparc/sparc.c (m8_costs): New struct.
13151 (sparc_option_override): Handle TARGET_CPU_m8.
13152 (sparc32_initialize_trampoline): Likewise.
13153 (sparc64_initialize_trampoline): Likewise.
13154 (sparc_issue_rate): Likewise.
13155 (sparc_register_move_cost): Likewise.
13156 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
13157 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
13158 (ASM_CPU64_DEFAULT_SPEC): Likewise.
13159 (CPP_CPU_SPEC): Handle M8.
13160 (ASM_CPU_SPEC): Likewise.
13161 (AS_M8_FLAG): Define.
13162 * config/sparc/sparc.md: Add m8 to the cpu attribute.
13163 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
13164 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
13166 * configure: Regenerate.
13167 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
13170 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13172 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
13174 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
13175 ("*movdi_insn_sp32"): Do not set v3pipe.
13176 ("*movsi_insn"): Likewise.
13177 ("*movdi_insn_sp64"): Likewise.
13178 ("*movsf_insn"): Likewise.
13179 ("*movdf_insn_sp32"): Likewise.
13180 ("*movdf_insn_sp64"): Likewise.
13181 ("*zero_extendsidi2_insn_sp64"): Likewise.
13182 ("*sign_extendsidi2_insn"): Likewise.
13183 ("*mov<VM32:mode>_insn"): Likewise.
13184 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13185 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13186 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13187 ("<vlop:code><VL:mode>3"): Likewise.
13188 ("*not_<vlop:code><VL:mode>3"): Likewise.
13189 ("*nand<VL:mode>_vis"): Likewise.
13190 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
13191 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
13192 ("one_cmpl<VL:mode>2"): Likewise.
13193 ("faligndata<VM64:mode>_vis"): Likewise.
13194 ("alignaddrsi_vis"): Likewise.
13195 ("alignaddrdi_vis"): Likweise.
13196 ("alignaddrlsi_vis"): Likewise.
13197 ("alignaddrldi_vis"): Likewise.
13198 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13199 ("bmaskdi_vis"): Likewise.
13200 ("bmasksi_vis"): Likewise.
13201 ("bshuffle<VM64:mode>_vis"): Likewise.
13202 ("cmask8<P:mode>_vis"): Likewise.
13203 ("cmask16<P:mode>_vis"): Likewise.
13204 ("cmask32<P:mode>_vis"): Likewise.
13205 ("pdistn<P:mode>_vis"): Likewise.
13206 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13208 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13210 * config/sparc/sparc.md ("subtype"): New insn attribute.
13211 ("*wrgsr_sp64"): Set insn subtype.
13212 ("*rdgsr_sp64"): Likewise.
13213 ("alignaddrsi_vis"): Likewise.
13214 ("alignaddrdi_vis"): Likewise.
13215 ("alignaddrlsi_vis"): Likewise.
13216 ("alignaddrldi_vis"): Likewise.
13217 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13218 ("fexpand_vis"): Likewise.
13219 ("fpmerge_vis"): Likewise.
13220 ("faligndata<VM64:mode>_vis"): Likewise.
13221 ("bshuffle<VM64:mode>_vis"): Likewise.
13222 ("cmask8<P:mode>_vis"): Likewise.
13223 ("cmask16<P:mode>_vis"): Likewise.
13224 ("cmask32<P:mode>_vis"): Likewise.
13225 ("fchksm16_vis"): Likewise.
13226 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
13227 ("fmean16_vis"): Likewise.
13228 ("fp<plusminus_insn>64_vis"): Likewise.
13229 ("<plusminus_insn>v8qi3"): Likewise.
13230 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13231 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
13232 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
13233 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
13234 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
13235 ("*movqi_insn"): Likewise.
13236 ("*movhi_insn"): Likewise.
13237 ("*movsi_insn"): Likewise.
13238 ("movsi_pic_gotdata_op"): Likewise.
13239 ("*movdi_insn_sp32"): Likewise.
13240 ("*movdi_insn_sp64"): Likewise.
13241 ("movdi_pic_gotdata_op"): Likewise.
13242 ("*movsf_insn"): Likewise.
13243 ("*movdf_insn_sp32"): Likewise.
13244 ("*movdf_insn_sp64"): Likewise.
13245 ("*zero_extendhisi2_insn"): Likewise.
13246 ("*zero_extendqihi2_insn"): Likewise.
13247 ("*zero_extendqisi2_insn"): Likewise.
13248 ("*zero_extendqidi2_insn"): Likewise.
13249 ("*zero_extendhidi2_insn"): Likewise.
13250 ("*zero_extendsidi2_insn_sp64"): Likewise.
13251 ("ldfsr"): Likewise.
13252 ("prefetch_64"): Likewise.
13253 ("prefetch_32"): Likewise.
13254 ("tie_ld32"): Likewise.
13255 ("tie_ld64"): Likewise.
13256 ("*tldo_ldub_sp32"): Likewise.
13257 ("*tldo_ldub1_sp32"): Likewise.
13258 ("*tldo_ldub2_sp32"): Likewise.
13259 ("*tldo_ldub_sp64"): Likewise.
13260 ("*tldo_ldub1_sp64"): Likewise.
13261 ("*tldo_ldub2_sp64"): Likewise.
13262 ("*tldo_ldub3_sp64"): Likewise.
13263 ("*tldo_lduh_sp32"): Likewise.
13264 ("*tldo_lduh1_sp32"): Likewise.
13265 ("*tldo_lduh_sp64"): Likewise.
13266 ("*tldo_lduh1_sp64"): Likewise.
13267 ("*tldo_lduh2_sp64"): Likewise.
13268 ("*tldo_lduw_sp32"): Likewise.
13269 ("*tldo_lduw_sp64"): Likewise.
13270 ("*tldo_lduw1_sp64"): Likewise.
13271 ("*tldo_ldx_sp64"): Likewise.
13272 ("*mov<VM32:mode>_insn"): Likewise.
13273 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13274 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13276 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13278 * config/sparc/sparc.md ("type"): New insn type viscmp.
13279 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
13281 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
13282 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
13283 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13284 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
13286 ("n7_vis_logical_11cycle"): Likewise.
13287 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
13288 * config/sparc/niagara2.md ("niag3_vis": Likewise.
13289 * config/sparc/niagara.md ("niag_vis"): Likewise.
13290 * config/sparc/ultra3.md ("us3_fga"): Likewise.
13291 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
13293 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13295 * config/sparc/sparc.md: New instruction type `bmask'.
13296 (bmaskdi_vis): Use the `bmask' type.
13297 (bmasksi_vis): Likewise.
13298 * config/sparc/ultra3.md (us3_array): Likewise.
13299 * config/sparc/niagara7.md (n7_array): Likewise.
13300 * config/sparc/niagara4.md (n4_array): Likewise.
13301 * config/sparc/niagara2.md (niag2_vis): Likewise.
13302 (niag3_vis): Likewise.
13303 * config/sparc/niagara.md (niag_vis): Likewise.
13305 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13307 * ipa-comdats.c: Remove optimize check from gate.
13308 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
13309 for functions not optimized.
13310 (ipa_fn_summary_read): Skip optimize check.
13311 (ipa_fn_summary_write): Likewise.
13312 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
13314 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
13316 (can_inline_edge_p): Check flag_pcc_struct_return for match.
13317 (check_callers): Give up on caller which is not optimized.
13318 (inline_small_functions): Likewise.
13319 (ipa_inline): Do not give up when not optimizing.
13320 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
13321 away unoptimizes cdtors.
13322 (whole_program_function_and_variable_visibility): Do
13323 ipa_discover_readonly_nonaddressable_vars in LTO mode.
13324 * ipa.c (process_references): Do not check optimize.
13325 (symbol_table::remove_unreachable_nodes): Update optimize check.
13326 (set_writeonly_bit): Update optimize check.
13327 (pass_ipa_cdtor_merge::gate): Do not check optimize.
13328 (pass_ipa_single_use::gate): Remove.
13330 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13332 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
13333 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
13334 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
13335 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
13336 permute_load, permute_store, adjust_extract, adjust_splat,
13337 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
13338 replace_swap_with_copy, dump_swap_insn_table,
13339 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
13340 recombine_lvx_pattern, recombine_stvx_pattern,
13341 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
13342 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
13343 to file rs6000-p8swap.c.
13344 * config/rs6000/rs6000-p8swap.c: New file.
13345 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
13346 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
13347 and rs6000*-*-* targets.
13349 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13351 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
13353 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13355 * lto-wrapper.c (merge_and_complain): Do not merge
13356 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
13357 fsigned_zeros, ftrapping_math, fwrapv.
13358 (append_compiler_options): Do not track these options.
13359 (append_linker_options): Likewie
13361 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13363 * cgraphunit.c (cgraph_node::finalize_function): When
13364 !flag_toplevel_reorde set no_reorder flag.
13365 (varpool_node::finalize_decl): Likewise.
13366 (symbol_table::compile): Drop no toplevel reorder path.
13368 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13370 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
13371 edges; zero probability is not better than uninitialized.
13373 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13375 * asan.h (asan_sanitize_allocas_p): Declare.
13376 * asan.c (asan_sanitize_allocas_p): New function.
13377 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
13378 (handle_builtin_alloca): Likewise.
13379 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
13380 if !asan_sanitize_allocas_p.
13381 * params.def (asan-instrument-allocas): Add new option.
13382 * params.h (ASAN_PROTECT_ALLOCAS): Define.
13383 * opts.c (common_handle_option): Disable allocas sanitization for
13386 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13388 * asan.c: Include gimple-fold.h.
13389 (get_last_alloca_addr): New function.
13390 (handle_builtin_stackrestore): Likewise.
13391 (handle_builtin_alloca): Likewise.
13392 (asan_emit_allocas_unpoison): Likewise.
13393 (get_mem_refs_of_builtin_call): Add new parameter, remove const
13394 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
13395 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
13396 (instrument_builtin_call): Pass gimple iterator to
13397 get_mem_refs_of_builtin_call.
13398 (last_alloca_addr): New global.
13399 * asan.h (asan_emit_allocas_unpoison): Declare.
13400 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
13401 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
13402 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
13403 if function calls alloca.
13404 * gimple-fold.c (replace_call_with_value): Remove static keyword.
13405 * gimple-fold.h (replace_call_with_value): Declare.
13406 * internal-fn.c: Include asan.h.
13407 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
13408 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
13410 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13412 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
13413 (C_SELFTEST_FLAGS): New.
13414 (CPP_SELFTEST_FLAGS): New.
13415 (SELFTEST_DEPS): New, from deps of s-selftest.
13416 (C_SELFTEST_DEPS): New, from deps of s-selftest.
13417 (CPP_SELFTEST_DEPS): New.
13418 (selftest): Add dependency on s-selftest-c++.
13419 (s-selftest): Rename to...
13420 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
13421 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
13422 than SELFTEST_FLAGS.
13423 (selftest-gdb): Rename to...
13424 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
13426 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
13427 (selftest-valgrind): Rename to...
13428 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
13430 (selftest-valgrind): Reintroduce as an alias for
13431 selftest-c-valgrind.
13432 (s-selftest-c++): New.
13433 (selftest-c++-gdb): New.
13434 (selftest-c++-valgrind): New.
13436 2017-07-06 Olivier Hainque <hainque@adacore.com>
13438 * gcc.c (process_command): When deciding if undefined variables
13439 should be ignored when processing specs, accept "gcc -v" as well.
13441 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13443 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
13444 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
13446 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13448 * config/arm/arm-cpus.in (armv8-r): Add new entry.
13449 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
13450 * config/arm/arm-tables.opt: Regenerate.
13451 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
13453 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
13455 2017-07-06 Carl Love <cel@us.ibm.com>
13457 * ChangeLog: Clean up from mid air collision
13459 2017-07-06 Carl Love <cel@us.ibm.com>
13461 * config/rs6000/rs6000-c.c: Add support for built-in functions
13462 vector signed int vec_subc (vector signed int, vector signed int);
13463 vector signed __int128 vec_subc (vector signed __int128,
13464 vector signed __int128);
13465 vector unsigned __int128 vec_subc (vector unsigned __int128,
13466 vector unsigned __int128);
13467 vector signed int vec_sube (vector signed int, vector signed int,
13468 vector signed int);
13469 vector unsigned int vec_sube (vector unsigned int,
13470 vector unsigned int,
13471 vector unsigned int);
13472 vector signed __int128 vec_sube (vector signed __int128,
13473 vector signed __int128,
13474 vector signed__int128);
13475 vector unsigned __int128 vec_sube (vector unsigned __int128,
13476 vector unsigned __int128,
13477 vector unsigned __int128);
13478 vector signed int vec_subec (vector signed int, vector signed int,
13479 vector signed int);
13480 vector unsigned int vec_subec (vector unsigned int,
13481 vector unsigned int,
13482 vector unsigned int);
13483 vector signed __int128 vec_subec (vector signed __int128,
13484 vector signed __int128,
13485 vector signed__int128);
13486 vector unsigned __int128 vec_subec (vector unsigned __int128,
13487 vector unsigned __int128,
13488 vector unsigned __int128);
13489 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
13490 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
13491 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
13492 BU_ALTIVEC_OVERLOAD_X definitions.
13493 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
13494 * doc/extend.texi: Update the built-in documentation file for the new
13495 built-in functions.
13497 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13500 * diagnostic-show-locus.c (layout::layout): Use start and finish
13501 spelling location for the start and finish of each range.
13502 * genmatch.c (linemap_client_expand_location_to_spelling_point):
13503 Add unused aspect param.
13504 * input.c (expand_location_1): Add "aspect" param, and use it
13505 to access the correct part of the location.
13506 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
13508 (expand_location_to_spelling_point): Likewise.
13509 (linemap_client_expand_location_to_spelling_point): Add "aspect"
13510 param, and pass it to expand_location_1.
13512 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
13514 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
13515 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
13516 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
13517 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
13518 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
13519 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
13520 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
13521 _mm_maskz_getmant_ss): New intrinsics.
13522 (__builtin_ia32_getexpss128_mask): Changed to ...
13523 __builtin_ia32_getexpss128_round ... this.
13524 (__builtin_ia32_getexpsd128_mask): Changed to ...
13525 __builtin_ia32_getexpsd128_round ... this.
13526 * config/i386/i386-builtin-types.def
13527 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
13528 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
13529 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
13530 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
13531 __builtin_ia32_getmantss_mask_round): New builtins.
13532 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
13533 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
13534 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
13535 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
13536 * config/i386/sse.md
13537 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
13538 avx512f_sgetexp<mode><mask_scalar_name>
13539 <round_saeonly_scalar_name> ... this.
13540 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
13541 %0, %1, %2<round_saeonly_op3>}): Changed to ...
13542 vgetexp<ssescalarmodesuffix>
13543 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13544 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
13545 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
13546 avx512f_vgetmant<mode><mask_scalar_name>
13547 <round_saeonly_scalar_name> ... this.
13548 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
13549 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
13550 vgetmant<ssescalarmodesuffix>
13551 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
13552 %0<mask_scalar_operand4>, %1, %2
13553 <round_saeonly_scalar_mask_op4>, %3} ... this.
13554 * config/i386/subst.md (mask_scalar_operand4,
13555 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
13556 round_saeonly_scalar_nimm_predicate): New subst attributes.
13558 2017-07-06 Julia Koval <julia.koval@intel.com>
13560 * config/i386/i386.c (ix86_erase_embedded_rounding):
13561 Remove code for old rounding pattern.
13563 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
13565 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
13567 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
13569 * doc/sourcebuild.texi (Test Directives, Variants of
13570 dg-require-support): Add documentation for dg-require-stack-check.
13572 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
13574 * config/i386/subst.md (mask_scalar, round_scalar,
13575 round_saeonly_scalar): New meta-templates.
13576 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
13577 round_scalar_mask_operand3, round_scalar_mask_op3,
13578 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
13579 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
13580 round_saeonly_scalar_constraint,
13581 round_saeonly_scalar_prefix): New subst attribute.
13582 * config/i386/sse.md
13583 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
13584 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
13585 <round_scalar_name> ... this.
13586 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
13587 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
13588 <round_scalar_name> ... this.
13589 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
13590 <sse>_vm<code><mode>3<mask_scalar_name>
13591 <round_saeonly_scalar_name> ... this.
13592 (v<plusminus_mnemonic><ssescalarmodesuffix>
13593 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13594 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13595 v<plusminus_mnemonic><ssescalarmodesuffix>
13596 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13597 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13598 (v<multdiv_mnemonic><ssescalarmodesuffix>
13599 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13600 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13601 v<multdiv_mnemonic><ssescalarmodesuffix>
13602 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13603 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13604 (v<maxmin_float><ssescalarmodesuffix>
13605 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
13606 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
13607 v<maxmin_float><ssescalarmodesuffix>
13608 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13609 %0<mask_scalar_operand3>, %1, %<iptr>2
13610 <round_saeonly_scalar_mask_op3>} ... this.
13612 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
13614 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
13615 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
13617 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
13618 Alan Hayward <alan.hayward@arm.com>
13619 David Sherwood <david.sherwood@arm.com>
13621 * combine.c (simplify_if_then_else): Remove "enum" before
13623 * compare-elim.c (can_eliminate_compare): Likewise.
13624 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13626 (aarch64_lookup_simd_builtin_type): Likewise.
13627 (aarch64_simd_builtin_type): Likewise.
13628 (aarch64_init_simd_builtin_types): Likewise.
13629 (aarch64_simd_expand_args): Likewise.
13630 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
13632 (aarch64_reverse_mask): Likewise.
13633 (aarch64_simd_emit_reg_reg_move): Likewise.
13634 (aarch64_gen_adjusted_ldpstp): Likewise.
13635 (aarch64_ccmp_mode_to_code): Likewise.
13636 (aarch64_operands_ok_for_ldpstp): Likewise.
13637 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13638 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
13640 (aarch64_min_divisions_for_recip_mul): Likewise.
13641 (aarch64_reassociation_width): Likewise.
13642 (aarch64_get_condition_code_1): Likewise.
13643 (aarch64_simd_emit_reg_reg_move): Likewise.
13644 (aarch64_simd_attr_length_rglist): Likewise.
13645 (aarch64_reverse_mask): Likewise.
13646 (aarch64_operands_ok_for_ldpstp): Likewise.
13647 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13648 (aarch64_gen_adjusted_ldpstp): Likewise.
13649 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
13651 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
13652 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13653 (arm_lookup_simd_builtin_type): Likewise.
13654 (arm_simd_builtin_type): Likewise.
13655 (arm_init_simd_builtin_types): Likewise.
13656 (arm_expand_builtin_args): Likewise.
13657 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
13658 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
13659 (ft32_setup_incoming_varargs): Likewise.
13660 (ft32_function_arg): Likewise.
13661 (ft32_function_arg_advance): Likewise.
13662 (ft32_pass_by_reference): Likewise.
13663 (ft32_arg_partial_bytes): Likewise.
13664 (ft32_valid_pointer_mode): Likewise.
13665 (ft32_addr_space_pointer_mode): Likewise.
13666 (ft32_addr_space_legitimate_address_p): Likewise.
13667 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
13669 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
13670 (ix86_emit_outlined_ms2sysv_restore): Likewise.
13671 (iamcu_alignment): Likewise.
13672 (canonicalize_vector_int_perm): Likewise.
13673 (ix86_noce_conversion_profitable_p): Likewise.
13674 (ix86_mpx_bound_mode): Likewise.
13675 (ix86_operands_ok_for_move_multiple): Likewise.
13676 * config/microblaze/microblaze-protos.h
13677 (microblaze_expand_conditional_branch_reg): Likewise.
13678 * config/microblaze/microblaze.c
13679 (microblaze_expand_conditional_branch_reg): Likewise.
13680 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13682 (rs6000_reassociation_width): Likewise.
13683 (rs6000_invalid_binary_op): Likewise.
13684 (fusion_p9_p): Likewise.
13685 (emit_fusion_p9_load): Likewise.
13686 (emit_fusion_p9_store): Likewise.
13687 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
13689 (riscv_hard_regno_mode_ok_p): Likewise.
13690 (riscv_address_insns): Likewise.
13691 (riscv_split_symbol): Likewise.
13692 (riscv_legitimize_move): Likewise.
13693 (riscv_function_value): Likewise.
13694 (riscv_hard_regno_nregs): Likewise.
13695 (riscv_expand_builtin): Likewise.
13696 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
13697 (riscv_build_integer): Likewise.
13698 (riscv_split_integer): Likewise.
13699 (riscv_legitimate_constant_p): Likewise.
13700 (riscv_cannot_force_const_mem): Likewise.
13701 (riscv_regno_mode_ok_for_base_p): Likewise.
13702 (riscv_valid_base_register_p): Likewise.
13703 (riscv_valid_offset_p): Likewise.
13704 (riscv_valid_lo_sum_p): Likewise.
13705 (riscv_classify_address): Likewise.
13706 (riscv_legitimate_address_p): Likewise.
13707 (riscv_address_insns): Likewise.
13708 (riscv_load_store_insns): Likewise.
13709 (riscv_force_binary): Likewise.
13710 (riscv_split_symbol): Likewise.
13711 (riscv_force_address): Likewise.
13712 (riscv_legitimize_address): Likewise.
13713 (riscv_move_integer): Likewise.
13714 (riscv_legitimize_const_move): Likewise.
13715 (riscv_legitimize_move): Likewise.
13716 (riscv_address_cost): Likewise.
13717 (riscv_subword): Likewise.
13718 (riscv_output_move): Likewise.
13719 (riscv_canonicalize_int_order_test): Likewise.
13720 (riscv_emit_int_order_test): Likewise.
13721 (riscv_function_arg_boundary): Likewise.
13722 (riscv_pass_mode_in_fpr_p): Likewise.
13723 (riscv_pass_fpr_single): Likewise.
13724 (riscv_pass_fpr_pair): Likewise.
13725 (riscv_get_arg_info): Likewise.
13726 (riscv_function_arg): Likewise.
13727 (riscv_function_arg_advance): Likewise.
13728 (riscv_arg_partial_bytes): Likewise.
13729 (riscv_function_value): Likewise.
13730 (riscv_pass_by_reference): Likewise.
13731 (riscv_setup_incoming_varargs): Likewise.
13732 (riscv_print_operand): Likewise.
13733 (riscv_elf_select_rtx_section): Likewise.
13734 (riscv_save_restore_reg): Likewise.
13735 (riscv_for_each_saved_reg): Likewise.
13736 (riscv_register_move_cost): Likewise.
13737 (riscv_hard_regno_mode_ok_p): Likewise.
13738 (riscv_hard_regno_nregs): Likewise.
13739 (riscv_class_max_nregs): Likewise.
13740 (riscv_memory_move_cost): Likewise.
13741 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
13742 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
13743 (rl78_addr_space_address_mode): Likewise.
13744 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13746 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13747 (rs6000_reassociation_width): Likewise.
13748 (rs6000_invalid_binary_op): Likewise.
13749 (fusion_p9_p): Likewise.
13750 (emit_fusion_p9_load): Likewise.
13751 (emit_fusion_p9_store): Likewise.
13752 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
13753 (ok_for_simple_move_operands): Likewise.
13754 (ok_for_simple_move_strict_operands): Likewise.
13755 (ok_for_simple_arith_logic_operands): Likewise.
13756 (visium_legitimize_reload_address): Likewise.
13757 (visium_select_cc_mode): Likewise.
13758 (output_cbranch): Likewise.
13759 (visium_split_double_move): Likewise.
13760 (visium_expand_copysign): Likewise.
13761 (visium_expand_int_cstore): Likewise.
13762 (visium_expand_fp_cstore): Likewise.
13763 * config/visium/visium.c (visium_pass_by_reference): Likewise.
13764 (visium_function_arg): Likewise.
13765 (visium_function_arg_advance): Likewise.
13766 (visium_libcall_value): Likewise.
13767 (visium_setup_incoming_varargs): Likewise.
13768 (visium_legitimate_constant_p): Likewise.
13769 (visium_legitimate_address_p): Likewise.
13770 (visium_legitimize_address): Likewise.
13771 (visium_secondary_reload): Likewise.
13772 (visium_register_move_cost): Likewise.
13773 (visium_memory_move_cost): Likewise.
13774 (prepare_move_operands): Likewise.
13775 (ok_for_simple_move_operands): Likewise.
13776 (ok_for_simple_move_strict_operands): Likewise.
13777 (ok_for_simple_arith_logic_operands): Likewise.
13778 (visium_function_value_1): Likewise.
13779 (rtx_ok_for_offset_p): Likewise.
13780 (visium_legitimize_reload_address): Likewise.
13781 (visium_split_double_move): Likewise.
13782 (visium_expand_copysign): Likewise.
13783 (visium_expand_int_cstore): Likewise.
13784 (visium_expand_fp_cstore): Likewise.
13785 (visium_split_cstore): Likewise.
13786 (visium_select_cc_mode): Likewise.
13787 (visium_split_cbranch): Likewise.
13788 (output_cbranch): Likewise.
13789 (visium_print_operand_address): Likewise.
13790 * expmed.c (flip_storage_order): Likewise.
13791 * expmed.h (emit_cstore): Likewise.
13792 (flip_storage_order): Likewise.
13793 * genrecog.c (validate_pattern): Likewise.
13794 * hsa-gen.c (gen_hsa_addr): Likewise.
13795 * internal-fn.c (expand_arith_overflow): Likewise.
13796 * ira-color.c (allocno_copy_cost_saving): Likewise.
13797 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13798 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
13799 (process_invariant_for_inheritance): Likewise.
13800 * lra-eliminations.c (move_plus_up): Likewise.
13801 * omp-low.c (lower_oacc_reductions): Likewise.
13802 * simplify-rtx.c (simplify_subreg): Likewise.
13803 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
13804 (TARGET_CHKP_BOUND_MODE): Likewise..
13805 * targhooks.c (default_chkp_bound_mode): Likewise.
13806 (default_setup_incoming_vararg_bounds): Likewise.
13807 * targhooks.h (default_chkp_bound_mode): Likewise.
13808 (default_setup_incoming_vararg_bounds): Likewise.
13809 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
13810 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
13811 (have_whole_vector_shift): Likewise.
13812 * tree-vect-stmts.c (vectorizable_load): Likewise.
13813 * doc/tm.texi: Regenerate.
13815 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13817 Graceful degrade if Binutils PR21472 is not available.
13820 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
13821 .rodata in flash test fails.
13822 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
13823 * confgure: Regenerate.
13824 * config.in: Regenerate.
13825 * config/avr/avr.c (avr_asm_named_section)
13826 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
13827 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
13828 (avr_asm_init_sections): Same.
13830 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13832 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
13833 (fma<VH:mode>4_intrinsic): Likewise.
13834 (*fmsub<VCVTF:mode>4): Likewise.
13835 (*fmsub<VH:mode>4_intrinsic): Likewise.
13837 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13840 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
13841 Don't depend on "optimize > 0".
13842 (out_movhi_r_mr, out_movqi_mr_r): Same.
13843 (out_movhi_mr_r, out_movqi_r_mr): Same.
13844 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
13845 io_address_operand on "optimize > 0".
13847 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13849 * tree-loop-distribution.c: Add general explanantion on the pass.
13850 (generate_loops_for_partition): Mark distributed loop.
13851 (pg_add_dependence_edges): New parameter. Handle alias data
13852 dependence specially and record it in the parameter if asked.
13853 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
13854 (init_partition_graph_vertices, add_partition_graph_edge): New.
13855 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
13856 (free_partition_graph_vdata, build_partition_graph): New.
13857 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
13858 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
13859 (data_ref_segment_size, latch_dominated_by_data_ref): New.
13860 (compute_alias_check_pairs, version_loop_by_alias_check): New.
13861 (version_for_distribution_p, finalize_partitions): New.
13862 (distribute_loop): Handle alias data dependence specially. Factor
13863 out loop fusion code as functions and call these functions.
13865 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13867 * tree-loop-distribution.c (classify_partition): New parameter and
13868 better handle reduction statement.
13869 (rdg_build_partitions): Revise comment.
13870 (distribute_loop): Compute statements in all partitions and pass it
13871 to classify_partition.
13873 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13875 * tree-loop-distribution.c (enum partition_type): New.
13876 (struct partition): New field type.
13877 (partition_merge_into): Add parameter. Update partition type.
13878 (data_dep_in_cycle_p, update_type_for_merge): New functions.
13879 (build_rdg_partition_for_vertex): Compute partition type.
13880 (rdg_build_partitions): Dump partition type.
13881 (distribute_loop): Update calls to partition_merge_into.
13883 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13885 * tree-loop-distribution.c (struct ddr_hasher): New.
13886 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
13888 (classify_partition): Call get_data_dependence.
13889 (pg_add_dependence_edges): Ditto.
13890 (distribute_loop): Release data dependence hash table.
13892 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13894 * tree-loop-distribution.c (ref_base_address): Delete.
13895 (similar_memory_accesses): Rename ...
13896 (share_memory_accesses): ... to this. Check if partitions access
13897 the same memory reference.
13898 (distribute_loop): Call share_memory_accesses.
13900 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13902 * tree-loop-distribution.c (struct partition): New field recording
13903 its data reference.
13904 (partition_alloc, partition_free): Init and release data refs.
13905 (partition_merge_into): Merge data refs.
13906 (build_rdg_partition_for_vertex): Collect data refs for partition.
13907 (pg_add_dependence_edges): Change parameters from vector to bitmap.
13909 (distribute_loop): Remve data refs from vertice data of partition
13912 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13914 * tree-loop-distribution.c (params.h): Include header file.
13915 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
13916 (datarefs_vec): New global var.
13917 (create_rdg_vertices): Use datarefs_vec directly.
13918 (free_rdg): Don't free data references.
13919 (build_rdg): Update use. Don't free data references.
13920 (distribute_loop): Compute global variable for data references.
13921 Bail out if there are too many data references.
13923 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13925 * tree-loop-distribution.c (loop_nest): New global var.
13926 (build_rdg): Use loop directly, rather than loop nest.
13927 (pg_add_dependence_edges): Remove loop nest parameter. Use global
13929 (distribute_loop): Compute global variable loop nest. Update use.
13931 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13933 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
13934 (partition_merge_into): New parameter. Dump reason for fusion.
13935 (distribute_loop): Update use of partition_merge_into.
13937 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13939 * tree-loop-distribution.c (bb_top_order_index): New.
13940 (bb_top_order_index_size, bb_top_order_cmp): New.
13941 (stmts_from_loop): Use topological order.
13942 (pass_loop_distribution::execute): Compute and release topological
13943 order for basic blocks.
13945 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13947 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
13950 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13952 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
13953 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
13954 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
13955 * internal-fn.def (LOOP_DIST_ALIAS): New.
13956 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
13957 (fold_loop_internal_call): ... this.
13958 (vect_loop_dist_alias_call): New function.
13959 (set_uid_loop_bbs): Call fold_loop_internal_call.
13960 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
13963 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13966 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
13967 Require dead FLAGS_REG at the beginning of a peephole.
13969 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13972 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
13973 arguments in the call to __builtin_ia32_sbb_u32.
13974 (_subborrow_u64): Swap _X and _Y arguments in the call to
13975 __builtin_ia32_sbb_u64.
13977 2017-07-04 Jakub Jelinek <jakub@redhat.com>
13980 * tree-vrp.c (compare_assert_loc): Turn into a function template
13981 with stable template parameter. Only test if a->e is NULL,
13982 !a->e == !b->e has been verified already. Use e == NULL or
13983 e != NULL instead of e or ! e tests. If stable is true, don't use
13984 iterative_hash_expr, on the other side allow a or b or both NULL
13985 and sort the NULLs last.
13986 (process_assert_insertions): Sort using compare_assert_loc<false>
13987 instead of compare_assert_loc, later sort using
13988 compare_assert_loc<true> before calling process_assert_insertions_for
13989 in a loop. Use break instead of continue once seen NULL pointer.
13991 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13993 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13994 Cortex-R7 and Cortex-R8 processors.
13996 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13998 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
13999 uninitialized while src is not.
14001 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
14003 * common/config/arm/arm-common.c: Adjust include path for
14005 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
14006 (arm-cpu.h): Create in build directory. Adjust dependency rules.
14007 (arm-cpu-data.h): Likewise.
14008 (arm-cpu-cdata.h): Likewise.
14009 * config/arm/arm-cpu.h: Delete.
14010 * config/arm/arm-cpu-cdata.h: Delete.
14011 * config/arm/arm-cpu-data.h: Delete.
14013 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
14015 * config/arm/arm-cpus.in (cortex-a55): New.
14016 (cortex-a75): Likewise.
14017 (cortex-a75.cortex-a55): Likewise.
14018 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
14020 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
14021 * config/arm/arm-cpu-cdata.h: Regenerate.
14022 * config/arm/arm-cpu-data.h: Regenerate.
14023 * config/arm/arm-cpu.h: Regenerate.
14024 * config/arm/arm-tables.opt: Regenerate.
14025 * config/arm/arm-tune.md: Regenerate.
14027 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14029 * haifa-sched.c (sched_create_recovery_edges): Update profile.
14031 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14033 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
14036 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
14038 PR tree-optimization/81292
14039 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
14040 full_string_p, also call adjust_related_strinfos if the adjustment
14041 is simple, otherwise invalidate related strinfos.
14043 2017-07-04 Martin Liska <mliska@suse.cz>
14046 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
14047 newly created variable as DECL_IGNORED_P.
14049 2017-07-04 Martin Liska <mliska@suse.cz>
14052 * ipa-inline.c (inline_small_functions):
14053 Use xstrdup_for_dump.
14055 2017-07-04 Tom de Vries <tom@codesourcery.com>
14057 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
14059 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14062 * config/darwin.c (darwin_function_switched_text_sections):
14065 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
14067 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
14069 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14071 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
14073 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14075 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
14076 min_profitable_iters, and th as inclusive lower bounds.
14077 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
14078 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
14079 for min_profitable_iters and min_profitable_estimate.
14080 (vect_transform_loop): Treat th as an inclusive lower bound.
14081 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
14083 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14086 * config/darwin.c (darwin_function_switched_text_sections):
14087 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
14088 in two pieces, and suppress the use of buf.
14090 2017-07-03 Nathan Sidwell <nathan@acm.org>
14092 * hash-table.h (hash_table_mod1): Fix indentation.
14094 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14096 PR middle-end/81290
14097 * predict.c (force_edge_cold): Be more careful about propagation
14099 * profile-count.h (profile_probability::guessed,
14100 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
14102 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
14104 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
14106 * doc/invoke.texi (rcpc architecture extension): Document it.
14108 2017-07-03 Richard Biener <rguenther@suse.de>
14110 PR tree-optimization/60510
14111 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
14112 the scalar reduction PHI and use it.
14113 (vectorizable_reduction): Properly guard the single_defuse_cycle
14114 path for non-SLP reduction chains where we cannot use it.
14115 Rework reduc_def/index and vector type deduction. Rework
14116 vector operand gathering during reduction op code-gen.
14117 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
14118 chains dissolve the chain and leave it to non-SLP reduction
14121 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14123 * tree-data-ref.h (dr_alignment): Declare.
14124 * tree-data-ref.c (dr_alignment): New function.
14125 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
14126 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
14128 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
14130 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14132 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
14133 and base_misalignment fields.
14134 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
14135 * tree-data-ref.c: Include builtins.h.
14136 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
14137 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
14138 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
14139 * tree-vect-data-refs.c: Include tree-cfg.h.
14140 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
14141 fields instead of calculating an alignment here.
14142 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
14143 innermost_loop_behavior fields.
14145 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14147 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
14149 (DR_STEP_ALIGNMENT): New macro.
14150 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
14151 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
14152 (create_data_ref): Print it.
14153 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
14154 to tell whether the step preserves vector (mis)alignment.
14155 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14156 Move the check for an integer step and generalise to all INTEGER_CST.
14157 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
14158 Print the outer step alignment.
14160 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14162 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
14163 with offset_alignment.
14164 (DR_ALIGNED_TO): Delete.
14165 (DR_OFFSET_ALIGNMENT): New macro.
14166 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
14167 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
14168 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
14169 (create_data_ref): Likewise.
14170 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14171 (vect_analyze_data_refs): Likewise.
14172 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
14173 creating dummy innermost behavior.
14175 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14177 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
14178 with a "innermost_loop_behavior *" and refeence tree.
14179 * tree-data-ref.c (dr_analyze_innermost): Likewise.
14180 (create_data_ref): Update call accordingly.
14181 * tree-predcom.c (find_looparound_phi): Likewise.
14183 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14185 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
14186 fields with dr_wrt_vec_loop.
14187 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
14188 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
14189 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
14190 (vect_dr_behavior): New function.
14191 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14192 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
14193 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
14194 track whether the step preserves the misalignment.
14195 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14196 Use vect_dr_behavior.
14197 (vect_setup_realignment): Update call accordingly.
14198 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
14199 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
14200 call to vect_create_addr_base_for_vector_ref.
14201 (vect_create_cond_for_align_checks): Likewise.
14202 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
14203 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
14204 (vect_recog_mask_conversion_pattern): Likewise.
14205 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
14206 (new_stmt_vec_info): Remove redundant zeroing.
14208 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14210 * common/config/arm/arm-common.c (arm_be8_option): New function.
14211 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
14212 (ISA_ARMv6): Add isa_bit_be8.
14213 * config/arm/arm.h (arm_be8_option): Add prototype.
14214 (BE8_SPEC_FUNCTION): New define.
14215 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
14216 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
14217 (mlittle-endian): Similarly.
14218 (mbe8, mbe32): New options.
14219 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
14220 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
14222 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14224 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
14226 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14228 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
14229 (cleanup_tree_cfg_bb): Use it.
14230 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
14232 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
14234 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14237 * loop-doloop.c (add_test): Update profile.
14239 2017-07-03 Martin Liska <mliska@suse.cz>
14242 * sanopt.c (rewrite_usage_of_param): New function.
14243 (sanitize_rewrite_addressable_params): Likewise.
14244 (pass_sanopt::execute): Call rewrite_usage_of_param.
14246 2017-07-03 Richard Biener <rguenther@suse.de>
14248 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
14249 back to using VIEW_CONVERT_EXPR.
14251 2017-07-03 Martin Liska <mliska@suse.cz>
14254 * doc/extend.texi: Document when a resolver function is
14255 generated for target_clones.
14257 2017-07-03 Martin Liska <mliska@suse.cz>
14259 * asan.c (asan_emit_stack_protection): Unpoison just red zones
14260 and shadow memory of auto variables which are subject of
14261 use-after-scope sanitization.
14262 (asan_expand_mark_ifn): Add do set only when is_poison.
14264 2016-07-03 Richard Biener <rguenther@suse.de>
14266 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
14268 (vect_force_simple_reduction): Record reduction def -> phi mapping.
14269 (vectorizable_reduction): Perform reduction PHI creation when
14270 visiting a reduction PHI and adjust and simplify code generation
14271 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
14272 (vect_transform_loop): Visit reduction PHIs.
14273 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
14274 defs into the SLP tree.
14275 (vect_build_slp_tree): Reduction defs terminate the recursion.
14276 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
14278 (vect_get_vec_defs_for_stmt_copy): Export.
14279 (vect_get_vec_defs): Likewise.
14280 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
14282 (vect_get_vec_defs_for_stmt_copy): Declare.
14283 (vect_get_vec_defs): Likewise.
14285 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14287 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
14288 parameter with a "loop" parameter and use it instead of the
14289 loop containing DR_STMT. Don't check simple_iv when doing
14290 BB analysis. Describe the two analysis modes in the comment.
14292 2017-07-03 Tom de Vries <tom@codesourcery.com>
14294 PR tree-optimization/69468
14295 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
14296 (find_same_succ_bb): Handle ignore_edge_flags.
14298 2017-07-03 Tom de Vries <tom@codesourcery.com>
14300 PR tree-optimization/81192
14301 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
14303 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
14305 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
14307 2017-07-03 Tom de Vries <tom@codesourcery.com>
14309 PR tree-optimization/81192
14310 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
14311 BB_SAME_SUCC (bb) == NULL.
14313 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14315 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
14318 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14320 * dumpfile.c: Include profile-count.h
14321 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
14323 (insert_cond_bb): Update profile.
14324 * tree-cfg.h (insert_cond_bb): Update prototype.
14325 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
14326 * tree-dump.c: Do not include tree-cfg.
14328 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14330 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
14332 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14334 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
14337 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14339 * tree-complex.c (expand_complex_div_wide): update profile.
14341 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14342 Alan Hayward <alan.hayward@arm.com>
14343 David Sherwood <david.sherwood@arm.com>
14345 * Makefile.in (MACHMODE_H): Remove insn-modes.h
14346 (CORETYPES_H): New define.
14347 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
14348 (insn-modes-inline.h, s-modes-inline-h): New rules.
14349 (generated_files): Add insn-modes-inline.h.
14350 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
14351 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
14352 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
14353 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
14354 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
14355 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
14356 (build/gencodes.o, build/genconditions.o): Likewise.
14357 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
14358 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
14359 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
14360 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
14361 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
14362 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
14363 * coretypes.h: Include everything up to real.h for generators.
14364 Include insn-modes.h first. Include wide-int-print.h after
14365 wide-int.h. Include insn-modes-inline.h and then machmode.h.
14366 * machmode.h: Don't include insn-modes.h here.
14367 * function-tests.c: Remove includes of signop.h, machmode.h,
14368 double-int.h and wide-int.h.
14370 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
14372 * optc-save-gen.awk: Likewise.
14373 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
14374 * godump.c: Remove include of wide-int-print.h.
14375 * pretty-print.h: Likewise.
14376 * wide-int-print.cc: Likewise.
14377 * wide-int.cc: Likewise.
14378 * hash-map-tests.c: Remove include of signop.h.
14379 * hash-set-tests.c: Likewise.
14380 * rtl-tests.c: Likewise.
14381 * mkconfig.sh: Remove include of machmode.h.
14382 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
14384 (emit_insn_modes_inline_h): ...this new function. Emit the code
14385 into an insn-modes-inline.h header file, adding appropriate
14386 include guards and end comments.
14387 (emit_insn_modes_c_header): Remove include of machmode.h.
14388 (emit_min_insn_modes_c_header): Include coretypes.h rather than
14390 (main): Handle -i flag and call emit_insn_modes_inline_h when
14393 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14395 * tree-ssa-strlen.c (strinfo): Rename the length field to
14396 nonzero_chars. Add a full_string_p field.
14397 (compare_nonzero_chars, zero_length_string_p): New functions.
14398 (get_addr_stridx): Add an offset_out parameter.
14399 Use compare_nonzero_chars.
14400 (get_stridx): Update accordingly. Use compare_nonzero_chars.
14401 (new_strinfo): Update after above changes to strinfo.
14402 (set_endptr_and_length): Set full_string_p.
14403 (get_string_length): Update after above changes to strinfo.
14404 (unshare_strinfo): Update call to new_strinfo.
14405 (maybe_invalidate): Likewise.
14406 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
14407 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
14408 as a uhwi instead of an shwi. Update after above changes to
14409 strinfo and new_strinfo.
14410 (zero_length_string): Assert that chainsi contains full strings.
14411 Use zero_length_string_p. Update call to new_strinfo.
14412 (adjust_related_strinfos): Update after above changes to strinfo.
14413 Copy full_string_p from origsi.
14414 (adjust_last_stmt): Use zero_length_string_p.
14415 (handle_builtin_strlen): Update after above changes to strinfo and
14416 new_strinfo. Install the lhs as the string length if the previous
14417 entry didn't describe a full string.
14418 (handle_builtin_strchr): Update after above changes to strinfo
14420 (handle_builtin_strcpy): Likewise.
14421 (handle_builtin_strcat): Likewise.
14422 (handle_builtin_malloc): Likewise.
14423 (handle_pointer_plus): Likewise.
14424 (handle_builtin_memcpy): Likewise. Track nonzero characters
14425 that aren't necessarily followed by a nul terminator.
14426 (handle_char_store): Likewise.
14428 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14430 PR tree-optimization/80769
14431 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
14432 for malloc and calloc. Document the new invariant that all related
14433 strinfos have delayed lengths or none do.
14434 (verify_related_strinfos): Move earlier in file.
14435 (set_endptr_and_length): New function, split out from...
14436 (get_string_length): ...here. Also set the lengths of related
14438 (zero_length_string): Assert that chainsi has known (rather than
14440 (adjust_related_strinfos): Likewise.
14442 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14444 PR tree-optimization/81136
14445 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
14446 assert that two references with the same misalignment have the same
14447 compile-time misalignment if those compile-time misalignments
14450 2017-07-01 Andi Kleen <ak@linux.intel.com>
14452 * print-tree.c (print_node): Print all attributes.
14454 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14456 * cfg.c (scale_bbs_frequencies): New function.
14457 * cfg.h (scale_bbs_frequencies): Declare it.
14458 * cfgloopanal.c (single_likely_exit): Cleanup.
14459 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
14461 (scale_loop_profile): Likewise.
14462 (loop_version): Likewise.
14463 (create_empty_loop_on_edge): Update.
14464 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
14465 scale_loop_frequencies, scale_loop_profile, loopify,
14466 loop_version): Update prototypes.
14467 * modulo-sched.c (sms_schedule): Update.
14468 * predict.c (unlikely_executed_edge_p): Also check probability.
14469 (probably_never_executed_edge_p): Fix typo.
14470 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14471 * tree-parloops.c (gen_parallel_loop): Update.
14472 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
14473 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14474 * tree-ssa-loop-split.c (split_loop): Update.
14475 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14476 * tree-vect-loop-manip.c (vect_do_peeling): Update.
14477 (vect_loop_versioning): Update.
14478 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14480 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14482 * trans-mem.c (split_bb_make_tm_edge): Update profile.
14484 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14486 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
14487 to keep profile consistent.
14489 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14491 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
14492 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
14493 * profile-count.h (max_safe_multiplier): Make unsigned.
14494 (profile_count::guessed_zero): New.
14496 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14498 * bb-reorder.c (fix_up_crossing_landing_pad,
14499 fix_crossing_conditional_branches): Use make_single_succ_edge
14500 to keep profile consistent.
14502 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14504 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
14507 2017-07-01 Jakub Jelinek <jakub@redhat.com>
14510 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
14511 the right scopes, make sure cond_jump isn't preserved between multiple
14512 iterations. Search for fallthru edge whenever there are 3+ edges and
14513 use find_fallthru_edge for it.
14515 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14517 Patch by Alexander Monakov <amonakov@ispras.ru>
14518 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
14519 probabilities consistently.
14521 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14523 * pa.c (pa_expand_compare_and_swap_loop): Update call of
14524 emit_cmp_and_jump_insns.
14526 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14529 * tree-inline.c (expand_call_inline): Combine profile statuses.
14531 2017-06-30 Andrew Pinski <apinski@cavium.com>
14533 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
14534 fold_stmt returned true.
14536 2017-06-30 Nathan Sidwell <nathan@acm.org>
14538 * ggc.h (empty_string): Delete.
14539 * cfgexpand.c (expand_asm_stmt): Use plain "".
14540 * optabs.c (expand_asm_memory_barrier): Likewise.
14541 * stringpool.c (empty_string): Delete.
14542 (digit_vector, digit_string): Delete.
14543 (ggc_alloc_string): Use plain "", don't optimize single digit
14544 strings. Use ggc_alloc_atomic.
14546 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
14548 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
14549 comparison set and one other set, use the cost of the non-comparison
14552 2017-06-30 Nathan Sidwell <nathan@acm.org>
14554 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
14557 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
14559 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
14560 loops. Remove now unneeded calls to gimple_switch_set_label() that
14561 just set removed labels to NULL_TREE.
14563 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
14565 * tree-ssanames.c (set_range_info_raw): Abstract from ...
14566 (set_range_info): ...here. Only call set_range_info_raw if domain
14568 (set_nonzero_bits): Call set_range_info_raw.
14569 * tree-ssanames.h (set_range_info_raw): New.
14571 2017-06-30 Jakub Jelinek <jakub@redhat.com>
14574 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
14575 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
14576 of nonimmediate_operand and <store_mask_constraint> instead of m
14577 for the input operand. For V8FI iterator, always split if input
14578 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
14579 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
14580 <store_mask_predicate> instead of register_operand and
14581 <store_mask_constraint> instead of v for the input operand. Make
14582 sure both operands aren't MEMs for if not <mask_applied>.
14584 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
14586 * lto-wrapper.c (copy_file) Close both file descriptors before
14589 2017-06-30 Martin Liska <mliska@suse.cz>
14592 * multiple_target.c (create_dispatcher_calls): Make ifunc
14593 also for function that don't have calls or are not referenced.
14595 2017-06-30 Richard Biener <rguenther@suse.de>
14597 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
14598 analyze the first scalar stmt. Move vector type computation
14599 for the BB case here from ...
14600 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
14601 live operation processing in the SLP case properly.
14603 2017-06-30 Richard Biener <rguenther@suse.de>
14605 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
14607 2017-06-30 Martin Liska <mliska@suse.cz>
14610 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
14611 before BUILT_IN_UNWIND_RESUME when ASAN is used.
14613 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
14615 * doc/invoke.texi (AArch64): Add missing options and remove redundant
14618 2017-06-30 Richard Biener <rguenther@suse.de>
14620 PR tree-optimization/81249
14621 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
14622 condition reduction result to original scalar type.
14624 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14626 * profile-count.h (enum profile_quality): Fix typos and whitespace
14629 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14631 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
14632 type for branch probabilities.
14634 2017-06-29 Julian Brown <julian@codesourcery.com>
14635 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14637 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
14638 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
14639 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
14640 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
14642 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14644 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
14645 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
14646 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
14647 CC usage from generic code to here.
14648 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
14649 CC usage into the target macros.
14651 2017-06-29 Maya Rashish <coypu@sdf.org>
14653 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
14656 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14658 * arm/arm-builtins.c: Include profile-count.h
14659 * except.c (sjlj_emit_function_enter): Use
14660 profile_probability::unlikely.
14662 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14664 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
14665 and tocrel_offset be pointer args rather than implicitly using
14667 (legitimate_constant_pool_address_p, rs6000_emit_move,
14668 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
14669 tocrel_offset and use in toc_relative_expr_p call.
14670 (print_operand, print_operand_address): Use static tocrel_base_oac
14671 and tocrel_offset_oac.
14672 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
14675 2017-06-29 Maya Rashish <coypu@sdf.org>
14677 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
14679 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
14681 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
14682 objects, take into account only the alignment of 'op0' and 'mode1' if
14685 2017-06-29 Steve Ellcey <sellcey@cavium.com>
14687 * ccmp.c (ccmp_tree_comparison_p): New function.
14688 (ccmp_candidate_p): Update to use above function.
14689 (get_compare_parts): New function.
14690 (expand_ccmp_next): Update to use new functions.
14691 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
14693 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
14694 take mode as argument.
14695 * ccmp.h (expand_ccmp_expr): Add mode as argument.
14696 * expr.c (expand_expr_real_1): Pass mode as argument.
14698 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
14700 * combine.c (combine_instructions): Print insns to dump_file, together
14703 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14705 * asan.c (asan_emit_stack_protection): Update.
14706 (create_cond_insert_point): Update.
14707 * auto-profile.c (afdo_propagate_circuit): Update.
14708 * basic-block.h (struct edge_def): Turn probability to
14709 profile_probability.
14710 (EDGE_FREQUENCY): Update.
14711 * bb-reorder.c (find_traces_1_round): Update.
14712 (better_edge_p): Update.
14713 (sanitize_hot_paths): Update.
14714 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
14715 (make_single_succ_edge): Update.
14716 (check_bb_profile): Update.
14717 (dump_edge_info): Update.
14718 (update_bb_profile_for_threading): Update.
14719 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
14720 probabilitycount to 0.
14721 * cfgbuild.c (compute_outgoing_frequencies): Update.
14722 * cfgcleanup.c (try_forward_edges): Update.
14723 (outgoing_edges_match): Update.
14724 (try_crossjump_to_edge): Update.
14725 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
14726 (expand_gimple_tailcall): Update.
14727 (construct_init_block): Use make_single_succ_edge.
14728 (construct_exit_block): Use make_single_succ_edge.
14729 * cfghooks.c (verify_flow_info): Update.
14730 (redirect_edge_succ_nodup): Update.
14731 (split_edge): Update.
14732 (account_profile_record): Update.
14733 * cfgloopanal.c (single_likely_exit): Update.
14734 * cfgloopmanip.c (scale_loop_profile): Update.
14735 (set_zero_probability): Remove.
14736 (duplicate_loop_to_header_edge): Update.
14737 * cfgloopmanip.h (loop_version): Update prototype.
14738 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
14739 (force_nonfallthru_and_redirect): Update.
14740 (update_br_prob_note): Update.
14741 (rtl_verify_edges): Update.
14742 (purge_dead_edges): Update.
14743 (rtl_lv_add_condition_to_bb): Update.
14744 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
14745 * cgraphunit.c (init_lowered_empty_function): Update.
14746 (cgraph_node::expand_thunk): Update.
14747 * cilk-common.c: Include profile-count.h
14748 * dojump.c (inv): Remove.
14749 (jumpifnot): Update.
14750 (jumpifnot_1): Update.
14751 (do_jump_1): Update.
14753 (do_jump_by_parts_greater_rtx): Update.
14754 (do_compare_rtx_and_jump): Update.
14755 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
14756 do_jump_1. do_compare_rtx_and_jump): Update prototype.
14757 * dwarf2cfi.c: Include profile-count.h
14758 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
14759 (sjlj_emit_dispatch_table): Likewise.
14760 * explow.c: Include profile-count.h
14761 * expmed.c (emit_store_flag_force): Update.
14762 (do_cmp_and_jump): Update.
14763 * expr.c (compare_by_pieces_d::generate): Update.
14764 (compare_by_pieces_d::finish_mode): Update.
14765 (emit_block_move_via_loop): Update.
14766 (store_expr_with_bounds): Update.
14767 (store_constructor): Update.
14768 (expand_expr_real_2): Update.
14769 (expand_expr_real_1): Update.
14770 * expr.h (try_casesi, try_tablejump): Update prototypes.
14771 * gimple-pretty-print.c (dump_probability): Update.
14772 (dump_profile): New.
14773 (dump_gimple_label): Update.
14774 (dump_gimple_bb_header): Update.
14775 * graph.c (draw_cfg_node_succ_edges): Update.
14776 * hsa-gen.c (convert_switch_statements): Update.
14777 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
14778 (find_if_case_1): Update.
14779 (find_if_case_2): Update.
14780 * internal-fn.c (expand_arith_overflow_result_store): Update.
14781 (expand_addsub_overflow): Update.
14782 (expand_neg_overflow): Update.
14783 (expand_mul_overflow): Update.
14784 (expand_vector_ubsan_overflow): Update.
14785 * ipa-cp.c (good_cloning_opportunity_p): Update.
14786 * ipa-split.c (split_function): Use make_single_succ_edge.
14787 * ipa-utils.c (ipa_merge_profiles): Update.
14788 * loop-doloop.c (add_test): Update.
14789 (doloop_modify): Update.
14790 * loop-unroll.c (compare_and_jump_seq): Update.
14791 (unroll_loop_runtime_iterations): Update.
14792 * lra-constraints.c (lra_inheritance): Update.
14793 * lto-streamer-in.c (input_cfg): Update.
14794 * lto-streamer-out.c (output_cfg): Update.
14795 * mcf.c (adjust_cfg_counts): Update.
14796 * modulo-sched.c (sms_schedule): Update.
14797 * omp-expand.c (expand_omp_for_init_counts): Update.
14798 (extract_omp_for_update_vars): Update.
14799 (expand_omp_ordered_sink): Update.
14800 (expand_omp_for_ordered_loops): Update.
14801 (expand_omp_for_generic): Update.
14802 (expand_omp_for_static_nochunk): Update.
14803 (expand_omp_for_static_chunk): Update.
14804 (expand_cilk_for): Update.
14805 (expand_omp_simd): Update.
14806 (expand_omp_taskloop_for_outer): Update.
14807 (expand_omp_taskloop_for_inner): Update.
14808 * omp-simd-clone.c (simd_clone_adjust): Update.
14809 * optabs.c (expand_doubleword_shift): Update.
14810 (expand_abs): Update.
14811 (emit_cmp_and_jump_insn_1): Update.
14812 (expand_compare_and_swap_loop): Update.
14813 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
14814 * predict.c (predictable_edge_p): Update.
14815 (edge_probability_reliable_p): Update.
14816 (set_even_probabilities): Update.
14817 (combine_predictions_for_insn): Update.
14818 (combine_predictions_for_bb): Update.
14819 (propagate_freq): Update.
14820 (estimate_bb_frequencies): Update.
14821 (force_edge_cold): Update.
14822 * profile-count.c (profile_count::dump): Add missing space into dump.
14823 (profile_count::debug): Add newline.
14824 (profile_count::differs_from_p): Explicitly convert to unsigned.
14825 (profile_count::stream_in): Update.
14826 (profile_probability::dump): New member function.
14827 (profile_probability::debug): New member function.
14828 (profile_probability::differs_from_p): New member function.
14829 (profile_probability::differs_lot_from_p): New member function.
14830 (profile_probability::stream_in): New member function.
14831 (profile_probability::stream_out): New member function.
14832 * profile-count.h (profile_count_quality): Rename to ...
14833 (profile_quality): ... this one.
14834 (profile_probability): New.
14835 (profile_count): Update.
14836 * profile.c (compute_branch_probabilities): Update.
14837 * recog.c (peep2_attempt): Update.
14838 * sched-ebb.c (schedule_ebbs): Update.
14839 * sched-rgn.c (find_single_block_region): Update.
14840 (compute_dom_prob_ps): Update.
14841 (schedule_region): Update.
14842 * sel-sched-ir.c (compute_succs_info): Update.
14843 * stmt.c (struct case_node): Update.
14844 (do_jump_if_equal): Update.
14845 (get_outgoing_edge_probs): Update.
14846 (conditional_probability): Update.
14847 (emit_case_dispatch_table): Update.
14848 (expand_case): Update.
14849 (expand_sjlj_dispatch_table): Update.
14850 (emit_case_nodes): Update.
14851 * targhooks.c: Update.
14852 * tracer.c (better_p): Update.
14853 (find_best_successor): Update.
14854 * trans-mem.c (expand_transaction): Update.
14855 * tree-call-cdce.c: Update.
14856 * tree-cfg.c (gimple_split_edge): Upate.
14857 (move_sese_region_to_fn): Upate.
14858 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
14859 * tree-eh.c (lower_resx): Upate.
14860 (cleanup_empty_eh_move_lp): Upate.
14861 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14862 * tree-inline.c (copy_edges_for_bb): Update.
14863 (copy_cfg_body): Update.
14864 * tree-parloops.c (gen_parallel_loop): Update.
14865 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
14866 (gimple_gen_time_profiler): Update.
14867 * tree-ssa-dce.c (remove_dead_stmt): Update.
14868 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14869 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
14870 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
14871 (unloop_loops): Update.
14872 (try_peel_loop): Update.
14873 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14874 * tree-ssa-loop-split.c (connect_loops): Update.
14875 (split_loop): Update.
14876 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14877 (hoist_guard): Update.
14878 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
14879 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
14880 (value_replacement): Update.
14881 * tree-ssa-reassoc.c (branch_fixup): Update.
14882 * tree-ssa-tail-merge.c (replace_block_by): Update.
14883 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
14884 (create_edge_and_update_destination_phis): Update.
14885 (compute_path_counts): Update.
14886 (recompute_probabilities): Update.
14887 (update_joiner_offpath_counts): Update.
14888 (freqs_to_counts_path): Update.
14889 (duplicate_thread_path): Update.
14890 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
14891 (struct switch_conv_info): Update.
14892 (gen_inbound_check): Update.
14893 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14894 (vect_do_peeling): Update.
14895 (vect_loop_versioning): Update.
14896 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14897 (optimize_mask_stores): Update.
14898 * ubsan.c (ubsan_expand_null_ifn): Update.
14899 * value-prof.c (gimple_divmod_fixed_value): Update.
14900 (gimple_divmod_fixed_value_transform): Update.
14901 (gimple_mod_pow2): Update.
14902 (gimple_mod_pow2_value_transform): Update.
14903 (gimple_mod_subtract): Update.
14904 (gimple_mod_subtract_transform): Update.
14905 (gimple_ic): Update.
14906 (gimple_stringop_fixed_value): Update.
14907 (gimple_stringops_transform): Update.
14908 * value-prof.h: Update.
14910 2017-06-29 Carl Love <cel@us.ibm.com>
14912 * config/rs6000/rs6000-c.c: Add support for built-in functions
14913 vector signed int vec_signed (vector float);
14914 vector signed long long vec_signed (vector double);
14915 vector signed int vec_signed2 (vector double, vector double);
14916 vector signed int vec_signede (vector double);
14917 vector signed int vec_signedo (vector double);
14918 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
14919 instruction generator.
14920 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14921 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
14922 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
14924 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
14925 vunsignede_v2df): Add define_expands.
14926 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
14927 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
14928 VEC_UNSIGNEDO): Add definitions.
14929 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14930 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
14931 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
14932 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
14933 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
14934 * config/rs6000/altivec.h (vec_signed, vec_signed2,
14935 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
14936 vec_unsignede, vec_unsignedo): Add builtin defines.
14937 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
14939 * doc/extend.texi: Update the built-in documentation file for the
14940 new built-in functions.
14942 2017-06-29 Richard Biener <rguenther@suse.de>
14944 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
14945 reduction chains to LOOP_VINFO_REDUCTIONS.
14946 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
14947 SLP reductions after processing reduction chains.
14949 2017-06-29 Nathan Sidwell <nathan@acm.org>
14951 * builtins.c (fold_builtin_FUNCTION): Use
14952 lang_hooks.decl_printable_name.
14954 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
14956 PR middle-end/81194
14957 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
14958 with only one label.
14959 * stmt.c (expand_case): Assert NCASES is greater than one.
14961 2017-06-29 Richard Biener <rguenther@suse.de>
14963 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
14965 (group_case_labels): Likewise.
14966 (find_taken_edge): Push sanity checking on val to workers...
14967 (find_taken_edge_cond_expr): ... here
14968 (find_taken_edge_switch_expr): ... and here, handle cases
14969 with just a default label.
14970 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
14971 (group_case_labels): Likewise.
14972 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
14973 group_case_labels does anything cleanup the CFG again.
14975 2017-06-29 Bin Cheng <bin.cheng@arm.com>
14977 PR tree-optimization/81196
14978 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
14979 exit condition comparing two IVs.
14981 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
14983 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
14984 profile to the dummy entry at the end of the list of architectures.
14985 * config/arm/arm-cpu-cdata.h: Regenerated.
14987 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14988 Michael Collison <michael.collison@arm.com>
14991 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
14993 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
14994 (*aarch64_reg_<mode>3_minus_mask): New pattern.
14995 (*aarch64_<optab>_reg_di3_mask2): New pattern.
14996 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
14997 of shift when the shift amount is masked with constant equal to
14998 the size of the mode.
14999 * config/aarch64/predicates.md (subreg_lowpart_operator): New
15002 2017-06-29 Martin Liska <mliska@suse.cz>
15004 * config/i386/i386.opt: Change range from [1,5] to [0,5].
15006 2017-06-29 Yury Gribov <tetra2005@gmail.com>
15009 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
15012 (edge_growth_cache_entry::edge_growth_cache_entry): New
15014 (reset_edge_growth_cache): Update to use constructor.
15016 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15018 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
15019 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15020 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15022 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
15024 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
15025 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
15027 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
15029 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
15030 (*-linux-uclibc*): Add t-uclibc tmake_file.
15031 * config/t-musl: New.
15032 * config/t-uclibc: New.
15034 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
15036 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
15038 (gen_comm_data): Emit architectural setting of arch_prof.
15039 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
15041 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
15042 (armv8-m.base, armv8-m.main): Likewise.
15043 * arm-protos.h (arm_build_target): Add profile field.
15044 (arch_option): Likewise.
15045 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
15047 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
15048 arm_active_target.profile.
15050 2017-06-28 Richard Biener <rguenther@suse.de>
15052 PR middle-end/81227
15053 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
15054 TYPE_OVERFLOW_WRAPS.
15055 * match.pd (negate_expr_p): Likewise.
15056 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
15057 fold_build2, not fold_binary.
15059 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15061 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
15062 Convert memory address to Pmode.
15063 (aarch64_print_operand): Assert MEM operands are always Pmode.
15065 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15068 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
15069 Remove redundant if.
15070 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
15071 * config/arm/aarch-common-protos.h
15072 (aarch_forward_to_shift_is_not_shifted_re): Remove.
15073 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
15075 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
15078 * multiple_target.c (create_dispatcher_calls): Set the default
15079 clone to be static, not public.
15081 2017-06-28 Richard Biener <rguenther@suse.de>
15083 * tree-vect-loop.c (vectorizable_reduction): Move special
15084 cond reduction IV var creation ...
15085 (vect_create_epilog_for_reduction): ... here. Remove induction_index
15086 parameter. Use STMT_VINFO_VECTYPE.
15087 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
15090 2017-06-28 Martin Liska <mliska@suse.cz>
15093 * ipa-visibility.c (non_local_p): Handle visibility.
15095 2017-06-28 Martin Liska <mliska@suse.cz>
15098 * common.opt: Add IntegerRange to various options.
15099 * opt-functions.awk (integer_range_info): New function.
15100 * optc-gen.awk: Add integer_range_info to cl_options struct.
15101 * opts-common.c (decode_cmdline_option): Handle
15102 CL_ERR_INT_RANGE_ARG.
15103 (cmdline_handle_error): Likewise.
15104 * opts.c (print_filtered_help): Show valid interval in
15105 when --help is provided.
15106 * opts.h (struct cl_option): Add range_min and range_max fields.
15107 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
15109 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
15111 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
15112 (x * C EQ/NE y * C): New transformation.
15114 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
15116 * genmultilib (combination_space): Accept '+' in option names.
15118 2017-06-28 Martin Liska <mliska@suse.cz>
15121 * asan.c (instrument_derefs): Bail out inner references
15122 that are hard register variables.
15124 2017-06-28 Jakub Jelinek <jakub@redhat.com>
15127 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
15128 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
15130 2017-06-28 Richard Biener <rguenther@suse.de>
15132 * tree-vectorizer.h (vect_get_vec_defs): Remove.
15133 (vect_get_slp_defs): Adjust.
15134 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
15136 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
15138 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
15139 get_initial_defs_for_reduction instead of vect_get_vec_defs.
15140 (vectorizable_reduction): Adjust.
15141 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
15143 (vect_get_slp_defs): Likewise.
15144 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
15145 (vectorizable_bswap): Adjust.
15146 (vectorizable_call): Likewise.
15147 (vectorizable_conversion): Likewise.
15148 (vectorizable_assignment): Likewise.
15149 (vectorizable_shift): Likewise.
15150 (vectorizable_operation): Likewise.
15151 (vectorizable_store): Likewise.
15152 (vectorizable_condition): Likewise.
15153 (vectorizable_comparison): Likewise.
15155 2017-06-28 Michael Collison <michael.collison@arm.com>
15158 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
15160 (arm_gen_movmemqi): Removed unused variable 'i'.
15161 Convert 'for' loop into 'while' loop.
15162 (arm_expand_prologue): Remove last unnecessary set of insn.
15163 (thumb_pop): Remove unused variable 'pushed_words'.
15164 (thumb_exit): Remove last unnecessary set of regs_to_pop.
15166 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15168 * config/s390/predicates.md: Use s390_rel_address_ok_p.
15169 * config/s390/s390-protos.h: Add prototype of
15170 s390_rel_address_ok_p.
15171 * config/s390/s390.c (s390_got_symbol): New function.
15172 (s390_rel_address_ok_p): New function.
15173 (legitimize_pic_address): Use s390_rel_address_ok_p.
15174 (s390_load_got): Use s390_got_symbol.
15175 (s390_option_override): Issue error if
15176 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
15177 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
15179 * config/s390/s390.opt: New option mpic-data-is-text-relative.
15181 2017-06-27 Andrew Pinski <apinski@cavium.com>
15183 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
15184 (X * copysign (1.0, X)): New pattern.
15185 (X * copysign (1.0, -X)): New pattern.
15186 (copysign (-1.0, CST)): New pattern.
15188 2017-06-27 Joseph Myers <joseph@codesourcery.com>
15190 * genmultilib (combination_space): Remove variable.
15191 Validate reuse rules against regular expression for any sequence
15192 of multilib options in any order.
15194 2017-06-27 Michael Collison <michael.collison@arm.com>
15196 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
15197 call aarch64_split_simd_combine.
15198 * (aarch64_combine_internal<mode>): Delete pattern.
15199 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
15200 Allow register and subreg operands.
15202 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15204 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
15205 specific need, just fallback on defaults.
15206 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
15208 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15209 Olivier Hainque <hainque@adacore.com>
15211 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
15213 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
15214 targets. Pick a default if no particular attempt applied.
15215 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
15218 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15220 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
15221 (x86_64-wrs-vxworks7): Likewise.
15223 2017-06-27 Marek Polacek <polacek@redhat.com>
15226 * ubsan.c (instrument_null): Check get_base_address's result for null.
15228 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15230 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
15232 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15234 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
15235 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
15236 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
15237 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
15238 New function types.
15239 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
15240 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
15241 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
15242 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
15243 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
15244 BUILT_IN_FEUPDATEENV): New builtins.
15245 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
15246 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
15247 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
15248 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
15250 (builtin_structptr_types): Adjust size.
15251 * tree.c (builtin_structptr_types): Add four entries.
15253 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15254 Olivier Hainque <hainque@adacore.com>
15256 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
15257 (TLS_SYM): New local macro, forcing reference to __tls__ on
15258 link command lines for VxWorks 7 RTPs, triggering initialization
15260 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
15261 OS features TLS support, true for RTPs on VxWorks 7.
15262 * config/vxworks.c (vxworks_override_options): Setup emutls
15265 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15267 * predict.c (test_prediction_value_range): Use -1U instead of -1
15268 to avoid narrowing conversion warning.
15269 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
15270 to avoid narrowing conversion warning.
15271 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
15273 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
15275 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15277 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
15278 64bit configurations.
15279 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
15280 (SIZE_TYPE): Likewise.
15281 * config/vxworks.c (vxworks_emutls_var_fields): Use
15282 long_unsigned_type_node instead of unsigned_type_node as the offset
15283 field type, which is "pointer" mode in emutls.c.
15285 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15288 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
15290 PR middle-end/81207
15291 * gimple-fold.c (replace_call_with_call_and_fold): Handle
15292 gimple_vuse copying separately from gimple_vdef copying.
15294 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15296 * value-prof.c (free_hist): Remove call to memset and the enclosing if
15299 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15300 Olivier Hainque <hainque@adacore.com>
15302 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
15303 for all vxworks7 targets.
15304 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
15305 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
15306 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
15307 variations for VX6/VX7 and 32/64bits later on in ...
15308 (VXWORKS_LIB_SPEC): Leverage new macros.
15309 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
15310 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
15312 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15314 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
15315 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
15317 2017-06-26 Carl Love <cel@us.ibm.com>
15319 * config/rs6000/rs6000-c.c: Add support for built-in functions
15320 vector bool char vec_reve (vector bool char);
15321 vector signed char vec_reve (vector signed char);
15322 vector unsigned char vec_reve (vector unsigned char);
15323 vector bool int vec_reve (vector bool int);
15324 vector signed int vec_reve (vector signed int);
15325 vector unsigned int vec_reve (vector unsigned int);
15326 vector bool long long vec_reve (vector bool long long);
15327 vector signed long long vec_reve (vector signed long long);
15328 vector unsigned long long vec_reve (vector unsigned long long);
15329 vector bool short vec_reve (vector bool short);
15330 vector signed short vec_reve (vector signed short);
15331 vector double vec_reve (vector double);
15332 vector float vec_reve (vector float);
15333 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
15334 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
15335 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
15336 (altivec_vreve): New pattern.
15337 * config/rs6000/altivec.h (vec_reve): New define.
15338 * doc/extend.texi (vec_rev): Update the built-in documentation file
15339 for the new built-in functions.
15341 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15343 PR tree-optimization/71815
15344 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
15346 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
15348 (slsr_process_phi): Likewise.
15349 (replace_uncond_cands_and_profitable_phis): Don't replace a
15350 multiply candidate with a stride of 1 (copy or cast).
15351 (phi_incr_cost): Call uses_consumed_by_stmt rather than
15353 (lowest_cost_path): Likewise.
15354 (total_savings): Likewise.
15356 2017-06-26 Richard Biener <rguenther@suse.de>
15359 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
15360 Use def_builtin_pure for all gather builtins.
15362 2017-06-26 Richard Biener <rguenther@suse.de>
15364 PR tree-optimization/81203
15365 * tree-tailcall.c (find_tail_calls): Do not move stmts into
15366 non-dominating BBs.
15368 2017-06-26 Marek Polacek <polacek@redhat.com>
15371 * doc/invoke.texi: Document -Wmultistatement-macros.
15373 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
15375 * doc/sourcebuild.texi (ARM-specific attributes): Document new
15376 arm_neon_ok_no_float_abi effective target.
15378 2017-06-26 Richard Biener <rguenther@suse.de>
15380 PR tree-optimization/80928
15381 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
15382 (copy_bbs): Set BB_DUPLICATED flag early.
15383 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
15385 (execute_on_shrinking_pred): Likewise.
15386 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
15387 BB_DUPLICATED blocks.
15388 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
15389 iterate over all PHIs considering removal of *gsi.
15391 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
15393 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
15396 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15398 * config/rs6000/rs6000-string.c: (expand_block_clear,
15399 do_load_for_compare, select_block_compare_mode,
15400 compute_current_alignment, expand_block_compare,
15401 expand_strncmp_align_check, expand_strn_compare,
15402 expand_block_move, rs6000_output_load_multiple)
15403 Move functions related to string/block move/compare
15404 to a separate file.
15405 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
15406 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
15407 for this function which is now used in two files.
15408 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
15409 * config.gcc: Add rs6000-string.o to extra_objs for
15410 targets powerpc*-*-* and rs6000*-*-*.
15412 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
15415 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
15416 32-bit, since indexed is not valid for DImode.
15417 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
15418 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
15419 (define_peephole2 for Altivec d-form load): Add 32-bit support.
15420 (define_peephole2 for Altivec d-form store): Likewise.
15423 * multiple_target.c (create_dispatcher_calls): Only create the
15424 dispatcher call if the function is the default clone of a
15425 versioned function.
15427 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
15429 PR middle-end/80902
15430 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
15431 a call, force the call to not be a tail call.
15433 2017-06-23 Jeff Law <law@redhat.com>
15435 * doc/contrib.texi: Add entry for Steven Pemberton's work on
15438 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
15440 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
15441 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
15442 handling for early expansion of vector shifts (sl,sr,sra,rl).
15443 (builtin_function_type): Add vector shift right instructions
15444 to the unsigned argument list.
15446 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
15448 rtl-optimizatoin/79286
15449 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
15450 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
15451 trap. PIC register plus a const unspec without offset can never trap.
15453 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
15455 * tree.h (builtin_structptr_type): New type.
15456 (builtin_structptr_types): Declare new array.
15457 * tree.c (builtin_structptr_types): New array.
15458 (free_lang_data, build_common_tree_nodes): Use it.
15460 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
15463 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
15466 2017-06-22 Matt Turner <mattst88@gmail.com>
15468 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
15469 Lake models to skylake case. Assume skylake for unknown
15470 models with clflushopt.
15472 2017-06-22 Jeff Law <law@redhat.com>
15474 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
15475 frame sizes that do not satisfy aarch64_uimm12_shift.
15477 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
15479 * profile-count.h (apply_probability,
15480 apply_scale, probability_in): Fix checks for zero.
15482 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15484 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
15485 * doc/cppdiropts.texi (-I @var{dir}): Document it.
15487 2016-06-22 Richard Biener <rguenther@suse.de>
15489 * tree-vect-loop.c (vect_model_reduction_cost): Handle
15490 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
15491 REDUC_MAX_EXPR support.
15492 (vectorizable_reduction): Likewise.
15493 (vect_create_epilog_for_reduction): Likewise.
15495 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
15497 * match.pd (A / (1 << B) -> A >> B): New.
15498 * generic-match-head.c: Include optabs-tree.h.
15499 * gimple-match-head.c: Likewise.
15500 * optabs-tree.h (target_supports_op_p): New.
15501 * optabs-tree.c (target_supports_op_p): New.
15503 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15505 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
15506 $gcc_cv_ld --help output.
15507 (gcc_cv_ld_demangle): Likewise.
15508 (gcc_cv_ld_eh_frame_hdr): Likewise.
15509 (gcc_cv_ld_pie): Likewise.
15510 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
15511 (gcc_cv_ld_buildid): Likewise.
15512 (gcc_cv_ld_sysroot): Likewise.
15513 (ld_bndplt_support): Likewise.
15514 (ld_pushpopstate_support): Likewise.
15515 * configure: Regenerate.
15516 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
15518 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15521 * config/i386/sse.md (round<mode>2): Renumber match_dup and
15522 operands indexes to avoid gap between operands and match_dups.
15524 2017-06-21 Andrew Pinski <apinski@cavium.com>
15526 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
15527 Increment Arith_shift and Arith_shift_reg by 1.
15528 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
15530 * config/aarch64/aarch64.c (thunderx_tunings): Enable
15531 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
15532 (aarch64_strip_extend): Add new argument and test for it.
15533 (aarch64_cheap_mult_shift_p): New function.
15534 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
15535 add a cost if it is true.
15536 Update calls to aarch64_strip_extend.
15537 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
15539 2017-06-21 Andrew Pinski <apinski@cavium.com>
15541 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
15543 (thunderxt88): Likewise.
15544 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
15545 (thunderx_prefetch_tune): New variable.
15546 (thunderx2t99_prefetch_tune): Update for the correct values.
15547 (thunderxt88_tunings): New variable.
15548 (thunderx_tunings): Use thunderx_prefetch_tune instead of
15549 generic_prefetch_tune.
15550 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
15552 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15554 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
15555 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
15556 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
15557 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
15558 (aarch64_atomic_cas<mode>, GPI): Likewise.
15560 2017-06-21 Martin Liska <mliska@suse.cz>
15562 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
15563 statements on cold and hot labels.
15564 * predict.c (tree_estimate_probability_bb): Remove the
15565 prediction from this place.
15567 2017-06-21 Martin Liska <mliska@suse.cz>
15569 PR tree-optimization/79489
15570 * gimplify.c (maybe_add_early_return_predict_stmt): New
15572 (gimplify_return_expr): Call the function.
15573 * predict.c (tree_estimate_probability_bb): Remove handling
15575 * predict.def: Update comment about early return predictor.
15576 * gimple-predict.h (is_gimple_predict): New function.
15577 * predict.def: Change default value of early return to 66.
15578 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
15580 * passes.def: Put pass_strip_predict_hints to the beginning of
15583 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
15585 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
15586 FUNCTION_DECL declarations.
15587 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
15589 (dwaf2out_decl): Likewise.
15590 * godump.c (go_early_global_decl): Skip call to the real debug hook
15591 for FUNCTION_DECL declarations.
15592 * passes.c (rest_of_decl_compilation): Skip call to the
15593 early_global_decl debug hook for FUNCTION_DECL declarations, unless
15594 -fdump-go-spec is passed.
15596 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15598 * config/i386/i386.c (struct builtin_isa): New field pure_p.
15599 Reorder for compactness.
15600 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
15601 (def_builtin_pure, def_builtin_pure2): New functions.
15602 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
15604 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15606 * match.pd (nop_convert): New predicate.
15607 ((A +- CST1) +- CST2): Allow some NOP conversions.
15609 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15612 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
15613 with ctors/dtors if GOVD_SHARED is set.
15615 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15617 * config/aarch64/aarch64.md (movti_aarch64):
15618 Emit mov rather than orr.
15619 (movtf_aarch64): Likewise.
15620 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
15621 Emit mov rather than orr.
15623 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15625 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
15626 Swap alternatives, make integer dup more expensive.
15628 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15630 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
15631 Return true for non-tls symbols.
15633 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
15635 * config/aarch64/aarch64-cores.def (cortex-a55): New.
15636 (cortex-a75): Likewise.
15637 (cortex-a75.cortex-a55): Likewise.
15638 * config/aarch64/aarch64-tune.md: Regenerate.
15639 * doc/invoke.texi (-mtune): Document new values for -mtune.
15641 2017-06-21 Tom de Vries <tom@codesourcery.com>
15643 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
15644 stack_size feature.
15645 (Effective-Target Keywords, Other attributes): Suggest using
15646 dg-add-options stack_size feature to get stack limit in stack_size
15647 effective target documentation.
15649 2017-06-21 Julian Brown <julian@codesourcery.com>
15650 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15652 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
15653 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
15654 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
15656 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
15657 attribute type list for neon_multiply.
15658 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
15659 attribute type list for neon_multiply.
15660 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
15661 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
15662 attribute type list for neon_multiply.
15663 * config/arm/types.md (crypto_pmull): Add.
15664 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
15665 attribute type list.
15667 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
15669 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
15672 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15674 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
15675 to make sure not to dereference a NULL cost_classes_ptr pointer.
15677 2017-06-20 Carl Love <cel@us.ibm.com>
15679 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15680 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15681 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15682 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15683 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15684 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15685 VMULOSW): New enum "unspec" values.
15686 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15687 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15688 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15689 altivec_vmulosw): New patterns.
15690 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15691 VMULOSW): Add definitions.
15693 2017-06-20 Julia Koval <julia.koval@intel.com>
15695 * config/i386/i386.c: Fix rounding expand for new pattern.
15696 * config/i386/subst.md: Fix pattern (parallel -> unspec).
15698 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15700 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
15701 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
15703 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15705 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
15708 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15710 * config/aarch64/aarch64-cores.def: Rearrange to sort by
15711 architecture, then by implementer ID.
15712 * config/aarch64/aarch64-tune.md: Regenerate.
15714 2017-06-20 Richard Biener <rguenther@suse.de>
15716 PR middle-end/81097
15717 * fold-const.c (split_tree): Fold to type before negating.
15719 2017-06-20 David Malcolm <dmalcolm@redhat.com>
15721 * diagnostic-show-locus.c
15722 (selftest::test_fixit_deletion_affecting_newline): New function.
15723 (selftest::diagnostic_show_locus_c_tests): Call it.
15725 2017-06-20 Andreas Schwab <schwab@suse.de>
15728 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
15731 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15733 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
15734 __ARM_FEATURE_COPROC according to support.
15736 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15738 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15739 Rewritten to avoid overflow for > 32-bit pointers.
15742 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
15743 by removing enum keyword.
15744 (ubsan_type_descriptor): Likewise. Formatting fix.
15747 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
15748 splitter): Require TARGET_SSE2 in the condition.
15750 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
15753 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
15754 for doing vector set of SFmode on ISA 3.0.
15755 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
15756 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
15758 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
15759 SFmode value into a V4SF variable that was extracted from another
15760 V4SF variable without converting the element to double precision
15761 and back to single precision vector format.
15762 (vsx_insert_extract_v4sf_p9_2): Likewise.
15764 2017-06-19 Jakub Jelinek <jakub@redhat.com>
15766 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
15767 in UWHI to avoid undefined overflow.
15770 * ubsan.h (enum ubsan_encode_value_phase): New.
15771 (ubsan_encode_value): Change second argument to
15772 enum ubsan_encode_value_phase with default value of
15773 UBSAN_ENCODE_VALUE_GENERIC.
15774 * ubsan.c (ubsan_encode_value): Change second argument to
15775 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
15776 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
15777 create_tmp_var_raw instead of create_tmp_var and use a
15779 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
15780 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
15781 ubsan_encode_value callers.
15784 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
15785 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
15786 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
15788 2017-06-19 Richard Biener <rguenther@suse.de>
15790 PR middle-end/81118
15791 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
15792 estimates if we changed anything.
15794 2017-06-19 Richard Biener <rguenther@suse.de>
15796 PR tree-optimization/80887
15797 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
15798 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
15799 simplified lookups, then reset mprts_hook.
15800 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
15802 (try_to_simplify): Likewise.
15804 2017-06-19 Martin Liska <mliska@suse.cz>
15807 * gimplify.c (gimplify_switch_expr):
15808 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
15810 2017-06-19 Martin Liska <mliska@suse.cz>
15812 * doc/install.texi: Document that PGO runs in 4 stages.
15814 2017-06-19 Martin Liska <mliska@suse.cz>
15817 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
15818 to dispatcher function name.
15819 * multiple_target.c (replace_function_decl): New function.
15820 (create_dispatcher_calls): Redirect both edges and references.
15822 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
15824 * profile-count.c (profile_count::dump): Dump quality.
15825 (profile_count::differs_from_p): Update for unsigned val.
15826 * profile-count.h (profile_count_quality): New enum.
15827 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
15829 2017-06-19 Richard Biener <rguenther@suse.de>
15831 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
15832 struct function as arg.
15833 (estimate_numbers_of_iterations): Export overload with loop arg.
15834 (free_numbers_of_iterations_estimates_loop): Use an overload of
15835 free_numbers_of_iterations_estimates instead.
15836 * tree-cfg.c (remove_bb): Adjust.
15837 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
15838 * tree-parloops.c (gen_parallel_loop): Likewise.
15839 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
15841 (tree_unroll_loops_completely): Likewise.
15842 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
15843 Use an overload instead and export.
15844 (estimated_loop_iterations): Adjust.
15845 (max_loop_iterations): Likewise.
15846 (likely_max_loop_iterations): Likewise.
15847 (estimate_numbers_of_iterations): Take struct function as arg
15849 (loop_exits_before_overflow): Adjust.
15850 (free_numbers_of_iterations_estimates_loop): Use an overload.
15851 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
15852 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
15854 2017-06-19 Richard Biener <rguenther@suse.de>
15857 * ipa-prop.c (find_constructor_constant_at_offset): Handle
15858 RANGE_EXPR conservatively.
15860 2017-06-16 Carl Love <cel@us.ibm.com>
15862 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15863 definitions for vec_float, vec_float2, vec_floato,
15864 vec_floate built-ins.
15865 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
15866 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
15868 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
15869 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
15870 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
15871 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
15872 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
15873 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
15874 vec_floato): Add builtin defines.
15875 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
15876 Update the built-in documentation file for the new built-in
15879 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15881 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
15882 (mthumb): Mark as the negative of -marm.
15884 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15886 * doc/invoke.texi (ARM Options, -mcpu): Document supported
15888 (ARM Options, -mtune): Document that this accepts the same
15889 extension options as -mcpu.
15890 (ARM Options, -mfpu): Document addition of -mfpu=auto.
15892 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15894 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
15895 permitted extensions.
15897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15899 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
15900 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
15901 (armv8-m.main): Add option +nodsp.
15902 * config/arm/arm-cpu-cdata.h: Regenerated.
15904 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15906 * config/arm/t-fuchsia: New file.
15907 * config.gcc (arm*-*-fuchsia*): Use it.
15909 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15911 * config/arm/t-symbian: Rewrite for new option infrastructure.
15913 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15915 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
15916 (MULTILIB_REQUIRED): Likewise.
15918 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15920 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
15921 (MULTILIB_RESUE): Likewise.
15922 (MULTILIB_MATCHES): Likewise.
15923 (MULTLIB_REQUIRED): Likewise.
15925 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15927 * config/arm/t-rtems: Rewrite for new option framework.
15929 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15931 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
15932 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
15933 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
15934 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
15935 * config/arm/t-multilib: ... here.
15936 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
15937 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
15938 armv7-a and armv8*-a when A-profile libraries have not been built.
15939 * config/arm/t-rmprofile: Rewrite.
15941 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15943 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
15944 with a backslash. Remove the backslash after substituting unescaped
15946 * doc/fragments.texi (MULTILIB_REUSE): Document it.
15948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15950 * config.gcc: (arm*-*-*): When building a-profile libraries, force
15951 the driver to pass through the default setting of -mfloat-abi.
15952 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
15954 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
15955 (all_feat_combs): New rule.
15956 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
15958 * config/arm/t-aprofile: Rewrite.
15960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15962 * config/arm/arm.h (FPUTYPE_AUTO): Define.
15963 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
15964 fpu is not specified by the user/command-line.
15965 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
15966 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
15967 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
15968 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
15969 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
15970 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
15972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15974 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
15975 * config/arm/t-arm-elf: Rewritten.
15977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15979 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
15980 have some floating-point instructions.
15981 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
15982 (TARGET_MAYBE_HARD_FLOAT): New macro.
15983 * config/arm/arm-builtins.c (arm_init_builtins): Use
15984 TARGET_MAYBE_HARD_FLOAT.
15985 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
15987 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15989 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
15990 (configargs.h): Include it.
15991 (arm_print_hint_for_fpu_option): New function.
15992 (arm_parse_fpu_option): New function.
15993 (candidate_extension): New class.
15994 (arm_canon_for_multilib): New function.
15995 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
15996 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
15997 (ARCH_CANONICAL_SPECS): New macro.
15998 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
16000 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16002 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
16003 are set after handling multilib fragments. Set target_cpu_default2
16006 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16008 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
16010 (arm*-*-*): Set target_cpu_default2 to a quoted string.
16011 * config/arm/parsecpu.awk (check_cpu): Validate any extension
16013 (check_arch): Likewise.
16014 * config/arm/arm.c (arm_configure_build_target): Handle
16015 TARGET_CPU_DEFAULT being a string constant. Scan any feature
16016 options in the default.
16018 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16020 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
16021 when an option is an alias of another.
16022 * config/arm/parsecpu.awk (optalias): New parser token.
16023 (gen_comm_data): Mark non-alias options as such. Emit entries
16024 for extension aliases.
16025 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
16026 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
16027 (armv6kz, armv6zk, armv6t2): Likewise.
16028 (armv7): Make vfpv3-d16 an alias.
16029 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
16031 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
16032 Sort in canonical order.
16033 (armv8-a): Sort in canonical order.
16034 (armv8.1-a, armv8.2-a): Likewise.
16035 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
16037 (cortex-a9): Sort in canonical order.
16038 * config/arm/arm.c (selftests.h): Include it.
16039 (arm_test_cpu_arch_data): New function.
16040 (arm_run_self_tests): New function.
16041 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
16042 (targetm): Move declaration to the end of the file.
16043 * arm-cpu-cdata.h: Regenerated.
16045 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16047 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
16048 call to target_mode_check describing the type of option passed.
16049 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
16050 (arm_target_thumb_only): Use arm_parse_arch_option_name or
16051 arm_parse_cpu_option_name to match parameters against list of
16053 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
16054 arm_arch_core_flags data structure.
16055 * config/arm/arm-cpu_cdata.h: Regenerated.
16057 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16059 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
16061 (arm_print_hint_for_cpu_option): Likewise.
16062 (arm_print_hint_for_arch_option): Likewise.
16063 (arm_parse_cpu_option_name): Likewise.
16064 (arm_parse_arch_option_name): Likewise.
16065 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
16066 of entries in the all_fpus list.
16067 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
16068 (arm_parse_cpu_option_name): Declare.
16069 (arm_parse_arch_option_name): Declare.
16070 (arm_parse_option_features): Declare.
16071 (arm_intialize_isa): Declare.
16072 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
16074 (gen_comm_data): ... here. Make definitions non-static.
16075 * config/arm/arm-cpu-data.h: Regenerated.
16076 * config/arm/arm-cpu-cdata.h: Regenerated.
16078 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16080 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
16081 (cpu_arch_extension): New structure.
16082 (cpu_arch_option, arch_option, cpu_option): New structures.
16083 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
16084 architecture types.
16085 (gen_data): Generate new format data tables.
16086 * config/arm/arm.c (cpu_tune): New structure.
16087 (cpu_option, processors): Delete.
16088 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
16089 (arm_print_hint_for_cpu_option): ... this and ...
16090 (arm_print_hint_for_arch_option): ... this.
16091 (arm_parse_arch_cpu_name): Delete. Replace with ...
16092 (arm_parse_cpu_option_name): ... this and ...
16093 (arm_parse_arch_option_name): ... this.
16094 (arm_unrecognized_feature): Change type of target parameter to
16096 (arm_parse_arch_cpu_features): Delete. Replace with ...
16097 (arm_parse_option_features): ... this.
16098 (arm_configure_build_target): Rework to use new configuration data
16100 (arm_print_tune_info): Rework for new configuration data tables.
16101 * config/arm/arm-cpu-data.h: Regenerated.
16102 * config/arm/arm-cpu.h: Regenerated.
16104 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16106 * Makefile.in (OBJS): Move sbitmap.o from here ...
16107 (OBJS-libcommon): ... to here.
16109 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16111 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
16112 (ISA_ALL_CRYPTO): New macro.
16113 (ISA_ALL_SIMD): New macro
16114 (ISA_ALL_FP): New macro.
16115 * config/arm/arm.c (fpu_bitlist): Update initializer.
16116 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
16118 (arm9e): Add fpu. Add option for nofp
16119 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
16120 (arm926ej-s, arm1026ej-s): Likewise.
16121 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
16122 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
16123 neon-fp16, neon-vfpv4, nofp and nosimd.
16124 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
16125 (cortex-a8): Add fpu. Add option for nofp.
16126 (cortex-a9): Add fpu. Add options for nosimd and nofp.
16127 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
16128 (cortex-r4f): Add fpu.
16129 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
16130 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
16132 (cortex-r8): Likewise.
16133 (cortex-m4): Add fpu. Add option for nofp.
16134 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
16135 (cortex-a17.cortex-a7): Likewise.
16136 (cortex-a32): Add fpu. Add options for crypto and nofp.
16137 (cortex-a35, cortex-a53): Likewise.
16138 (cortex-a57): Add fpu. Add option for crypto.
16139 (cortex-a72, cortex-a73): Likewise.
16140 (exynos-m1): Likewise.
16141 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
16142 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
16143 (cortex-m33): Add fpu. Add option for nofp.
16144 * config/arm/arm-cpu-cdata.h: Regenerated
16145 * config/arm/arm-cpu-data.h: Regenerated.
16147 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16149 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
16150 (armv5te, armv5tej): Likewise.
16151 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
16152 (armv7): Add options fp and vfpv3-d16.
16153 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
16154 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
16156 (armv7ve): Likewise.
16157 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
16158 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
16159 (armv8-a): Add nocrypto option.
16160 (armv8.1-a, armv8.2-a): Likewise.
16161 (armv8-m.main): add options fp, fp.dp and nofp.
16163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16165 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
16167 (armv8-a+crc): Delete.
16168 (armv8.1-a): Add options simd, crypto and nofp.
16169 (armv8.2-a): Add options fp16, simd, crypto and nofp.
16170 (armv8.2-a+fp16): Delete.
16171 (armv8-m.main): Add option dsp.
16172 (armv8-m.main+dsp): Delete.
16173 (cortex-a8): Add fpu. Add nofp option.
16174 (cortex-a9): Add fpu. Add nofp and nosimd options.
16175 * config/arm/parsecpu.awk (gen_data): Generate option tables and
16176 link to main cpu and architecture data structures.
16177 (gen_comm_data): Only put isa attributes from the main architecture
16179 (option): New statement for architecture and CPU entries.
16180 * arm.c (struct cpu_option): New structure.
16181 (struct processors): Add entry for options.
16182 (arm_unrecognized_feature): New function.
16183 (arm_parse_arch_cpu_name): Ignore any characters after the first
16185 (arm_parse_arch_cpu_feature): New function.
16186 (arm_configure_build_target): Separate out any CPU and architecture
16187 features and parse separately. Don't error out if -mfpu=auto is
16188 used with only an architecture string.
16189 (arm_print_asm_arch_directives): New function.
16190 (arm_file_start): Call it.
16191 * config/arm/arm-cpu-cdata.h: Regenerated.
16192 * config/arm/arm-cpu-data.h: Likewise.
16193 * config/arm/arm-tables.opt: Likewise.
16195 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16197 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
16198 assembler when it is not -mfpu=auto.
16200 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16202 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
16203 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
16204 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
16205 (ASM_CPU_SPEC): Rewrite.
16206 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
16207 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
16208 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
16209 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
16210 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
16211 copied string is NUL-terminated. Also strip any characters prefixed
16213 (arm_rewrite_selected_arch): New function.
16214 (arm_rewrite_march): New function.
16216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16218 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
16219 (x_arm_cpu_string, x_arm_tune_string): Likewise.
16220 (march, mcpu, mtune): Convert to string-based options.
16221 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
16222 (arm_parse_arch_cpu_name): New function.
16223 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
16224 identify selected architecture or CPU.
16225 (arm_option_save): New function.
16226 (TARGET_OPTION_SAVE): Redefine.
16227 (arm_option_restore): Restore string options.
16228 (arm_option_print): Print string options.
16230 2017-06-16 Martin Sebor <msebor@redhat.com>
16232 PR tree-optimization/80933
16233 PR tree-optimization/80934
16234 * builtins.c (fold_builtin_3): Do not handle bcmp here.
16235 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
16236 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
16237 (gimple_fold_builtin): Call them.
16239 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16241 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
16242 as unlikely; update profile.
16244 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16246 * predict.c (force_edge_cold): Handle declaring edges impossible
16249 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16251 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
16253 (try_unroll_loop_completely): Fix reporting.
16255 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16257 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
16259 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
16262 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
16263 if given a non-constant argument for an intrinsic which requires a
16266 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16268 * profile.c (compare_freqs): New function.
16269 (branch_prob): Sort edge list.
16270 (find_spanning_tree): Assume that the list is priority sorted.
16272 2017-06-16 Richard Biener <rguenther@suse.de>
16274 PR tree-optimization/81090
16275 * passes.def (pass_record_bounds): Remove.
16276 * tree-pass.h (make_pass_record_bounds): Likewise.
16277 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
16278 make_pass_record_bounds): Likewise.
16279 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
16280 not free niter estimates at the beginning but at the end.
16281 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
16283 2017-06-16 Richard Biener <rguenther@suse.de>
16285 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
16286 initializer to workaround ICE in host GCC 4.8.
16288 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16290 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
16292 (clone_inlined_nodes): Update.
16294 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16296 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
16297 prefetch settings, and enable prefetching by default at -O3.
16299 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16301 * config/aarch64/aarch64.c (aarch64_override_options_internal):
16302 Set flag_prefetch_loop_arrays according to tuning data.
16304 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16306 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
16307 New tune structure.
16308 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
16309 [Unrelated to main purpose of the patch] Place the pointer field last
16310 to enable type checking errors when tune structure are wrongly merged.
16311 * config/aarch64/aarch64.c (generic_prefetch_tune,)
16312 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
16313 (thunderx2t99_prefetch_tune): New tune constants.
16314 (tune_params *_tunings): Update all tunings (no functional change).
16315 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
16316 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
16317 from tunings structures.
16319 2017-06-16 Jakub Jelinek <jakub@redhat.com>
16322 * ubsan.c (instrument_null): Add T argument, use it instead
16323 of computing it based on IS_LHS.
16324 (instrument_object_size): Likewise.
16325 (pass_ubsan::execute): Adjust instrument_null and
16326 instrument_object_size callers to pass gimple_get_lhs or
16327 gimple_assign_rhs1 result to it. Use instrument_null instead of
16328 calling get_base_address and instrument_mem_ref. Handle
16329 aggregate call arguments for object-size sanitization.
16331 2017-06-16 Yury Gribov <tetra2005@gmail.com>
16333 PR tree-optimization/81089
16334 * tree-vrp.c (is_masked_range_test): Validate operands of
16337 2017-06-15 Martin Sebor <msebor@redhat.com>
16340 * dumpfile.c (dump_register): Avoid calling memset to initialize
16341 a class with a default ctor.
16342 * gcc.c (struct compiler): Remove const qualification.
16343 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
16344 * hash-table.h: Ditto.
16345 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
16347 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
16348 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
16350 * params.h (struct param_info): Make struct members non-const.
16351 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
16352 with default initialization.
16353 * vec.h (vec_copy_construct, vec_default_construct): New helper
16355 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
16356 with vec_copy_construct.
16357 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
16358 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
16359 * doc/invoke.texi (-Wclass-memaccess): Document.
16361 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16363 * emit-rtl.h (is_leaf): Update comment about local
16364 register allocator.
16366 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16369 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
16370 for a variable to have a section before checking if the section has a
16372 Set section to.persistent if persistent attribute is set.
16373 Warn if .persistent attribute is used on an automatic variable.
16375 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
16377 PR rtl-optimization/80474
16378 * reorg.c (update_block): Do not ignore instructions in a delay slot.
16380 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
16382 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
16385 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
16387 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
16388 (casesi): Emit bounds checking as RTL.
16389 (casesi_internal_mips16_<mode>): Remove bounds checking.
16391 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
16393 * config/xtensa/xtensa.c (xtensa_option_override): Append
16394 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
16395 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
16396 xtensa_doloop_hooks): Define unconditionally.
16397 (xtensa_reorg_loops): Only call reorg_loops in the presence of
16399 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
16400 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
16401 for it in xtensa_option_override.
16402 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
16403 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
16405 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
16407 * doc/cppopts.texi: Document '-' special value to -MF.
16409 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
16411 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
16412 (cortex_a53_fconst): Likewise.
16413 (cortex_a53_fpmul): Likewise.
16414 (cortex_a53_f_load_64): Likewise.
16415 (cortex_a53_f_load_many): Likewise.
16416 (cortex_a53_advsimd_alu): Likewise.
16417 (cortex_a53_advsimd_alu_q): Likewise.
16418 (cortex_a53_advsimd_mul): Likewise.
16419 (cortex_a53_advsimd_mul_q): Likewise.
16420 (fpmac bypass): Add new bypass for fpmac-fpmac case.
16421 Add missing fmul, r2f_cvt and fconst cases.
16423 2017-06-14 Richard Biener <rguenther@suse.de>
16425 PR middle-end/81088
16426 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
16428 (fold_binary_loc): When associating do not treat pre-existing
16429 TREE_OVERFLOW on literal constants as a reason to allow
16430 TREE_OVERFLOW on associated literal constants.
16432 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16434 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
16435 (MASK_FEATURES): New macro.
16436 * config/sparc/sparc.c (sparc_option_override): Remove the special
16437 handling of -mfpu and generalize it to all MASK_FEATURES switches.
16439 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16441 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
16442 a division of 0 if non-call exceptions are enabled.
16444 2017-06-14 Andrew Pinski <apinski@cavium.com>
16445 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16448 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
16449 Improve vector initialization code gen for only variable case.
16451 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16453 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
16455 2017-06-14 Richard Biener <rguenther@suse.de>
16457 PR tree-optimization/81083
16458 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
16461 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16463 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
16464 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
16465 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
16466 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
16467 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
16468 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
16470 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16472 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
16473 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
16475 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16477 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
16479 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16481 * config/rs6000/t-rtems: Don't handle SPE.
16483 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16485 * config/rs6000/t-linux: Don't handle SPE.
16487 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16489 * config/rs6000/eabispe.h: Delete file.
16491 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16493 * config/rs6000/t-spe: Delete file.
16495 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16497 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
16498 (rs6000_legitimate_offset_address_p): Return false for anything in
16499 V2SImode or V2SFmode.
16501 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16503 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
16504 except V2SF and V2SI. Rearrange the vector modes, and add comments.
16505 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
16507 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
16508 (rs6000_legitimate_offset_address_p): Ditto.
16509 (rs6000_emit_move): Ditto.
16510 (rs6000_init_builtins): Remove V4HI_type_node.
16512 2017-06-13 Martin Liska <mliska@suse.cz>
16515 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
16516 (gate_asan): Likewise.
16517 * asan.h (asan_no_sanitize_address_p): Remove the function.
16518 (sanitize_flags_p): New function.
16519 * builtins.def: Fix coding style.
16520 * common.opt: Use renamed enum value.
16521 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
16522 * doc/extend.texi: Document no_sanitize attribute.
16523 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
16524 to SANITIZE_UNDEFINED_NONDEFAULT.
16525 * gcc.c (sanitize_spec_function): Use the renamed enum value.
16526 * gimple-fold.c (optimize_atomic_compare_exchange_p):
16527 Use sanitize_flags_p.
16528 * gimplify.c (gimplify_function_tree): Likewise.
16529 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
16530 * opts.c (parse_no_sanitize_attribute): New function.
16531 (common_handle_option): Use renamed enum value.
16532 * opts.h (parse_no_sanitize_attribute): Declare.
16533 * tree.c (sanitize_flags_p): New function.
16534 * tree.h: Declared here.
16535 * tsan.c: Use sanitize_flags_p.
16536 * ubsan.c (ubsan_expand_null_ifn): Likewise.
16537 (instrument_mem_ref): Likewise.
16538 (instrument_bool_enum_load): Likewise.
16539 (do_ubsan_in_current_function): Remove the function.
16540 (pass_ubsan::execute): Use sanitize_flags_p.
16541 * ubsan.h: Remove do_ubsan_in_current_function
16542 * tree-cfg.c (print_no_sanitize_attr_value): New function.
16543 (dump_function_to_file): Use it here.
16545 2017-06-13 Martin Jambor <mjambor@suse.cz>
16547 PR tree-optimization/80803
16548 PR tree-optimization/81063
16549 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
16550 (propagate_subaccesses_across_link): Enqueue subtree whenever
16551 necessary instead of relying on the caller.
16553 2017-06-13 Martin Jambor <mjambor@suse.cz>
16555 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
16556 that have a first_link.
16557 (sort_and_splice_var_accesses): Do not check first_link before
16559 (subtree_mark_written_and_enqueue): Likewise.
16560 (propagate_all_subaccesses): Likewise and do not stop at first
16561 parent with a first_link.
16563 2017-06-13 Martin Jambor <mjambor@suse.cz>
16565 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
16568 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16570 * match.pd: New pattern.
16572 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16574 * tree-vrp.c (is_masked_range_test): New function.
16575 (register_edge_assert_for): Determine ranges for
16578 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16580 PR tree-optimization/67328
16581 * fold-const.c (maskable_range_p): New function.
16582 (build_range_check): Generate bittests if possible.
16584 2017-06-13 Martin Liska <mliska@suse.cz>
16586 * gimple-pretty-print.c (dump_probability): Add new argument.
16587 (dump_edge_probability): Dump both probability and count.
16588 (dump_gimple_label): Likewise.
16589 (dump_gimple_bb_header): Likewise.
16591 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
16594 * config/avr/avr-devices.c: Fix indentation.
16595 * config/avr/gen-avr-mmcu-specs.c: Dito.
16597 2017-06-13 Richard Biener <rguenther@suse.de>
16599 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
16600 instead get vector type from stmt_info.
16601 (vectorizable_reduction): Adjust. Remove dead code.
16603 2017-06-13 Richard Biener <rguenther@suse.de>
16605 PR middle-end/81065
16606 * fold-const.c (extract_muldiv_1): Remove bogus distribution
16607 case of C * (x * C2 + C3).
16608 (fold_addr_of_array_ref_difference): Properly fold index difference.
16610 2017-06-12 David S. Miller <davem@davemloft.net>
16613 * config/sparc/sparc.md (return expander): Emit frame blockage if
16614 function uses alloca.
16616 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16618 * combine.c (make_field_assignment): Check len rather than the mode
16619 precision when calling force_to_mode.
16621 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
16623 Support multilibs and devices that see flash in RAM address range.
16626 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
16627 (avr_mcu_t) <flash_pm_offset>: New field.
16628 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
16629 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
16630 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
16631 (AVR_TINY_PM_OFFSET): Remove macro.
16632 * config/avr/avr.opt (-mshort-calls): New option.
16633 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16634 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
16635 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
16636 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
16637 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
16638 instead of avr_arch->have_jmp_call.
16639 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
16640 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
16641 avr_arch->flash_pm_offset to define.
16642 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
16643 new field flash_pm_offset. Add entry for avrxmega3.
16644 (avr_texinfo): Add entry for avrxmega3.
16645 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
16646 attiny212, attiny214,
16647 attiny412, attiny414, attiny416, attiny417,
16648 attiny814, attiny816, attiny817,
16649 attiny1614, attiny1616, attiny1617,
16650 attiny3214, attiny3216, attiny3217.
16651 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
16652 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
16653 (avr_print_operand_address) [AVR_TINY]: Same.
16654 (avr_asm_init_sections) <readonly_data_section>: Only patch
16655 callback if avr_arch->flash_pm_offset = 0.
16656 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
16657 for rodata if avr_arch->flash_pm_offset != 0.
16658 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
16659 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
16660 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
16661 (m_options): Append opt_rcall.
16662 (m_dirnames): Append dir_rcall.
16663 * config/avr/t-multilib: Regenerate.
16665 * configure.ac [target=avr]: Check whether avrxmega3 default
16666 linker description file works as needed.
16667 * configure: Regenerate.
16668 * doc/avr-mmcu.texi: Regenerate.
16669 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
16670 <__AVR_ARCH__>: Document avrxmega3 and 103.
16671 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
16672 <__AVR_SHORT_CALLS__>: Document it.
16673 <__AVR_PM_BASE_ADDRESS__>: Document it.
16674 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
16675 (AVR Variable Attributes) <progmem>: Document this is
16676 not needed for avrxmega3.
16677 (AVR Named Address Spaces) <__flash>: Dito.
16679 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
16681 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
16683 2017-06-12 Doug Rupp <rupp@adacore.com>
16685 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
16686 Append vxworks-stdint.h to the tm_file list.
16687 * config/vxworks-stdint.h: New file.
16689 2017-06-12 Martin Liska <mliska@suse.cz>
16691 PR tree-optimization/81041
16692 * tree-profile.c (gimple_gen_ic_func_profiler):
16693 Create an extra BB in profile-generate
16694 (gimple_gen_time_profiler): Likewise.
16696 2017-06-12 Jakub Jelinek <jakub@redhat.com>
16698 PR tree-optimization/81003
16699 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
16700 (update_range_test): Use it instead of force_gimple_operand_gsi.
16702 2017-06-12 Richard Biener <rguenther@suse.de>
16704 PR tree-optimization/81053
16705 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
16706 with backedge value not defined in loop. Simplify def stmt
16709 2017-06-11 Tom de Vries <tom@codesourcery.com>
16712 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
16714 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
16715 nvptx_cannot_force_const_mem.
16717 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16719 * opts.c (finish_options): Move test for flag_split_stack after
16720 it has been initialized.
16722 2017-06-11 Jason Merrill <jason@redhat.com>
16724 * tree.h (id_equal): New.
16725 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
16726 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
16727 instead of strcmp of IDENTIFIER_POINTER.
16729 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16731 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
16732 (mark_all_inlined_calls_cdtor): Fix formating.
16733 (inline_transform): Rescale profile before inlining.
16735 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16737 * cgraph.h (cgraph_edge::clone): Update prototype.
16738 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
16739 (cgraph_node::create_clone): Update.
16740 (cgraph_node::create_version_clone): Update.
16741 * tree-inline.c (copy_bb): Update.
16742 (expand_call_inline): Update.
16744 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16746 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
16747 factored out from ...
16748 (rs6000_emit_prologue): ... here.
16750 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16752 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
16753 factored out from ...
16754 (rs6000_emit_prologue): ... here.
16756 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16758 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
16760 (handle_missing_profiles): Fix computation of tp_first_run.
16761 (counts_to_freqs): Do not touch freqs when count is 0.
16763 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16765 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
16768 2017-06-10 Tom de Vries <tom@codesourcery.com>
16770 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
16771 attributes): Document signal effective target.
16773 2017-06-10 Tom de Vries <tom@codesourcery.com>
16775 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
16776 Document effective target stack_size.
16778 2017-06-09 David Malcolm <dmalcolm@redhat.com>
16780 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
16781 to the edit_context if they can be auto-applied.
16783 2017-06-9 Ian Lance Taylor <iant@golang.org>
16785 * opts.c (finish_options): If -fsplit-stack, disable implicit
16786 -forder-blocks-and-partition.
16787 * doc/invoke.texi (Optimize Options): Document that when using
16788 -fsplit-stack -forder-blocks-and-partition is not implicitly
16791 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16793 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
16794 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
16795 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
16796 * builtins.def (abort, trap, unreachable): Declare cold.
16797 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
16798 * tree-core.h (ECF_COLD): New.
16799 * tree.c (set_call_expr_flags): Handle ECF_COLD.
16800 (build_common_builtin_nodes): Mark unreachable and abort as cold.
16802 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16804 * predict.c (unlikely_executed_stmt_p): Cleanup.
16806 2017-06-09 Richard Biener <rguenther@suse.de>
16808 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
16809 model if the ref is always written to.
16811 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16813 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
16815 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16817 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
16820 2017-06-09 Tom de Vries <tom@codesourcery.com>
16823 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
16824 "target cannot support label values" when encountering LABEL_REF.
16826 2017-06-09 Martin Liska <mliska@suse.cz>
16828 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
16829 (gimple_gen_ic_func_profiler): Emit direct comparison
16830 of __gcov_indirect_call_callee with NULL.
16831 (gimple_gen_time_profiler): Change probability from
16832 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
16834 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16836 * profile.c (edge_gcov_counts): Turn to pointer.
16837 (compute_branch_probabilities, compute_branch_probabilities): Update.
16838 (branch_prob): Do not clear edge_gcov_count.
16839 * profile.h (edge_gcov_counts): Turn to pointer.
16840 (edge_gcov_count): Update.
16842 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16844 * gimple.h (gimple_check_failed): Mark cold.
16846 2017-06-09 Richard Biener <rguenther@suse.de>
16848 PR tree-optimization/66623
16849 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
16850 refactor check_reduction into two parts, properly computing
16851 whether we have to check reduction validity for outer loop
16854 2017-06-09 Richard Biener <rguenther@suse.de>
16856 PR tree-optimization/79483
16857 * graphite-scop-detection.c (order): New global.
16858 (get_order): Compute bb to order mapping that satisfies code
16859 generation constraints.
16860 (cmp_pbbs): New helper.
16861 (build_scops): Start domwalk at entry block, sort generated
16864 2017-06-09 Richard Biener <rguenther@suse.de>
16866 PR middle-end/81007
16867 * ipa-polymorphic-call.c
16868 (ipa_polymorphic_call_context::restrict_to_inner_class):
16869 Skip FIELD_DECLs with error_mark_node type.
16870 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
16873 2017-06-09 Martin Liska <mliska@suse.cz>
16875 * predict.c (struct branch_predictor): New struct.
16876 (test_prediction_value_range): New test.
16877 (predict_c_tests): New function.
16878 * selftest-run-tests.c (selftest::run_tests): Run the function.
16879 * selftest.h: Declare new tests.
16881 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
16884 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
16885 gen_add3_insn did not fail.
16886 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
16887 r0, construct that number in a temporary reg and add that reg to r0.
16888 If asked to put the result in r0 as well, fail.
16890 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
16892 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16893 for early expansion of vec_eqv.
16895 2017-06-08 Jakub Jelinek <jakub@redhat.com>
16897 PR middle-end/81005
16898 * ubsan.c (instrument_null): Avoid pointless code temporary.
16899 (pass_ubsan::execute): Instrument aggregate arguments of calls.
16901 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
16905 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
16908 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
16911 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16913 * predict.c (unlikely_executed_edge_p): Move ahead.
16914 (probably_never_executed_edge_p): Use it.
16916 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16918 PR middle-end/79988
16919 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
16920 gimple_call_builtin_p call.
16922 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16924 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
16925 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
16926 rtl_check_failed_type2, rtl_check_failed_code1,
16927 rtl_check_failed_code2, rtl_check_failed_code_mode,
16928 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
16929 rtvec_check_failed_bounds, rtl_check_failed_flag,
16930 _fatal_insn_not_found, _fatal_insn): Likewise.
16931 * tree.h (tree_contains_struct_check_failed,
16932 tree_check_failed, tree_not_check_failed,
16933 tree_class_check_failed, tree_range_check_failed,
16934 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
16935 tree_vec_elt_check_failed, phi_node_elt_check_failed,
16936 tree_operand_check_failed, omp_clause_check_failed,
16937 omp_clause_operand_check_failed, omp_clause_range_check_failed):
16940 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16942 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
16943 flag_branch_probabilities.
16944 * ipa-inline.c (edge_badness): Likewise.
16945 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16946 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
16947 * predict.c (maybe_hot_frequency_p): Likewise.
16948 (probably_never_executed): Likewise.
16949 * sched-ebb.c (schedule_ebbs): Likewise.
16950 * sched-rgn.c (find_single_block_region): Likewise.
16951 * tracer.c (tail_duplicate): Likewise.
16953 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16955 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
16956 longer requires x_flag_profile_use.
16958 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16960 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
16961 instead of flag_reorder_blocks_and_partition.
16962 * dbxout.c (dbxout_function_end): Likewise.
16963 * dwarf2out.c (gen_subprogram_die): Likewise.
16964 * haifa-sched.c (sched_create_recovery_edges): Likewise.
16965 * hw-doloop.c (reorg_loops): Likewise.
16966 * varasm.c (assemble_start_function,
16967 assemble_end_function): Likewise.
16968 (decide_function_section): Do not check for
16969 flag_reorder_blocks_and_partition.
16971 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16973 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16975 (chkp_get_hard_register_fake_addr_expr): Ditto.
16976 (chkp_build_addr_expr): Add check for hard reg case.
16977 (chkp_parse_array_and_component_ref): Ditto.
16978 (chkp_find_bounds_1): Ditto.
16979 (chkp_process_stmt): Don't generate bounds store for
16982 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16984 * predict.c (maybe_hot_bb_p): Do not check profile status.
16985 (maybe_hot_edge_p): Likewise.
16986 (probably_never_executed): Check for zero counts even if profile
16988 (unlikely_executed_edge_p): New function.
16989 (unlikely_executed_stmt_p): New function.
16990 (unlikely_executed_bb_p): New function.
16991 (set_even_probabilities): Use unlikely predicates.
16992 (combine_predictions_for_bb): Likewise.
16993 (predict_paths_for_bb): Likewise.
16994 (predict_paths_leading_to_edge): Likewise.
16995 (determine_unlikely_bbs): New function.
16996 (estimate_bb_frequencies): Use it.
16997 (compute_function_frequency): Use zero counts even if profile is
16999 * profile-count.h: Fix typo.
17001 2017-08-08 Julia Koval <julia.koval@intel.com>
17003 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
17004 _mm512_mask_cvtsepi16_storeu_epi8,
17005 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
17006 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
17007 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
17008 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
17009 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
17010 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
17011 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
17012 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
17013 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
17014 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
17015 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
17016 __builtin_ia32_pmovuswb256mem_mask,
17017 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
17018 __builtin_ia32_pmovwb512mem_mask): New builtins.
17020 2017-08-08 Julia Koval <julia.koval@intel.com>
17022 PR target/73350,80862
17023 * config/i386/subst.md (round): Fix round pattern.
17024 * config/i386/i386.c (ix86_erase_embedded_rounding):
17025 Fix erasing rounding for the fixed pattern.
17027 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17029 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
17031 2017-06-08 Martin Liska <mliska@suse.cz>
17033 PR gcov-profile/80911
17034 * gcov.c (block_info::block_info): New constructor.
17036 2017-06-07 Carl Love <cel@us.ibm.com>
17038 * config/rs6000/rs6000-c: The return type of the following
17039 built-in functions was implemented as int not long long. Fix sign
17040 of return value for the unsigned version of vec_mulo and vec_mule.
17041 vector unsigned long long vec_bperm (vector unsigned long long,
17042 vector unsigned char)
17043 vector signed long long vec_mule (vector signed int,
17045 vector unsigned long long vec_mule (vector unsigned int,
17046 vector unsigned int)
17047 vector signed long long vec_mulo (vector signed int,
17049 vector unsigned long long vec_mulo (vector unsigned int,
17050 vector unsigned int)
17051 * doc/extend.texi: Fix the documentation for the built-in
17054 2017-06-07 Carl Love <cel@us.ibm.com>
17057 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
17060 2017-06-07 Carl Love <cel@us.ibm.com>
17062 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
17063 support, Generate doublehv for signed int/float for BE case only.
17065 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
17067 * doc/invoke.texi (mcx16): Rewrite.
17069 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17071 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
17072 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
17073 *mov<mode>_softfloat, and an anonymous splitter): Use
17074 nonimmediate_operand instead of rs6000_nonimmediate_operand.
17076 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17078 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
17080 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17081 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
17082 (rs6000_debug_reg_global): Adjust.
17083 (rs6000_init_hard_regno_mode_ok): Adjust.
17084 (rs6000_dbx_register_number): Adjust.
17085 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
17086 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17087 Remove SPE_ACC and SPEFSCR.
17088 (REG_ALLOC_ORDER): Ditto.
17089 (FRAME_POINTER_REGNUM): Change to 111.
17090 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
17091 (REG_CLASS_NAMES): Ditto.
17092 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
17093 (REGISTER_NAMES): Ditto.
17094 (ADDITIONAL_REG_NAMES): Ditto.
17095 (rs6000_reg_names): Ditto.
17096 * config/rs6000/rs6000.md: Renumber some register number
17099 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17101 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
17103 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17104 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
17106 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
17107 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17108 Delete the SPE high registers.
17109 (REG_ALLOC_ORDER): Ditto.
17110 (enum reg_class): Remove SPE_HIGH_REGS.
17111 (REG_CLASS_NAMES): Ditto.
17112 (REG_CLASS_CONTENTS): Delete the SPE high registers.
17113 (REGISTER_NAMES): Ditto.
17114 (rs6000_reg_names): Ditto.
17115 * doc/tm.texi.in: Remove SPE as example.
17116 * doc/tm.texi: Regenerate.
17118 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17120 * config/rs6000/8540.md (ppc8540_brinc): Delete.
17121 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
17122 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
17123 * config/rs6000/rs6000.md (type): Remove "brinc".
17125 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17127 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
17128 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
17129 * config/rs6000/linuxspe.h: Delete file.
17130 * config/rs6000/rs6000.md: Don't include spe.md.
17131 * config/rs6000/spe.h: Delete file.
17132 * config/rs6000/spe.md: Delete file.
17133 * config/rs6000/t-rs6000: Remove spe.md.
17135 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17137 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
17138 (reg_or_none500mem_operand): Delete.
17139 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
17140 instead of reg_or_none500mem_operand.
17142 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17144 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17145 handling of SPE flags.
17146 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
17148 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17150 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17152 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
17153 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
17154 paired_divv2sf3): Similar.
17155 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
17156 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
17157 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
17159 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
17160 Rename the paired_* instruction patterns.
17161 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
17163 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
17164 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
17165 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
17166 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
17167 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
17168 PAIRED_VECTOR_MODE.
17169 (struct machine_function): Delete field spe_insn_chain_scanned_p.
17170 (spe_func_has_64bit_regs_p): Delete.
17171 (spe_expand_predicate_builtin): Delete.
17172 (spe_expand_evsel_builtin): Delete.
17173 (TARGET_DWARF_REGISTER_SPAN): Do not define.
17174 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
17175 (invalid_e500_subreg): Delete.
17176 (rs6000_legitimize_address): Always force_reg op2 as well, for
17177 paired single memory accesses.
17178 (rs6000_member_type_forces_blk): Delete.
17179 (rs6000_spe_function_arg): Delete.
17180 (rs6000_expand_unop_builtin): Delete SPE handling.
17181 (rs6000_expand_binop_builtin): Ditto.
17182 (spe_expand_stv_builtin): Delete.
17183 (bdesc_2arg_spe): Delete.
17184 (spe_expand_builtin): Delete.
17185 (spe_expand_predicate_builtin): Delete.
17186 (spe_expand_evsel_builtin): Delete.
17187 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
17188 (spe_init_builtins): Delete.
17189 (spe_func_has_64bit_regs_p): Delete.
17190 (savres_routine_name): Delete "info" parameter. Adjust callers.
17191 (rs6000_emit_stack_reset): Ditto.
17192 (rs6000_dwarf_register_span): Delete.
17193 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
17194 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
17195 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
17197 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
17199 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
17200 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
17201 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
17202 mulv2sf3, divv2sf3): Delete expanders.
17204 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17206 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
17208 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17210 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
17211 * config/rs6000/rs6000.c: Ditto.
17213 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17215 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
17216 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
17217 comparison_operator.
17219 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17221 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
17222 * config/rs6000/rs6000.opt: Ditto.
17223 * config/rs6000/t-rtems: Ditto.
17225 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17227 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
17228 TARGET_E500_SINGLE by 0, simplify.
17229 * config/rs6000/rs6000.c: Ditto.
17230 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
17231 (spe_build_register_parallel): Delete.
17232 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
17233 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
17234 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
17235 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
17236 (E500_CONVERT): Delete.
17237 * config/rs6000/spe.md: Remove many patterns and all define_constants.
17239 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17241 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
17242 * config/rs6000/dfp.md: Ditto.
17243 (negdd2, *negdd2_fpr): Merge.
17244 (absdd2, *absdd2_fpr): Merge.
17245 (negtd2, *negtd2_fpr): Merge.
17246 (abstd2, *abstd2_fpr): Merge.
17247 * config/rs6000/e500.h: Delete file.
17248 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
17249 TARGET_FPRS by 1 and simplify.
17250 * config/rs6000/rs6000-c.c: Ditto.
17251 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
17252 TARGET_DF_SPE by 0.
17253 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
17255 * config/rs6000/rs6000.md: Ditto.
17256 (floatdidf2, *floatdidf2_fpr): Merge.
17257 (move_from_CR_gt_bit): Delete.
17258 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
17259 (E500_CR_IOR_COMPARE): Delete.
17260 (All patterns that require !TARGET_FPRS): Delete.
17261 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
17263 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17265 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
17267 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17269 * graphds.c (add_edge): Intitialize edge's attached data.
17270 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
17271 pointer parameter. Call pointed function on each edge during
17272 graph traversing. Skip traversing the edge when the function
17274 (graphds_dfs, graphds_scc): Ditto.
17275 (for_each_edge): New parameter. Pass the new parameter to callback
17277 * graphds.h (skip_edge_callback): New function pointer type.
17278 (graphds_dfs, graphds_scc): New function pointer parameter.
17279 (graphds_edge_callback, for_each_edge): New parameter.
17281 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17283 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
17284 out code checking if runtime alias check is possible to below ...
17285 Call the new function.
17286 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
17287 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
17289 2017-06-07 Marek Polacek <polacek@redhat.com>
17292 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
17293 TYPE_OVERFLOW_WRAPS check.
17295 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17297 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
17298 if versioning is required.
17299 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
17300 peeling with the check for versioning.
17302 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17304 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
17305 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
17306 Set true to new parameter if new ssa variable is defined.
17307 (vect_gen_vector_loop_niters): Refactor. Set range information
17308 for the new vector loop bound variable.
17309 (vect_do_peeling): Ditto.
17311 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17313 * tree-affine.c (ssa.h): Include header file.
17314 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
17315 has wrapping overflow behavior.
17317 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17319 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
17321 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17323 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
17324 (tree_to_aff_combination): ... here.
17326 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17328 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
17329 reg_pressure model function.
17330 (ivopts_global_cost_for_size): Delete.
17331 (determine_set_costs, iv_ca_recount_cost): Call new model function
17332 ivopts_estimate_reg_pressure.
17334 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17336 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
17337 expensive than udiv. Remove floating point cases from mod.
17339 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17341 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
17342 Increase idiv cost.
17344 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17346 * config/aarch64/aarch64.md
17347 (copysignsf3): Fix mask generation.
17349 2017-06-07 Jakub Jelinek <jakub@redhat.com>
17351 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
17353 (class dump_manager): Add register_dumps method.
17354 * dumpfile.c: Include langhooks.h.
17355 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
17356 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
17357 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
17358 (dump_manager::dump_register): Start with 512 entries instead of 32.
17359 (dump_manager::register_dumps): New method.
17360 * toplev.c (general_init): Instead of invoking register_dumps
17361 langhook, invoke register_dumps method on the dump manager.
17362 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
17365 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
17367 * doc/md.texi: Clarify the restrictions on a define_insn condition.
17368 Say that # requires an associated define_split to exist, and that
17369 the define_split must be suitable for use after register allocation.
17371 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17373 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
17374 (compute_outgoing_frequencies): Also initialize zero counts.
17375 (find_many_sub_basic_blocks): Do not produce uninitialized profile
17376 around loops; preserve more of profile when nothing changes.
17378 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
17380 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
17382 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
17383 * config/arm/arm-cpu-cdata.h: Regenerate.
17384 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
17385 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
17386 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
17387 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
17388 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
17390 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
17392 * config/arm/t-rmprofile: Likewise.
17393 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
17395 2017-06-06 David S. Miller <davem@davemloft.net>
17398 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
17399 blockage if function uses alloca.
17401 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17403 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
17404 New "uid" fields to hold pretty-print IDs of group and ref.
17405 Memory references are now identified as <group_id>:<ref_id>
17406 instead of using [random] addresses.
17407 (dump_mem_details): Simplify, no functional change.
17408 (dump_mem_ref): Simplify and make output more concise.
17409 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
17410 (find_or_create_group): Initialize group uid.
17411 (record_ref): Initialize ref uid. Improve debug output.
17412 (prune_group_by_reuse, should_issue_prefetch_p,)
17413 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
17414 (mark_nontemporal_store, determine_loop_nest_reuse):
17415 Improve debug output.
17417 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17419 * dbgcnt.def (prefetch): New debug counter.
17420 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
17421 (schedule_prefetches): Stop issueing prefetches if debug counter
17424 2017-06-06 Tom de Vries <tom@codesourcery.com>
17426 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
17427 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
17429 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17431 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
17432 Use aarch64_reg_or_zero predicate for operand 4.
17433 (aarch64_compare_and_swap<mode> define_insn_and_split):
17434 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
17435 (aarch64_store_exclusive<mode>): Likewise for operand 2.
17437 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17439 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
17440 (arm_compute_save_core_reg_mask): This.
17441 (thumb1_compute_save_reg_mask): Rename into ...
17442 (thumb1_compute_save_core_reg_mask): This.
17443 (arm_compute_save_reg0_reg12_mask): Adapt comment.
17444 (arm_compute_frame_layout): Likewise.
17446 2017-06-06 Richard Biener <rguenther@suse.de>
17448 PR tree-optimization/80974
17449 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
17450 keep or clear leaders SSA info.
17452 2017-06-06 Tom de Vries <tom@codesourcery.com>
17454 * config/nvptx/nvptx.c (split_mode_p): New function.
17455 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
17457 2017-06-06 Tom de Vries <tom@codesourcery.com>
17459 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
17461 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17464 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
17467 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17469 * shrink-wrap.c (handle_simple_exit): Update profile.
17470 (try_shrink_wrapping): Upate profile.
17472 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17474 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
17475 (tree_guess_outgoing_edge_probabilities): New.
17476 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
17477 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
17479 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17481 * ipa-split.c (split_function): Initialize return bb profile.
17483 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17485 * profile.c (compute_branch_probabilities): Also initialize
17486 EXIT_BLOCK profile.
17488 2017-06-06 Richard Biener <rguenther@suse.de>
17490 PR tree-optimization/80928
17491 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
17492 (vect_analyze_loop_operations): Properly guard analysis for
17494 (vect_transform_loop): Likewise.
17495 (vect_analyze_loop_2): Also reset SLP type on PHIs.
17496 (vect_model_induction_cost): Do not cost for pure SLP.
17497 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
17498 of induction in inner loop vectorization.
17499 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
17500 (vect_get_and_check_slp_defs): Handle vect_induction_def.
17501 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
17503 (vect_analyze_slp_cost_1): Cost induction.
17504 (vect_detect_hybrid_slp_stmts): Handle PHIs.
17505 (vect_get_slp_vect_defs): Likewise.
17506 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
17507 (vect_transform_stmt): Handle SLP reductions.
17508 * tree-vectorizer.h (vectorizable_induction): Adjust.
17510 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17512 * config/rs6000/rs6000.c (make_resolver_func): Update
17513 init_lowered_empty_function call.
17515 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
17517 * doc/invoke.texi: Document the -fprofile-abs-path option.
17518 * common.opt (fprofile-abs-path): New option.
17519 * gcov-io.h (gcov_write_filename): Declare.
17520 * gcov-io.c (gcov_write_filename): New function.
17521 * coverage.c (coverage_begin_function): Use gcov_write_filename.
17522 * profile.c (output_location): Likewise.
17524 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17526 * shring-wrap.c: Revert accidental commit.
17528 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
17530 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
17532 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17534 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
17536 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
17537 profile in callgraph edge.
17538 * profile-count.h (apply_probability): If THIS is 0, then result is 0
17539 (apply_scale): Likewise.
17540 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
17541 Also scale profile when inlining function with zero profile.
17542 (initialize_cfun): Update exit block profile even when it is zero.
17543 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
17544 when profile is read.
17546 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17548 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
17549 (CLONE_*): New constants to define the processors we can generate
17550 code for with the target_clone attribute.
17551 (rs6000_clone_map): New array to identify which clone processors
17552 the current program is running on.
17553 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
17554 target_clone attribute.
17555 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
17556 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
17557 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
17558 (cpu_expand_builtin): Add support for target_clone attribute.
17559 (rs6000_valid_attribute_p): Allow "default" attribute.
17560 (get_decl_name): New debug function to simplify printing the
17561 current function name in debugging statements.
17562 (rs6000_clone_priority): New functions to support the target_clone
17563 attribute, and be able to generate code to switch between ISA 2.05
17564 through ISA 3.0 (power6 through power9).
17565 (rs6000_compare_version_priority): Likewise.
17566 (rs6000_get_function_versions_dispatcher): Likewise.
17567 (make_resolver_func): Likewise.
17568 (add_condition_to_bb): Likewise.
17569 (dispatch_function_versions): Likewise.
17570 (rs6000_generate_version_dispatcher_body): Likewise.
17571 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
17572 (fusion_gpr_load_p): Fix a spacing issue.
17573 * doc/extend.texi (Common Function Attributes): Document that the
17574 PowerPC supports the target_clone attribute.
17576 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17578 * config/arm/arm.h: explain F symbol found in description of ARM
17579 register allocation in its legend.
17581 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17583 * config/mips/frame-header-opt.c: Include profile-count.h.
17584 * config/riscv/riscv.c: Include profile-count.h
17586 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17588 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
17590 (sm_set_flag_if_changed): Add bbs field.
17591 (execute_sm_if_changed_flag_set): Pass BBS.
17592 (execute_sm): Update.
17594 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17596 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
17599 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17601 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
17602 (peephole2): New peephole2 to emit the above.
17603 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
17605 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17607 * config/aarch64/aarch64.c (define_peephole2 above
17608 *sub_<shift>_<mode>): New peephole.
17610 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17612 * config/i386/i386.c (make_resolver_func): Update.
17613 * Makefile.in: Add profile-count.h and profile-count.o
17614 * auto-profile.c (afdo_indirect_call): Update to new API.
17615 (afdo_set_bb_count): Update.
17616 (afdo_propagate_edge): Update.
17617 (afdo_propagate_circuit): Update.
17618 (afdo_calculate_branch_prob): Update.
17619 (afdo_annotate_cfg): Update.
17620 * basic-block.h: Include profile-count.h
17621 (struct edge_def): Turn count to profile_count.
17622 (struct basic_block_def): Likewie.
17623 (REG_BR_PROB_BASE): Move to profile-count.h
17624 (RDIV): Move to profile-count.h
17625 * bb-reorder.c (max_entry_count): Turn to profile_count.
17626 (find_traces): Update.
17627 (rotate_loop):Update.
17628 (connect_traces):Update.
17629 (sanitize_hot_paths):Update.
17630 * bt-load.c (migrate_btr_defs): Update.
17631 * cfg.c (RDIV): Remove.
17632 (init_flow): Use alloc_block.
17633 (alloc_block): Uninitialize count.
17634 (unchecked_make_edge): Uninitialize count.
17635 (check_bb_profile): Update.
17636 (dump_edge_info): Update.
17637 (dump_bb_info): Update.
17638 (update_bb_profile_for_threading): Update.
17639 (scale_bbs_frequencies_int): Update.
17640 (scale_bbs_frequencies_gcov_type): Update.
17641 (scale_bbs_frequencies_profile_count): New.
17642 * cfg.h (update_bb_profile_for_threading): Update.
17643 (scale_bbs_frequencies_profile_count): Declare.
17644 * cfgbuild.c (compute_outgoing_frequencies): Update.
17645 (find_many_sub_basic_blocks): Update.
17646 * cfgcleanup.c (try_forward_edges): Update.
17647 (try_crossjump_to_edge): Update.
17648 * cfgexpand.c (expand_gimple_tailcall): Update.
17649 (construct_exit_block): Update.
17650 * cfghooks.c (verify_flow_info): Update.
17651 (dump_bb_for_graph): Update.
17652 (split_edge): Update.
17653 (make_forwarder_block): Update.
17654 (duplicate_block): Update.
17655 (account_profile_record): Update.
17656 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
17657 (get_estimated_loop_iterations): Update.
17658 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
17659 (single_likely_exit): Update.
17660 * cfgloopmanip.c (scale_loop_profile): Update.
17662 (set_zero_probability): Update.
17663 (lv_adjust_loop_entry_edge): Update.
17664 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17665 (purge_dead_edges): Update.
17666 (rtl_account_profile_record): Update.
17667 * cgraph.c (cgraph_node::create): Uninitialize count.
17668 (symbol_table::create_edge): Uninitialize count.
17669 (cgraph_update_edges_for_call_stmt_node): Update.
17670 (cgraph_edge::dump_edge_flags): Update.
17671 (cgraph_node::dump): Update.
17672 (cgraph_edge::maybe_hot_p): Update.
17673 * cgraph.h: Include profile-count.h
17674 (create_clone), create_edge, create_indirect_edge): Update.
17675 (cgraph_node): Turn count to profile_count.
17676 (cgraph_edge0: Likewise.
17677 (make_speculative, clone): Update.
17678 (create_edge): Update.
17679 (init_lowered_empty_function): Update.
17680 * cgraphclones.c (cgraph_edge::clone): Update.
17681 (duplicate_thunk_for_node): Update.
17682 (cgraph_node::create_clone): Update.
17683 * cgraphunit.c (cgraph_node::analyze): Update.
17684 (cgraph_node::expand_thunk): Update.
17685 * final.c (dump_basic_block_info): Update.
17686 * gimple-streamer-in.c (input_bb): Update.
17687 * gimple-streamer-out.c (output_bb): Update.
17688 * graphite.c (print_global_statistics): Update.
17689 (print_graphite_scop_statistics): Update.
17690 * hsa-brig.c: Include basic-block.h.
17691 * hsa-dump.c: Include basic-block.h.
17692 * hsa-gen.c (T sum_slice): Update.
17693 (convert_switch_statements):Update.
17694 * hsa-regalloc.c: Include basic-block.h.
17695 * ipa-chkp.c (chkp_produce_thunks): Update.
17696 * ipa-cp.c (struct caller_statistics): Update.
17697 (init_caller_stats): Update.
17698 (gather_caller_stats): Update.
17699 (ipcp_cloning_candidate_p): Update.
17700 (good_cloning_opportunity_p): Update.
17701 (get_info_about_necessary_edges): Update.
17702 (dump_profile_updates): Update.
17703 (update_profiling_info): Update.
17704 (update_specialized_profile): Update.
17705 (perhaps_add_new_callers): Update.
17706 (decide_about_value): Update.
17707 (ipa_cp_c_finalize): Update.
17708 * ipa-devirt.c (struct odr_type_warn_count): Update.
17709 (struct decl_warn_count): Update.
17710 (struct final_warning_record): Update.
17711 (possible_polymorphic_call_targets): Update.
17712 (ipa_devirt): Update.
17713 * ipa-fnsummary.c (redirect_to_unreachable): Update.
17714 * ipa-icf.c (sem_function::merge): Update.
17715 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
17716 * ipa-inline.c (compute_uninlined_call_time): Update.
17717 (compute_inlined_call_time): Update.
17718 (want_inline_small_function_p): Update.
17719 (want_inline_self_recursive_call_p): Update.
17720 (edge_badness): Update.
17721 (lookup_recursive_calls): Update.
17722 (recursive_inlining): Update.
17723 (inline_small_functions): Update.
17724 (dump_overall_stats): Update.
17725 (dump_inline_stats): Update.
17726 * ipa-profile.c (ipa_profile_generate_summary): Update.
17727 (ipa_propagate_frequency): Update.
17728 (ipa_profile): Update.
17729 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17730 * ipa-utils.c (ipa_merge_profiles): Update.
17731 * loop-doloop.c (doloop_modify): Update.
17732 * loop-unroll.c (report_unroll): Update.
17733 (unroll_loop_runtime_iterations): Update.
17734 * lto-cgraph.c (lto_output_edge): Update.
17735 (lto_output_node): Update.
17736 (input_node): Update.
17737 (input_edge): Update.
17738 (merge_profile_summaries): Update.
17739 * lto-streamer-in.c (input_cfg): Update.
17740 * lto-streamer-out.c (output_cfg): Update.
17741 * mcf.c (create_fixup_graph): Update.
17742 (adjust_cfg_counts): Update.
17743 (sum_edge_counts): Update.
17744 * modulo-sched.c (sms_schedule): Update.
17745 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
17746 * predict.c (maybe_hot_count_p): Update.
17747 (probably_never_executed): Update.
17748 (dump_prediction): Update.
17749 (combine_predictions_for_bb): Update.
17750 (propagate_freq): Update.
17751 (handle_missing_profiles): Update.
17752 (counts_to_freqs): Update.
17753 (rebuild_frequencies): Update.
17754 (force_edge_cold): Update.
17755 * predict.h: Include profile-count.h
17756 (maybe_hot_count_p, counts_to_freqs): UPdate.
17757 * print-rtl-function.c: Do not include cfg.h
17758 * print-rtl.c: Include basic-block.h
17759 * profile-count.c: New file.
17760 * profile-count.h: New file.
17761 * profile.c (is_edge_inconsistent): Update.
17762 (correct_negative_edge_counts): Update.
17763 (is_inconsistent): Update.
17764 (set_bb_counts): Update.
17765 (read_profile_edge_counts): Update.
17766 (compute_frequency_overlap): Update.
17767 (compute_branch_probabilities): Update; Initialize and deinitialize
17769 (branch_prob): Update.
17770 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
17771 (edge_gcov_count): New.
17772 (bb_gcov_count): New.
17773 * shrink-wrap.c (try_shrink_wrapping): Update.
17774 * tracer.c (better_p): Update.
17775 * trans-mem.c (expand_transaction): Update.
17776 (ipa_tm_insert_irr_call): Update.
17777 (ipa_tm_insert_gettmclone_call): Update.
17778 * tree-call-cdce.c: Update.
17779 * tree-cfg.c (gimple_duplicate_sese_region): Update.
17780 (gimple_duplicate_sese_tail): Update.
17781 (gimple_account_profile_record): Update.
17782 (execute_fixup_cfg): Update.
17783 * tree-inline.c (copy_bb): Update.
17784 (copy_edges_for_bb): Update.
17785 (initialize_cfun): Update.
17786 (freqs_to_counts): Update.
17787 (copy_cfg_body): Update.
17788 (expand_call_inline): Update.
17789 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17790 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
17791 (try_unroll_loop_completely): Update.
17792 (try_peel_loop): Update.
17793 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17794 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
17795 * tree-ssa-loop-split.c (connect_loops): Update.
17796 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
17797 * tree-ssa-reassoc.c (branch_fixup): Update.
17798 * tree-ssa-tail-merge.c (replace_block_by): Update.
17799 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
17800 (compute_path_counts): Update.
17801 (update_profile): Update.
17802 (recompute_probabilities): Update.
17803 (update_joiner_offpath_counts): Update.
17804 (estimated_freqs_path): Update.
17805 (freqs_to_counts_path): Update.
17806 (clear_counts_path): Update.
17807 (ssa_fix_duplicate_block_edges): Update.
17808 (duplicate_thread_path): Update.
17809 * tree-switch-conversion.c (case_bit_test_cmp): Update.
17810 (struct switch_conv_info): Update.
17811 * tree-tailcall.c (decrease_profile): Update.
17812 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17813 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17814 * value-prof.c (check_counter): Update.
17815 (gimple_divmod_fixed_value): Update.
17816 (gimple_mod_pow2): Update.
17817 (gimple_mod_subtract): Update.
17818 (gimple_ic_transform): Update.
17819 (gimple_stringop_fixed_value): Update.
17820 * value-prof.h (gimple_ic): Update.
17822 2017-06-02 Carl Love <cel@us.ibm.com>
17824 * config/rs6000/rs6000-c: Add support for built-in functions
17825 vector double vec_doublee (vector signed int);
17826 vector double vec_doublee (vector unsigned int);
17827 vector double vec_doublee (vector float);
17828 vector double vec_doubleh (vector signed int);
17829 vector double vec_doubleh (vector unsigned int);
17830 vector double vec_doubleh (vector float);
17831 vector double vec_doublel (vector signed int);
17832 vector double vec_doublel (vector unsigned int);
17833 vector double vec_doublel (vector float);
17834 vector double vec_doubleo (vector signed int);
17835 vector double vec_doubleo (vector unsigned int);
17836 vector double vec_doubleo (vector float);.
17837 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
17838 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
17840 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
17841 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
17842 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
17844 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
17845 vec_doublel, vec_doubleh.
17846 * doc/extend.texi: Update the built-in documentation file for the
17847 new built-in functions.
17849 2017-06-02 David Malcolm <dmalcolm@redhat.com>
17852 * ipa-inline-analysis.c (free_growth_caches): Set
17853 edge_removal_hook_holder to NULL after removing it.
17855 2017-06-02 Sudakshina Das <sudi.das@arm.com>
17857 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
17858 comparision with zero.
17860 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17861 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17862 for early expansion of vec_min and vec_max builtins.
17863 (builtin_function_type): Add min/max unsigned variants to those
17864 identified as having unsigned arguments.
17866 2017-06-02 Olivier Hainque <hainque@adacore.com>
17868 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
17870 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17872 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
17873 Use VALL_F16 iterator rather than VALL.
17875 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17877 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
17878 Emit CBNZ inside loop when doing a strong exchange and comparing
17879 against zero. Generate the CC flags after the loop.
17881 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
17883 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
17884 (dl_section_ref): New.
17885 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
17886 On AIX, append an expression to subtract the size of the
17887 section length to dl_section_ref.
17889 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17891 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17892 for early expansion of vector absolute builtins.
17894 2017-06-02 Richard Biener <rguenther@suse.de>
17896 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
17897 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
17899 2017-06-02 Richard Biener <rguenther@suse.de>
17901 PR tree-optimization/80948
17902 * tree-tailcall.c (find_tail_calls): Track stmts to move in
17903 stmt order as well.
17905 2017-06-02 Richard Biener <rguenther@suse.de>
17907 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
17909 * tree-vect-stmts.c (process_use): Do not mark backedge defs
17910 for inductions as relevant.
17912 2017-06-02 Richard Biener <rguenther@suse.de>
17914 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
17915 (vectorizable_induction): ... this. Remove dead code.
17917 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
17919 * builtins. (expand_builtin_alloca): Remove second parameter and
17920 infer its value from the first parameter instead.
17921 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
17923 2017-06-02 Jakub Jelinek <jakub@redhat.com>
17925 PR rtl-optimization/80903
17926 * loop-doloop.c (add_test): Unshare sequence.
17928 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
17930 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
17932 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
17934 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
17936 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
17937 xlogue_layout::get_instance, logue_layout::xlogue_layout,
17938 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
17939 (xlogue_layout::get_stub_rtx): Make static.
17940 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
17941 (xlogue_layout::compute_stub_managed_regs): Rename to...
17942 (xlogue_layout::count_stub_managed_regs): ...this.
17943 (xlogue_layout::is_stub_managed_reg): New function.
17944 (xlogue_layout::m_stub_names): Rename to...
17945 (xlogue_layout::s_stub_names): ...this, make static.
17946 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
17947 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
17948 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
17949 xlogue_layout::s_stub_names): Instantiate statics.
17950 (stub_managed_regs): Remove.
17951 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
17952 (disable_call_ms2sysv_xlogues): Rename to...
17953 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
17954 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
17956 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
17957 change after reload_completed.
17958 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
17960 (ix86_expand_prologue): Likewise.
17961 (ix86_expand_epilogue): Likewise.
17962 (ix86_expand_split_stack_prologue): Likewise.
17963 (ix86_compute_frame_layout): Remove frame parameter ...
17964 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
17965 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
17967 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
17968 (ix86_frame): Move from here ...
17969 * config/i386/i386.h (ix86_frame): ... to here.
17970 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
17971 complete ix86_frame data structure instead. Remove some_ld_name.
17973 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17975 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
17976 symbols that hold a DECL_VALUE_EXPR.
17978 2017-06-01 Martin Jambor <mjambor@suse.cz>
17980 PR tree-optimization/80898
17981 * tree-sra.c (process_subtree_disqualification): Removed.
17982 (disqualify_candidate): Do not acll
17983 process_subtree_disqualification.
17984 (subtree_mark_written_and_enqueue): New function.
17985 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
17986 RHS has been disqualified and re-queue LHS if necessary. Apart
17987 from that, ignore disqualified RHS.
17989 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17991 * config/s390/s390.c (s390_emit_epilogue): Disable early return
17992 address fetch for z10 or later.
17994 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17996 * config/arc/arc.md (tst_movb): Add guard when splitting.
17998 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18000 * config/arc/arc.c (arc_can_eliminate): Test against
18001 arc_frame_pointer_needed.
18003 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18005 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
18006 to prevent store reordering.
18007 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
18008 (type): Add block type.
18009 (stack_tie): Define special instruction to be used in
18012 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18014 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
18015 constraint. It is not valid for the pattern.
18016 (noncommutative_binary_comparison): Likewise.
18018 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18020 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
18023 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18025 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
18026 be used by the reg-alloc.
18028 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18030 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
18031 reg-alloc when having mul64 or mul32x16 instructions.
18032 (mulsidi3): Likewise.
18033 (umulsidi3): Likewise.
18034 (mulsi32x16): New pattern.
18035 (mulsi64): Likewise.
18036 (mulsidi64): Likewise.
18037 (umulsidi64): Likewise.
18038 (MUL32x16_REG): Define.
18039 (mul64_600): Use MUL32x16_REG.
18040 (mac64_600): Likewise.
18041 (umul64_600): Likewise.
18042 (umac64_600): Likewise.
18044 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18046 * config/arc/arc.md (mulsi3_700): Make it commutative.
18048 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
18050 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
18052 (*sign_extendsidi2_insn): Likewise for movstosw.
18054 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18056 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
18057 the type of the input discriminant value. Convert the
18058 discriminant value of signedness vary.
18060 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
18062 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
18063 Add to -Wall section.
18065 2017-06-01 Richard Biener <rguenther@suse.de>
18067 PR middle-end/66313
18068 * fold-const.c (fold_plusminus_mult_expr): If the factored
18069 factor may be zero use a wrapping type for the inner operation.
18070 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
18071 and handle moved defs.
18072 (process_assignment): Properly guard the unary op case. Return a
18073 tri-state indicating that moving the stmt before the call may allow
18074 to continue. Pass through to_move.
18075 (find_tail_calls): Handle moving unrelated defs before
18078 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
18081 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
18082 splitter result in the canonical way.
18084 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
18086 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
18087 also for 32bit target. Update insn attributes.
18088 (zero-extendsidi2 splitter): Allow all registers for operand 1.
18090 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
18092 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
18093 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
18094 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
18095 (_mm_maskz_min_ss): New intrinsics.
18097 2017-05-31 Martin Liska <mliska@suse.cz>
18099 * tree-vect-loop.c (vect_create_epilog_for_reduction):
18100 Change comment style to one we normally use.
18101 (vectorizable_reduction): Likewise.
18102 (vectorizable_induction): Likewise.
18103 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18104 (vectorizable_call): Likewise.
18105 (vectorizable_simd_clone_call): Likewise.
18106 (vectorizable_conversion): Likewise.
18107 (vectorizable_assignment): Likewise.
18108 (vectorizable_shift): Likewise.
18109 (vectorizable_operation): Likewise.
18110 (vectorizable_store): Likewise.
18111 (vectorizable_load): Likewise.
18112 * tree-vectorizer.h: Likewise.
18114 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
18116 * passes.c (emergency_dump_function): New.
18117 * tree-pass.h (emergency_dump_function): Declare.
18118 * plugin.c (plugins_internal_error_function): Remove.
18119 * plugin.h (plugins_internal_error_function): Remove declaration.
18120 * toplev.c (internal_error_function): New static function. Use it...
18121 (general_init): ...here.
18123 2017-05-31 Graham Markall <graham.markall@embecosm.com>
18125 * config/arc/arc.c (arc_print_operand): Handle constant operands.
18126 (arc_rtx_costs): Add costs for new patterns.
18127 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
18128 * config/arc/predicates.md: Add _1_2_3_operand predicate.
18130 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18132 * tree-ssa-strlen.c (get_next_strinfo): New function.
18133 (get_stridx_plus_constant): Use it.
18134 (zero_length_string): Likewise.
18135 (adjust_related_strinfos): Likewise.
18136 (adjust_last_stmt): Likewise.
18138 2017-05-31 Richard Biener <rguenther@suse.de>
18141 * config/i386/i386.c (ix86_expand_builtin): Remove assert
18142 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
18144 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18146 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
18147 loop_vinfo argument and use of dependence distance vectors.
18148 Check instead whether the two references differ only in their
18149 initial value and assume that they have the same alignment if the
18150 difference is a multiple of the vector alignment.
18151 (vect_analyze_data_refs_alignment): Update call accordingly.
18153 2017-05-31 Martin Liska <mliska@suse.cz>
18156 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
18158 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18160 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
18161 (create_intersect_range_checks): Move from ...
18162 * tree-data-ref.c (create_intersect_range_checks_index)
18163 (create_intersect_range_checks): ... to here.
18164 (create_runtime_alias_checks): New function factored from ...
18165 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
18166 here. Call above function.
18167 * tree-data-ref.h (create_runtime_alias_checks): New function.
18169 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18171 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
18172 segment length for dr_b and compute it in wide_int.
18174 2017-05-31 Richard Biener <rguenther@suse.de>
18176 PR tree-optimization/80906
18177 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
18178 and pass through iv_map.
18179 (copy_bb_and_scalar_dependences): Adjust.
18180 (translate_pending_phi_nodes): Likewise.
18181 (copy_loop_close_phi_args): Handle code-generating IVs instead
18184 2017-05-30 David Malcolm <dmalcolm@redhat.com>
18186 * diagnostic-color.c (color_dict): Add "type-diff".
18187 (parse_gcc_colors): Update comment.
18188 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
18189 -fdiagnostics-show-template-tree and -fno-elide-type.
18190 (GCC_COLORS): Add type-diff to example.
18192 (-fdiagnostics-show-template-tree): New.
18193 (-fno-elide-type): New.
18194 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
18195 the pp_format_decoder callback. Call any m_format_postprocessor's
18197 (pretty_printer::pretty_printer): Initialize
18198 m_format_postprocessor.
18199 (pretty_printer::~pretty_printer): Delete any
18200 m_format_postprocessor.
18201 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
18202 (class format_postprocessor): New class.
18203 (struct pretty_printer::format_decoder): Document the new parameters.
18204 (struct pretty_printer::m_format_postprocessor): New field.
18205 * tree-diagnostic.c (default_tree_printer): Update for new
18206 bool and const char ** params.
18207 * tree-diagnostic.h (default_tree_printer): Likewise.
18209 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
18211 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
18212 (lwa_operand): Delete rs6000_gen_cell_microcode test.
18213 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18214 rs6000_gen_cell_microcode code.
18215 (rs6000_final_prescan_insn): Delete.
18216 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
18217 "warn-cell-microcode" entries.
18218 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
18219 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
18220 throughout. Change cc_reg_not_micro_cr0_operand to
18221 cc_reg_not_cr0_operand throughout.
18222 (*extendhi<mode>2_noload): Delete.
18223 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
18224 (mwarn-cell-microcode): Delete.
18225 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
18226 -mgen-cell-microcode and -mwarn-cell-microcode.
18228 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
18231 * config/i386/constraints.md (Yd): New constraint.
18233 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
18234 and (?Yd, r) alternatives. Update insn attributes.
18235 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
18236 and (?*Yd, r) alternatives. Update insn attributes.
18237 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
18239 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
18241 * gimplify.c (gimplify_modify_expr): Don't create a
18242 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
18245 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
18247 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
18249 2017-05-30 Richard Biener <rguenther@suse.de>
18251 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
18252 and reduc_def fields.
18253 (STMT_VINFO_REDUC_TYPE): New define.
18254 (STMT_VINFO_REDUC_DEF): Likewise.
18255 (vect_force_simple_reduction): Adjust prototype.
18256 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
18257 (vect_is_simple_reduction): Remove check_reduction argument.
18258 (vect_force_simple_reduction): Adjust and set
18259 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18260 (vectorizable_reduction): Do not re-do reduction analysis
18261 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18262 * tree-parloops.c (gather_scalar_reductions): Adjust.
18264 2017-05-30 Richard Biener <rguenther@suse.de>
18266 PR middle-end/80901
18267 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
18270 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
18272 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
18273 Introduce unknown_misalignment parameter and remove vf.
18274 (vect_peeling_hash_get_lowest_cost):
18275 Pass unknown_misalignment parameter.
18276 (vect_enhance_data_refs_alignment):
18277 Fix unsupportable data ref treatment.
18279 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18281 * tree-vect-data-refs.c (vect_get_data_access_cost):
18282 Workaround for SLP handling.
18283 (vect_enhance_data_refs_alignment):
18284 Compute costs for doing no peeling at all, compare to the best
18285 peeling costs so far and avoid peeling if cheaper.
18287 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18289 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
18290 Return peeling info and set costs to zero for unlimited cost
18292 (vect_enhance_data_refs_alignment): Also inspect all datarefs
18293 with unknown misalignment. Compute and costs for unknown
18294 misalignment, compare them to the costs for known misalignment
18295 and choose the cheapest for peeling.
18297 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18299 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
18300 (vect_get_peeling_costs_all_drs): Create function.
18301 (vect_peeling_hash_get_lowest_cost):
18302 Use vect_get_peeling_costs_all_drs.
18303 (vect_peeling_supportable): Create function.
18304 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
18306 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18308 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
18309 DR_HAS_NEGATIVE_STEP.
18310 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
18311 (vect_enhance_data_refs_alignment): Use.
18312 (vect_duplicate_ssa_name_ptr_info): Use.
18313 * tree-vectorizer.h (dr_misalignment): Use.
18314 (known_alignment_for_access_p): Use.
18316 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18319 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
18321 (has_section_name): New function.
18323 2017-05-30 Martin Liska <mliska@suse.cz>
18326 * auto-profile.c (get_function_decl_from_block): Fix
18329 2017-05-30 Richard Biener <rguenther@suse.de>
18331 PR middle-end/80876
18332 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
18334 2017-05-30 Martin Liska <mliska@suse.cz>
18336 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
18337 * dumpfile.h (struct dump_file_info): Remove ctors.
18339 2017-05-30 Martin Liska <mliska@suse.cz>
18341 * predict.def: Fix GNU coding style.
18343 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18345 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18346 Mark 'to' argument with ATTRIBUTE_UNUSED.
18348 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18350 * config/xtensa/xtensa.c (xtensa_emit_call): Use
18351 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
18352 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
18355 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
18357 * doc/install.texi (Options specification): Restore entry of
18358 --enable-sjlj-exceptions.
18360 2017-05-27 Michael Eager <eager@eagercon.com>
18363 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
18365 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
18367 * config/microblaze/microblaze.h
18368 (FIXED_REGISTERS): Update in macro.
18369 (CALL_USED_REGISTERS): Update in macro.
18371 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
18373 * doc/install.texi: Add links to macOS binary distributions.
18375 2017-05-27 Jakub Jelinek <jakub@redhat.com>
18379 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18381 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18383 2017-05-26 Martin Liska <mliska@suse.cz>
18385 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
18387 2017-05-26 Martin Liska <mliska@suse.cz>
18389 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
18390 always leading ';; '.
18391 (dump_bb_info): Likewise.
18392 (brief_dump_cfg): Likewise.
18393 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
18394 * dumpfile.c: Remove usage of TDF_VERBOSE.
18395 * dumpfile.h (enum dump_kind): Likewise.
18396 (dump_gimple_bb_header): Do not use TDF_COMMENT.
18397 * print-tree.c (debug_verbose): Remove.
18398 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
18399 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
18400 * tree-diagnostic.c (default_tree_printer): Replace
18401 TDF_DIAGNOSTIC with TDF_SLIM.
18403 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18405 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
18406 in parameter loop, rather than loop_vinfo.
18407 (create_intersect_range_checks): Ditto.
18408 (vect_create_cond_for_alias_checks): Update call to above functions.
18410 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18412 PR tree-optimization/80815
18413 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
18414 for merging runtime alias checks. Handle negative DR_STEPs.
18416 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18418 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
18420 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
18421 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
18422 out code pruning runtime alias checks.
18423 * tree-data-ref.c (prune_runtime_alias_test_list): New function
18424 factored out from above.
18425 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18427 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18429 (prune_runtime_alias_test_list): New decalaration.
18431 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18433 * tree-vect-data-refs.c (compare_tree): Rename and move ...
18434 * tree-data-ref.c (data_ref_compare_tree): ... to here.
18435 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
18436 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
18437 (operator==, comp_dr_with_seg_len_pair): Ditto.
18438 (vect_prune_runtime_alias_test_list): Ditto.
18440 2017-05-26 Martin Liska <mliska@suse.cz>
18443 * params.def: Bound partial-inlining-entry-probability param.
18445 2017-05-26 Marek Polacek <polacek@redhat.com>
18448 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
18451 2017-05-26 Richard Biener <rguenther@suse.de>
18453 PR tree-optimization/80842
18454 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
18457 2017-05-26 Richard Biener <rguenther@suse.de>
18459 PR tree-optimization/80844
18460 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
18462 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
18464 * doc/md.texi (Machine Constraints): Update x86 family
18465 machine constraints section to match 'config/i386/constraints.md'.
18467 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
18469 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
18471 2017-05-25 Nathan Sidwell <nathan@acm.org>
18473 * doc/invoke.texi (--enable-languages): Update documentation.
18475 2017-05-25 Martin Liska <mliska@suse.cz>
18477 * dumpfile.c: Add TDF_FOLDING.
18478 * dumpfile.h (enum dump_kind): Likewise.
18479 * genmatch.c (dt_simplify::gen_1): Use it.
18481 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18483 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
18485 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18487 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18488 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
18490 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18492 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
18493 * match.pd (X == C): Rewrite it here.
18494 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
18495 with_certain_nonzero_bits2): New predicates.
18496 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
18498 2017-05-24 Nathan Sidwell <nathan@acm.org>
18500 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
18503 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
18506 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18508 * config/powerpcspe: New port. Files are copied from the rs6000
18509 port, with "rs6000" in filenames replaced by "powerpcspe".
18511 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
18513 PR rtl-optimization/80754
18514 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
18516 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
18518 * config/sparc/sparc.md (length): Return the correct value for -mflat
18519 sibcalls to match output_sibcall.
18521 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18525 * config/rs6000/rs6000.c (struct machine_function): Add new field
18527 (rs6000_get_separate_components): Init that field, use it.
18528 (rs6000_components_for_bb): Use the field.
18530 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
18532 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
18534 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
18536 PR middle-end/80823
18537 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
18539 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18542 * config/s390/s390.c (s390_check_qrst_address): Check incoming
18543 address against address_operand predicate.
18544 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
18546 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
18548 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
18549 parameters passed indirectly.
18551 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
18553 * config/i386/i386.md (*movdi_internal): Remove SSE4
18554 alternative 18 (?r, *v). Update insn attributes.
18555 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
18556 Update insn attributes.
18557 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
18558 Update insn attributes.
18559 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
18560 alternative 1 (r, v). Remove isa attribute.
18561 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
18562 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
18563 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18565 2017-05-23 Tom de Vries <tom@codesourcery.com>
18567 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
18570 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18572 * cgraphunit.c (symbol_table::process_new_functions): Update.
18573 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
18574 (inline_generate_summary): Rename to ...
18575 (ipa_fn_summary_generate): ... this one.
18576 (inline_read_summary): Rename to ...
18577 (ipa_fn_summary_read): ... this one.
18578 (inline_write_summary): Rename to ...
18579 (ipa_fn_summary_write): ... this one.
18580 (inline_free_summary): Rename to ...
18581 (ipa_free_fn_summary): ... this one.
18582 (pass_data_local_fn_summary, pass_local_fn_summary,
18583 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
18584 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
18585 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
18586 make_pass_ipa_fn_summary): New.
18587 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
18588 inline_write_summary, inline_free_summary): Remove.
18589 (ipa_free_fn_summary) : New.
18590 * ipa-inline.c (ipa_inline): Update.
18591 (pass_ipa_inline): Do not generate summaries.
18592 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
18594 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
18595 and add pass_ipa_fn_summary.
18596 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
18598 (make_pass_inline_parameters): Remove.
18600 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
18602 * omp-low.c (struct omp_context): Remove "default_kind" member.
18605 * omp-offload.c (execute_oacc_device_lower): Remove the
18606 parallelism dimensions function attributes for unparallelized
18607 OpenACC kernels constructs.
18609 2017-05-23 Martin Liska <mliska@suse.cz>
18611 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
18613 (cgraph_edge::make_speculative): Likewise.
18614 (cgraph_edge::resolve_speculation): Likewise.
18615 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
18616 (cgraph_node::dump): Likewise.
18617 * cgraph.h: Likewise.
18618 * cgraphunit.c (analyze_functions): Likewise.
18619 (symbol_table::compile): Likewise.
18620 * ipa-cp.c (print_all_lattices): Likewise.
18621 (determine_versionability): Likewise.
18622 (initialize_node_lattices): Likewise.
18623 (ipcp_verify_propagated_values): Likewise.
18624 (estimate_local_effects): Likewise.
18625 (update_profiling_info): Likewise.
18626 (create_specialized_node): Likewise.
18627 (perhaps_add_new_callers): Likewise.
18628 (decide_about_value): Likewise.
18629 (decide_whether_version_node): Likewise.
18630 (identify_dead_nodes): Likewise.
18631 (ipcp_store_bits_results): Likewise.
18632 * ipa-devirt.c (dump_targets): Likewise.
18633 (ipa_devirt): Likewise.
18634 * ipa-icf.c (sem_item::dump): Likewise.
18635 (sem_function::equals): Likewise.
18636 (sem_variable::equals): Likewise.
18637 (sem_item_optimizer::read_section): Likewise.
18638 (sem_item_optimizer::execute): Likewise.
18639 (congruence_class::dump): Likewise.
18640 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
18641 (dump_inline_summary): Likewise.
18642 (estimate_node_size_and_time): Likewise.
18643 (inline_analyze_function): Likewise.
18644 * ipa-inline-transform.c (inline_call): Likewise.
18645 * ipa-inline.c (report_inline_failed_reason): Likewise.
18646 (want_early_inline_function_p): Likewise.
18647 (edge_badness): Likewise.
18648 (update_edge_key): Likewise.
18649 (inline_small_functions): Likewise.
18650 * ipa-profile.c (ipa_profile): Likewise.
18651 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
18652 (ipa_make_edge_direct_to_target): Likewise.
18653 (remove_described_reference): Likewise.
18654 (ipa_impossible_devirt_target): Likewise.
18655 (propagate_controlled_uses): Likewise.
18656 (ipa_print_node_params): Likewise.
18657 (ipcp_transform_function): Likewise.
18658 * ipa-pure-const.c (pure_const_read_summary): Likewise.
18659 (propagate_pure_const): Likewise.
18660 * ipa-reference.c (generate_summary): Likewise.
18661 (read_write_all_from_decl): Likewise.
18662 (propagate): Likewise.
18663 (ipa_reference_read_optimization_summary): Likewise.
18664 * ipa-utils.c (ipa_merge_profiles): Likewise.
18665 * ipa.c (walk_polymorphic_call_targets): Likewise.
18666 (symbol_table::remove_unreachable_nodes): Likewise.
18667 (ipa_single_use): Likewise.
18668 * passes.c (execute_todo): Likewise.
18669 * predict.c (drop_profile): Likewise.
18670 * symtab.c (symtab_node::get_dump_name): New function.
18671 (symtab_node::dump_name): Likewise.
18672 (symtab_node::dump_asm_name): Likewise.
18673 (symtab_node::dump_references): Likewise.
18674 (symtab_node::dump_referring): Likewise.
18675 (symtab_node::dump_base): Likewise.
18676 (symtab_node::debug_symtab): Likewise.
18677 * tree-sra.c (convert_callers_for_node): Likewise.
18678 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
18679 * value-prof.c (init_node_map): Likewise.
18681 2017-05-23 Martin Liska <mliska@suse.cz>
18683 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
18684 and symtab_node::debug_symtab to symbol_table::debug.
18685 * cgraphunit.c (analyze_functions): Use the renamed function.
18686 (symbol_table::compile): Likewise.
18687 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
18688 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
18689 * passes.c (execute_todo): Likewise.
18690 * symtab.c (symbol_table::dump): New function.
18691 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
18693 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18695 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
18696 that nonconst implies exec.
18698 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18700 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18701 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18702 (inline_edge_summary_vec): Turn into ...
18703 (ipa_call_summaries): ... this one.
18704 (redirect_to_unreachable, edge_set_predicate,
18705 evaluate_properties_for_edge, inline_summary_alloc,
18706 reset_ipa_call_summary, reset_inline_summary,
18707 inline_summary_t::duplicate): Update.
18708 (inline_edge_duplication_hook): Turn to ...
18709 (ipa_call_summary_t::duplicate): ... this one.
18710 (inline_edge_removal_hook): Turn to ...
18711 (ipa_call_summary_t::remove): ... this one.
18712 (dump_inline_edge_summary): Turn to ...
18713 (dump_ipa_call_summary): ... this one.
18714 (estimate_function_body_sizes): Update.
18715 (inline_update_callee_summaries): Update.
18716 (remap_edge_change_prob): Update.
18717 (remap_edge_summaries): Update.
18718 (inline_merge_summary): Update.
18719 (do_estimate_edge_time): Update.
18720 (inline_generate_summary): Update.
18721 (inline_read_section): Update.
18722 (inline_read_summary): Update.
18723 (inline_free_summary): Update.
18724 * ipa-inline.c (can_inline_edge_p): Update.
18725 (compute_inlined_call_time): Update.
18726 (want_inline_small_function_p): Update.
18727 (edge_badness): Update.
18728 (early_inliner): Update.
18729 * ipa-inline.h (inline_edge_summary): Turn to ...
18730 (ipa_call_summary): ... this one.
18731 (ipa_call_summary_t): New class.
18732 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18733 (ipa_call_summaries): New.
18734 (inline_edge_summary): Remove.
18735 (estimate_edge_growth): Update.
18736 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18737 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18738 * ipa-split.c (execute_split_functions): Update.
18739 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18741 2017-05-23 Tom de Vries <tom@codesourcery.com>
18743 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18744 attributes): Document rdrand effective target.
18746 2017-05-23 Tom de Vries <tom@codesourcery.com>
18748 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18749 attributes): Sort alphabetically.
18751 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
18753 * config/avr/genmultilib.awk: Use gsub instead of gensub.
18755 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
18758 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
18759 V2DF/V2DI splat into two separate patterns, one that handles
18760 registers, and the other that only handles memory. Drop support
18761 for splatting from a GPR on ISA 2.07 and then splitting the
18762 splat into direct move and splat.
18763 (vsx_splat_<mode>_reg): Likewise.
18764 (vsx_splat_<mode>_mem): Likewise.
18766 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
18768 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
18770 2017-05-22 Jakub Jelinek <jakub@redhat.com>
18772 PR middle-end/80809
18773 * omp-low.c (finish_taskreg_remap): New function.
18774 (finish_taskreg_scan): If unit size of ctx->record_type
18775 is non-constant, unshare the size expression and replace
18776 decls in it with possible outer var refs.
18778 PR middle-end/80809
18779 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
18780 GOVD_SHARED rather than GOVD_PRIVATE with it.
18781 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
18782 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
18784 PR middle-end/80853
18785 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
18786 as last argument to build_outer_var_ref for pointer bases of array
18787 section reductions.
18789 2017-05-19 Martin Sebor <msebor@redhat.com>
18791 * print-tree.c (print_node): Print DECL_READ_P flag.
18793 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18795 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
18796 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
18797 * cgraph.c: Likewise.
18798 * cgraphunit.c: Likewise.
18799 * gengtype.c: Likewise.
18800 * ipa-cp.c: Likewise.
18801 * ipa-devirt.c: Likewise.
18802 * ipa-icf.c: Likewise.
18803 * ipa-predicate.c: Likewise.
18804 * ipa-profile.c: Likewise.
18805 * ipa-prop.c: Likewise.
18806 * ipa-split.c: Likewise.
18808 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
18809 edge_predicate_pool, dump_inline_hints,
18810 inline_summary::account_size_time, redirect_to_unreachable,
18811 edge_set_predicate, set_hint_predicate,
18812 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
18813 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
18814 inline_summary_t::remove, remap_hint_predicate_after_duplication,
18815 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
18816 ipa_call_summary_t::remove, initialize_growth_caches,
18817 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
18818 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
18819 mark_modified, unmodified_parm_1, unmodified_parm,
18820 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
18821 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
18822 compute_bb_predicates, will_be_nonconstant_expr_predicate,
18823 will_be_nonconstant_predicate, record_modified_bb_info,
18824 get_minimal_bb, record_modified, param_change_prob,
18825 phi_result_unknown_predicate, predicate_for_phi_result,
18826 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
18827 estimate_function_body_sizes, compute_inline_parameters,
18828 compute_inline_parameters_for_curren, pass_data_inline_parameters,
18829 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
18830 inline_update_callee_summaries, remap_edge_change_prob,
18831 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
18832 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
18833 inline_analyze_function, inline_summary_t::insert,
18834 inline_generate_summary, read_ipa_call_summary, inline_read_section,
18835 inline_read_summary, write_ipa_call_summary, inline_write_summary,
18836 inline_free_summary): Move to ipa-fnsummary.h
18837 (predicate_t): Remove.
18838 * ipa-fnsummary.c: New file.
18839 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
18840 (enum inline_hints_vals, inline_hints, agg_position_info,
18841 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
18842 inline_summaries, ipa_call_summary, ipa_call_summary_t,
18843 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
18844 dump_inline_summary, dump_inline_hints, inline_generate_summary,
18845 inline_read_summary, inline_write_summary, inline_free_summary,
18846 inline_analyze_function, initialize_inline_failed,
18847 inline_merge_summary, inline_update_overall_summary,
18848 compute_inline_parameters): Move to ipa-fnsummary.h
18849 * ipa-fnsummary.h: New file.
18850 * ipa-inline-transform.h: Include ipa-inline.h.
18851 * ipa-inline.c: LIkewise.
18853 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18855 * ipa-inline.c (edge_badness): Use inlined_time instead of
18856 inline_summaries->get.
18858 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18860 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
18862 2017-05-22 Nathan Sidwell <nathan@acm.org>
18864 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
18865 (fdump-lang): Document 'raw' option.
18866 * dumpfile.h (TDI_tu): Delete.
18867 * dumpfile.c (dump_files): Remove translation-unit.
18868 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
18870 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
18872 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
18873 command option from $(AWK) call.
18874 * config/avr/genmultilib.awk: Simplify and rewrite so that it
18875 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
18876 [FORMAT]: Remove handling of variable.
18877 * config/avr/t-multilib: Regenerate.
18879 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18881 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
18883 (dump_inline_summary): Do not print self_time.
18884 (estimate_function_body_sizes): Do not set self_time.
18885 (compute_inline_parameters): Likewise.
18886 (inline_read_section, inline_write_summary): Do not stream self_time.
18887 * ipa-inline.h (inline_summary): Drop self_time.
18889 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18891 * ipa-inline-analysis.c (account_size_time): Rename to ...
18892 (inline_summary::account_size_time): ... this one.
18893 (reset_ipa_call_summary): Turn to ...
18894 (ipa_call_summary::reset): ... this one.
18895 (reset_inline_summary): Turn to ...
18896 (inline_summary::reset): ... this one.
18897 (inline_summary_t::remove): Update.
18898 (inline_summary_t::duplicate): Update.
18899 (ipa_call_summary_t::remove): Update.
18900 (dump_inline_summary): Update.
18901 (estimate_function_body_sizes): Update.
18902 (compute_inline_parameters): Update.
18903 (estimate_node_size_and_time): Update.
18904 (inline_merge_summary): Update.
18905 (inline_update_overall_summary): Update.
18906 (inline_read_section): Update.
18907 (inline_write_summary): Update.
18908 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
18909 add account_size_time and reset member functions.
18910 (ipa_call_summary): Add reset function.
18911 * ipa-predicate.h (predicate::operator &): Constify.
18913 2017-05-22 Richard Biener <rguenther@suse.de>
18915 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
18917 2017-05-19 Jason Merrill <jason@redhat.com>
18919 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
18921 2017-05-19 Marek Polacek <polacek@redhat.com>
18924 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
18925 TYPE_OVERFLOW_WRAPS checks.
18927 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
18929 * tree-core.h (enum omp_clause_default_kind): Add
18930 "OMP_CLAUSE_DEFAULT_PRESENT".
18931 * tree-pretty-print.c (dump_omp_clause): Handle it.
18932 * gimplify.c (enum gimplify_omp_var_data): Add
18933 "GOVD_MAP_FORCE_PRESENT".
18934 (gimplify_adjust_omp_clauses_1): Map it to
18935 "GOMP_MAP_FORCE_PRESENT".
18936 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
18938 * gimplify.c (oacc_default_clause): Clarify.
18940 2017-05-19 Nathan Sidwell <nathan@acm.org>
18942 LANG_HOOK_REGISTER_DUMPS
18943 * toplev.c (general_init): Call register dump lang hook.
18944 * doc/invoke.texi: Document -fdump-lang option family.
18945 * dumpfile.c (dump_files): Remove class dump here.
18946 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
18947 * dumpfile.h (tree_dump_index): Remove TDI_class.
18948 * langhooks-def.h (lhd_register_dumps): Declare.
18949 (LANG_HOOKS_REGISTER_DUMPS): Define.
18950 (LANG_HOOKS_INITIALIZER): Add it.
18951 * langhooks.c (lhd_register_dumps): Define.
18952 * langhooks.h (struct lang_hooks): Add register_dumps.
18954 2017-05-19 Nathan Sidwell <nathan@acm.org>
18956 * context.h (context::set_passes): New.
18957 * context.c (context::context): Do not create pass manager.
18958 * toplev.c (general_init): Create pass manager here.
18960 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
18962 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
18963 use this splitter if two add or or instructions would also work for
18964 the constant we want to generate.
18966 2017-05-19 Richard Biener <rguenther@suse.de>
18969 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
18970 predicate evaluation.
18972 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18974 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
18976 * ipa-inline.c (want_inline_small_function_p): Do not cast to
18979 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18981 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18982 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18983 (inline_edge_summary_vec): Turn into ...
18984 (ipa_call_summaries): ... this one.
18985 (redirect_to_unreachable, edge_set_predicate,
18986 evaluate_properties_for_edge, inline_summary_alloc,
18987 reset_ipa_call_summary, reset_inline_summary,
18988 inline_summary_t::duplicate): Update.
18989 (inline_edge_duplication_hook): Turn to ...
18990 (ipa_call_summary_t::duplicate): ... this one.
18991 (inline_edge_removal_hook): Turn to ...
18992 (ipa_call_summary_t::remove): ... this one.
18993 (dump_inline_edge_summary): Turn to ...
18994 (dump_ipa_call_summary): ... this one.
18995 (estimate_function_body_sizes): Update.
18996 (inline_update_callee_summaries): Update.
18997 (remap_edge_change_prob): Update.
18998 (remap_edge_summaries): Update.
18999 (inline_merge_summary): Update.
19000 (do_estimate_edge_time): Update.
19001 (inline_generate_summary): Update.
19002 (inline_read_section): Update.
19003 (inline_read_summary): Update.
19004 (inline_free_summary): Update.
19005 * ipa-inline.c (can_inline_edge_p): Update.
19006 (compute_inlined_call_time): Update.
19007 (want_inline_small_function_p): Update.
19008 (edge_badness): Update.
19009 (early_inliner): Update.
19010 * ipa-inline.h (inline_edge_summary): Turn to ...
19011 (ipa_call_summary): ... this one.
19012 (ipa_call_summary_t): New class.
19013 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19014 (ipa_call_summaries): New.
19015 (inline_edge_summary): Remove.
19016 (estimate_edge_growth): Update.
19017 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19018 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19019 * ipa-split.c (execute_split_functions): Update.
19020 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19022 2017-05-19 Richard Biener <rguenther@suse.de>
19024 PR middle-end/80764
19025 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
19027 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
19029 * config/rs6000/rs6000.c (struct machine_function): Add field
19030 fpr_is_wrapped_separately.
19031 (rs6000_get_separate_components): Use 64 components. Handle the
19032 new FPR components.
19033 (rs6000_components_for_bb): Handle the FPR components.
19034 (rs6000_emit_prologue_components): Handle the FPR components.
19035 (rs6000_emit_epilogue_components): Handle the FPR components.
19036 (rs6000_set_handled_components): Handle the FPR components.
19037 (rs6000_emit_prologue): Don't output prologue code for those FPRs
19038 that are already separately shrink-wrapped.
19039 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
19040 that are already separately shrink-wrapped.
19042 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19045 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
19048 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
19049 (define_peephole2 for Altivec d-form load): Add peepholes to catch
19050 cases where the register allocator uses a move and an offsettable
19051 memory operation to/from a FPR register on ISA 2.06/2.07.
19052 (define_peephole2 for Altivec d-form store): Likewise.
19054 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
19057 * config/i386/mmx.md (*mov<mode>_internal): Enable
19058 alternatives 11, 12, 13 and 14 also for 32bit targets.
19059 Remove alternatives 15, 16, 17 and 18.
19060 * config/i386/sse.md (vec_concatv2di): Change
19061 alternative (!x, *y) to (x, ?!*Yn).
19063 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
19065 * dumpfile.h (enum dump_kind): Remove stray comma.
19067 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19069 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
19070 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
19071 predicate::num_conditions
19072 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
19073 (CHANGED): turn into predicate::changed.
19074 (agg_position_info): Move to ipa-predicate.h
19075 (add_condition, predicate::add_clause, predicate::operator &=,
19076 predicate::or_with, predicate::evaluate, predicate::probability,
19077 dump_condition, dump_clause, predicate::dump,
19078 predicate::remap_after_duplication, predicate::remap_after_inlining,
19079 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
19080 (evaluate_conditions_for_known_args): Update.
19081 (set_cond_stmt_execution_predicate): Update.
19082 * ipa-inline.h: Include ipa-predicate.h
19083 (condition, inline_param_summary, conditions, agg_position_info,
19084 predicate): Move to ipa-predicate.h
19085 * ipa-predicate.c: New file.
19086 * ipa-predicate.h: New file.
19088 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
19090 * final.c (leaf_function_p): Check we are not in a sequence.
19092 2017-05-18 Martin Liska <mliska@suse.cz>
19094 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
19095 * dumpfile.c (dump_register): Use new enum dump_kind.
19096 (get_dump_file_name): Likewise.
19097 (dump_enable_all): Likewise.
19098 (dump_switch_p_1): Likewise.
19099 (enable_rtl_dump_file): Remove usage of TDF_RTL.
19100 * dumpfile.h (enum dump_kind): New enum type.
19101 (struct dump_file_info): Create constructor and
19102 format fields and comments.
19103 * passes.c (pass_manager::register_one_dump_file):
19105 * statistics.c (statistics_early_init): Likewise.
19106 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
19107 TDF_TREE with TDF_SLIM.
19108 (gather_memory_references_ref): Likewise.
19110 2017-05-18 Martin Liska <mliska@suse.cz>
19112 * vec.h (struct vnull): Use it.
19114 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19116 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
19117 (true_predicate, false_predicate, true_predicate_p,
19118 false_predicate_p): Remove.
19119 (single_cond_predicate, not_inlined_predicate): Turn to member function
19121 (add_condition): Update.
19122 (add_clause): Turn to...
19123 (predicate::add_clause): ... this one; update; allow passing NULL
19125 (and_predicates): Turn to ...
19126 (predicate::operator &=): ... this one.
19127 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
19128 (or_predicates): Turn to ...
19129 (predicate::or_with): ... this one.
19130 (evaluate_predicate): Turn to ...
19131 (predicate::evaluate): ... this one.
19132 (predicate_probability): Turn to ...
19133 (predicate::probability): ... this one.
19134 (dump_condition): Update.
19135 (dump_predicate): Turn to ...
19136 (predicate::dump): ... this one.
19137 (account_size_time): Update.
19138 (edge_set_predicate): Update.
19139 (set_hint_predicate): UPdate.
19140 (evaluate_conditions_for_known_args): Update.
19141 (evaluate_properties_for_edge): Update.
19142 (remap_predicate_after_duplication): Turn to...
19143 (predicate::remap_after_duplication): ... this one.
19144 (remap_hint_predicate_after_duplication): Update.
19145 (inline_summary_t::duplicate): UPdate.
19146 (dump_inline_edge_summary): Update.
19147 (dump_inline_summary): Update.
19148 (set_cond_stmt_execution_predicate): Update.
19149 (set_switch_stmt_execution_predicate): Update.
19150 (compute_bb_predicates): Update.
19151 (will_be_nonconstant_expr_predicate): Update.
19152 (will_be_nonconstant_predicate): Update.
19153 (phi_result_unknown_predicate): Update.
19154 (predicate_for_phi_result): Update.
19155 (array_index_predicate): Update.
19156 (estimate_function_body_sizes): Update.
19157 (estimate_node_size_and_time): Update.
19158 (estimate_ipcp_clone_size_and_time): Update.
19159 (remap_predicate): Rename to ...
19160 (predicate::remap_after_inlining): ... this one.
19161 (remap_hint_predicate): Update.
19162 (inline_merge_summary): Update.
19163 (inline_update_overall_summary): Update.
19164 (estimate_size_after_inlining): Update.
19165 (read_predicate): Rename to ...
19166 (predicate::stream_in): ... this one.
19167 (read_inline_edge_summary): Update.
19168 (write_predicate): Rename to ...
19169 (predicate::stream_out): ... this one.
19170 (write_inline_edge_summary): Update.
19171 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
19172 (clause_t): Turn to uint32_t
19173 (predicate): Turn to class; implement constructor and operators
19175 (size_time_entry): Update.
19176 (inline_summary): Update.
19177 (inline_edge_summary): Update.
19179 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
19181 * fold-const.c (fold_binary_loc): Move transformation...
19182 * match.pd (C - X CMP X): ... here.
19184 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
19186 * config/sparc/sparc.c (sparc_option_override): Set function
19187 alignment for -mcpu=niagara7 to 64 to match the I$ line.
19188 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
19190 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
19192 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
19194 2017-05-18 Marek Polacek <polacek@redhat.com>
19197 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
19198 (pass_ubsan::execute): Call gimple_assign_single_p instead of
19199 gimple_assign_load_p.
19201 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
19203 PR middle-end/80692
19204 * real.c (do_compare): Give decimal_do_compare preference over
19205 comparing just the signs.
19207 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
19209 * doc/md.texi (Canonicalization of Instructions): Describe the
19210 canonical form of instructions that inherently set a condition
19213 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
19215 PR middle-end/80775
19216 * tree-cfg.c: Move deletion of unreachable case statements to after
19217 the merging of consecutive case labels.
19219 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
19221 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
19222 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
19223 restoring of callee-saved registers.
19225 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
19227 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
19228 * config/visium/visium.c (single_set_and_flags): Likewise.
19229 * config/visium/visium.md (Substitutions): Likewise.
19231 2017-05-17 Martin Liska <mliska@suse.cz>
19233 * cfg.c: Introduce dump_flags_t type and
19234 use it instead of int type.
19236 * cfghooks.c: Likewise.
19237 * cfghooks.h (struct cfg_hooks): Likewise.
19238 * cfgrtl.c: Likewise.
19239 * cfgrtl.h: Likewise.
19240 * cgraph.c (cgraph_node::get_body): Likewise.
19241 * coretypes.h: Likewise.
19242 * domwalk.c: Likewise.
19243 * domwalk.h: Likewise.
19244 * dumpfile.c (struct dump_option_value_info): Likewise.
19245 (dump_enable_all): Likewise.
19246 (dump_switch_p_1): Likewise.
19247 (opt_info_switch_p): Likewise.
19248 * dumpfile.h (enum tree_dump_index): Likewise.
19249 (struct dump_file_info): Likewise.
19250 * genemit.c: Likewise.
19251 * generic-match-head.c: Likewise.
19252 * gengtype.c (open_base_files): Likewise.
19253 * gimple-pretty-print.c: Likewise.
19254 * gimple-pretty-print.h: Likewise.
19255 * graph.c (print_graph_cfg): Likewise.
19256 * graphite-scop-detection.c (dot_all_sese): Likewise.
19257 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
19258 * loop-unroll.c (report_unroll): Likewise.
19259 * passes.c (pass_manager::register_one_dump_file): Likewise.
19260 * print-tree.c: Likewise.
19261 * statistics.c: Likewise.
19262 * tree-cfg.c: Likewise.
19263 * tree-cfg.h: Likewise.
19264 * tree-dfa.c: Likewise.
19265 * tree-dfa.h: Likewise.
19266 * tree-dump.c (dump_function): Likewise.
19267 * tree-dump.h (struct dump_info): Likewise.
19268 * tree-pretty-print.c: Likewise.
19269 * tree-pretty-print.h: Likewise.
19270 * tree-ssa-live.c: Likewise.
19271 * tree-ssa-live.h: Likewise.
19272 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
19273 * tree-vect-loop.c: Likewise.
19274 * tree-vect-slp.c: Likewise.
19276 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
19277 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19279 PR tree-optimization/80457
19280 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
19281 of all arguments to a statement as scalar_to_vec operations.
19282 (vectorizable_call): Adjust call to vect_model_simple_cost for
19284 (vectorizable_conversion): Likewise.
19285 (vectorizable_assignment): Likewise.
19286 (vectorizable_shift): Likewise.
19287 (vectorizable_operation): Likewise.
19288 (vectorizable_comparison): Likewise.
19289 (vect_is_simple_cond): Record the def types for operands.
19290 (vectorizable_condition): Likewise, call vect_model_simple_cost.
19291 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
19292 for statement argument count.
19294 2017-05-16 Carl Love <cel@us.ibm.com>
19296 * config/rs6000/rs6000-c: Add support for built-in functions
19297 vector unsigned long long vec_bperm (vector unsigned long long,
19298 vector unsigned char)
19299 vector signed long long vec_mule (vector signed int,
19301 vector unsigned long long vec_mule (vector unsigned int,
19302 vector unsigned int)
19303 vector signed long long vec_mulo (vector signed int,
19305 vector unsigned long long vec_mulo (vector unsigned int,
19306 vector unsigned int)
19307 vector signed char vec_sldw (vector signed char,
19308 vector signed char,
19310 vector unsigned char vec_sldw (vector unsigned char,
19311 vector unsigned char,
19313 vector signed short vec_sldw (vector signed short,
19314 vector signed short,
19316 vector unsigned short vec_sldw (vector unsigned short,
19317 vector unsigned short,
19319 vector signed int vec_sldw (vector signed int,
19322 vector unsigned int vec_sldw (vector unsigned int,
19323 vector unsigned int,
19325 vector signed long long vec_sldw (vector signed long long,
19326 vector signed long long,
19328 vector unsigned long long vec_sldw (vector unsigned long long,
19329 vector unsigned long long,
19331 * config/rs6000/rs6000-c: Add support for built-in functions
19332 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
19333 * config/rs6000/altivec.h: Add defintion for vec_sldw.
19334 * doc/extend.texi: Update the built-in documentation for the
19335 new built-in functions.
19337 2017-05-16 Marek Polacek <polacek@redhat.com>
19341 * tree.c (save_expr): Don't fold the expression.
19343 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
19345 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
19346 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
19347 and (?*y,m). Update insn attributes.
19349 2017-05-16 Martin Liska <mliska@suse.cz>
19351 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
19352 flags argument of print_gimple_stmt, print_gimple_expr,
19353 print_generic_stmt and print_generic_expr.
19354 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
19355 * coretypes.h: Likewise.
19356 * except.c (dump_eh_tree): Likewise.
19357 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
19358 * gimple-pretty-print.h: Likewise.
19359 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
19360 (backprop::push_to_worklist): Likewise.
19361 (backprop::pop_from_worklist): Likewise.
19362 (backprop::process_use): Likewise.
19363 (backprop::intersect_uses): Likewise.
19364 (note_replacement): Likewise.
19365 * gimple-ssa-store-merging.c
19366 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
19367 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
19368 (pass_store_merging::execute): Likewise.
19369 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
19370 (ssa_base_cand_dump_callback): Likewise.
19371 (dump_incr_vec): Likewise.
19372 (replace_refs): Likewise.
19373 (replace_mult_candidate): Likewise.
19374 (create_add_on_incoming_edge): Likewise.
19375 (create_phi_basis): Likewise.
19376 (insert_initializers): Likewise.
19377 (all_phi_incrs_profitable): Likewise.
19378 (introduce_cast_before_cand): Likewise.
19379 (replace_one_candidate): Likewise.
19380 * gimplify.c (gimplify_expr): Likewise.
19381 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
19382 (set_rename): Likewise.
19383 (rename_uses): Likewise.
19384 (copy_loop_phi_nodes): Likewise.
19385 (add_close_phis_to_merge_points): Likewise.
19386 (copy_loop_close_phi_args): Likewise.
19387 (copy_cond_phi_args): Likewise.
19388 (graphite_copy_stmts_from_block): Likewise.
19389 (translate_pending_phi_nodes): Likewise.
19390 * graphite-poly.c (print_pdr): Likewise.
19391 (dump_gbb_cases): Likewise.
19392 (dump_gbb_conditions): Likewise.
19393 (print_scop_params): Likewise.
19394 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
19395 (build_cross_bb_scalars_use): Likewise.
19396 (gather_bbs::before_dom_children): Likewise.
19397 * hsa-dump.c (dump_hsa_immed): Likewise.
19398 * ipa-cp.c (print_ipcp_constant_value): Likewise.
19399 (get_replacement_map): Likewise.
19400 * ipa-inline-analysis.c (dump_condition): Likewise.
19401 (estimate_function_body_sizes): Likewise.
19402 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
19403 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
19404 * ipa-prop.c (ipa_dump_param): Likewise.
19405 (ipa_print_node_jump_functions_for_edge): Likewise.
19406 (ipa_modify_call_arguments): Likewise.
19407 (ipa_modify_expr): Likewise.
19408 (ipa_dump_param_adjustments): Likewise.
19409 (ipa_dump_agg_replacement_values): Likewise.
19410 (ipcp_modif_dom_walker::before_dom_children): Likewise.
19411 * ipa-pure-const.c (check_stmt): Likewise.
19412 (pass_nothrow::execute): Likewise.
19413 * ipa-split.c (execute_split_functions): Likewise.
19414 * omp-offload.c (dump_oacc_loop_part): Likewise.
19415 (dump_oacc_loop): Likewise.
19416 * trans-mem.c (tm_log_emit): Likewise.
19417 (tm_memopt_accumulate_memops): Likewise.
19418 (dump_tm_memopt_set): Likewise.
19419 (dump_tm_memopt_transform): Likewise.
19420 * tree-cfg.c (gimple_verify_flow_info): Likewise.
19421 (print_loop): Likewise.
19422 * tree-chkp-opt.c (chkp_print_addr): Likewise.
19423 (chkp_gather_checks_info): Likewise.
19424 (chkp_get_check_result): Likewise.
19425 (chkp_remove_check_if_pass): Likewise.
19426 (chkp_use_outer_bounds_if_possible): Likewise.
19427 (chkp_reduce_bounds_lifetime): Likewise.
19428 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
19429 (chkp_mark_completed_bounds): Likewise.
19430 (chkp_register_incomplete_bounds): Likewise.
19431 (chkp_mark_invalid_bounds): Likewise.
19432 (chkp_maybe_copy_and_register_bounds): Likewise.
19433 (chkp_build_returned_bound): Likewise.
19434 (chkp_get_bound_for_parm): Likewise.
19435 (chkp_build_bndldx): Likewise.
19436 (chkp_get_bounds_by_definition): Likewise.
19437 (chkp_generate_extern_var_bounds): Likewise.
19438 (chkp_get_bounds_for_decl_addr): Likewise.
19439 * tree-chrec.c (chrec_apply): Likewise.
19440 * tree-data-ref.c (dump_data_reference): Likewise.
19441 (dump_subscript): Likewise.
19442 (dump_data_dependence_relation): Likewise.
19443 (analyze_overlapping_iterations): Likewise.
19444 * tree-inline.c (expand_call_inline): Likewise.
19445 (tree_function_versioning): Likewise.
19446 * tree-into-ssa.c (dump_defs_stack): Likewise.
19447 (dump_currdefs): Likewise.
19448 (dump_names_replaced_by): Likewise.
19449 (dump_update_ssa): Likewise.
19450 (update_ssa): Likewise.
19451 * tree-object-size.c (pass_object_sizes::execute): Likewise.
19452 * tree-parloops.c (build_new_reduction): Likewise.
19453 (try_create_reduction_list): Likewise.
19454 (ref_conflicts_with_region): Likewise.
19455 (oacc_entry_exit_ok_1): Likewise.
19456 (oacc_entry_exit_single_gang): Likewise.
19457 * tree-pretty-print.h: Likewise.
19458 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
19459 (get_scalar_evolution): Likewise.
19460 (add_to_evolution): Likewise.
19461 (get_loop_exit_condition): Likewise.
19462 (analyze_evolution_in_loop): Likewise.
19463 (analyze_initial_condition): Likewise.
19464 (analyze_scalar_evolution): Likewise.
19465 (instantiate_scev): Likewise.
19466 (number_of_latch_executions): Likewise.
19467 (gather_chrec_stats): Likewise.
19468 (final_value_replacement_loop): Likewise.
19469 (scev_const_prop): Likewise.
19470 * tree-sra.c (dump_access): Likewise.
19471 (disqualify_candidate): Likewise.
19472 (create_access): Likewise.
19473 (reject): Likewise.
19474 (maybe_add_sra_candidate): Likewise.
19475 (create_access_replacement): Likewise.
19476 (analyze_access_subtree): Likewise.
19477 (analyze_all_variable_accesses): Likewise.
19478 (sra_modify_assign): Likewise.
19479 (initialize_constant_pool_replacements): Likewise.
19480 (find_param_candidates): Likewise.
19481 (decide_one_param_reduction): Likewise.
19482 (replace_removed_params_ssa_names): Likewise.
19483 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
19484 * tree-ssa-copy.c (dump_copy_of): Likewise.
19485 (copy_prop_visit_cond_stmt): Likewise.
19486 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
19487 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
19488 (record_equivalences_from_stmt): Likewise.
19489 * tree-ssa-dse.c (compute_trims): Likewise.
19490 (delete_dead_call): Likewise.
19491 (delete_dead_assignment): Likewise.
19492 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
19493 (forward_propagate_into_cond): Likewise.
19494 (pass_forwprop::execute): Likewise.
19495 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
19496 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
19498 (move_computations_worker): Likewise.
19499 (execute_sm): Likewise.
19500 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
19501 (remove_exits_and_undefined_stmts): Likewise.
19502 (remove_redundant_iv_tests): Likewise.
19503 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
19504 (adjust_iv_update_pos): Likewise.
19505 * tree-ssa-math-opts.c (bswap_replace): Likewise.
19506 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
19507 (value_replacement): Likewise.
19508 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
19509 * tree-ssa-pre.c (print_pre_expr): Likewise.
19510 (get_representative_for): Likewise.
19511 (create_expression_by_pieces): Likewise.
19512 (insert_into_preds_of_block): Likewise.
19513 (eliminate_insert): Likewise.
19514 (eliminate_dom_walker::before_dom_children): Likewise.
19515 (eliminate): Likewise.
19516 (remove_dead_inserted_code): Likewise.
19517 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
19518 * tree-ssa-reassoc.c (get_rank): Likewise.
19519 (eliminate_duplicate_pair): Likewise.
19520 (eliminate_plus_minus_pair): Likewise.
19521 (eliminate_not_pairs): Likewise.
19522 (undistribute_ops_list): Likewise.
19523 (eliminate_redundant_comparison): Likewise.
19524 (update_range_test): Likewise.
19525 (optimize_range_tests_var_bound): Likewise.
19526 (optimize_vec_cond_expr): Likewise.
19527 (rewrite_expr_tree): Likewise.
19528 (rewrite_expr_tree_parallel): Likewise.
19529 (linearize_expr): Likewise.
19530 (break_up_subtract): Likewise.
19531 (linearize_expr_tree): Likewise.
19532 (attempt_builtin_powi): Likewise.
19533 (attempt_builtin_copysign): Likewise.
19534 (transform_stmt_to_copy): Likewise.
19535 (transform_stmt_to_multiply): Likewise.
19536 (dump_ops_vector): Likewise.
19537 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
19538 (print_scc): Likewise.
19539 (set_ssa_val_to): Likewise.
19540 (visit_reference_op_store): Likewise.
19541 (visit_use): Likewise.
19542 (sccvn_dom_walker::before_dom_children): Likewise.
19543 (run_scc_vn): Likewise.
19544 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
19546 (expr_hash_elt::print): Likewise.
19547 (const_and_copies::pop_to_marker): Likewise.
19548 (const_and_copies::record_const_or_copy_raw): Likewise.
19549 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
19550 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
19551 (dump_predicates): Likewise.
19552 (find_uninit_use): Likewise.
19553 (warn_uninitialized_phi): Likewise.
19554 (pass_late_warn_uninitialized::execute): Likewise.
19555 * tree-ssa.c (verify_vssa): Likewise.
19556 (verify_ssa): Likewise.
19557 (maybe_optimize_var): Likewise.
19558 * tree-vrp.c (dump_value_range): Likewise.
19559 (dump_all_value_ranges): Likewise.
19560 (dump_asserts_for): Likewise.
19561 (register_edge_assert_for_2): Likewise.
19562 (vrp_visit_cond_stmt): Likewise.
19563 (vrp_visit_switch_stmt): Likewise.
19564 (vrp_visit_stmt): Likewise.
19565 (vrp_visit_phi_node): Likewise.
19566 (simplify_cond_using_ranges_1): Likewise.
19567 (fold_predicate_in): Likewise.
19568 (evrp_dom_walker::before_dom_children): Likewise.
19569 (evrp_dom_walker::push_value_range): Likewise.
19570 (evrp_dom_walker::pop_value_range): Likewise.
19571 (execute_early_vrp): Likewise.
19573 2017-05-16 Richard Biener <rguenther@suse.de>
19575 * dwarf2out.c (loc_list_from_tree_1): Do not create
19576 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
19578 2017-05-16 Richard Biener <rguenther@suse.de>
19580 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
19582 (note_variable_value_in_expr): If we resolved the decl ref
19583 do not push to the stack.
19585 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
19587 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
19588 operations in fast-math mode.
19589 (vaddq_f16): Likewise.
19590 (vmul_f16): Likewise.
19591 (vmulq_f16): Likewise.
19592 (vsub_f16): Likewise.
19593 (vsubq_f16): Likewise.
19594 * config/arm/neon.md (add<mode>3): New.
19596 (fma:<VH:mode>3): New. Also remove outdated comment.
19599 2017-05-16 Martin Liska <mliska@suse.cz>
19603 * ipa-devirt.c (warn_types_mismatch): Fix typo.
19604 (odr_types_equivalent_p): Likewise.
19606 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
19608 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
19610 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
19613 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
19614 non-interunit SSE move alternatives with '?'.
19615 (zero-extendsidi peephole2): New peephole to skip intermediate
19616 general register in SSE zero-extend sequence.
19618 2017-05-15 Jeff Law <law@redhat.com>
19620 * reorg.c (relax_delay_slots): Create a new variable to hold
19621 the temporary target rather than clobbering TARGET_LABEL.
19623 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
19624 missing argument to extract_bit_field call.
19625 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
19627 2017-05-15 Martin Liska <mliska@suse.cz>
19630 * gcc.c (process_command): Do not allow empty argument of -o option.
19632 2017-05-15 Renlin Li <renlin.li@arm.com>
19634 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
19635 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
19636 * config/aarch64/constraints.md (Usf): Add long call check.
19637 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
19638 (call_value): Likewise.
19639 (sibcall): Likewise.
19640 (sibcall_value): Likewise.
19642 (call_value_insn): New.
19643 (sibcall_insn): Update rtx pattern.
19644 (sibcall_value_insn): Likewise.
19645 (call_internal): Remove.
19646 (call_value_internal): Likewise.
19647 (sibcall_internal): Likewise.
19648 (sibcall_value_internal): Likewise.
19649 (call_reg): Likewise.
19650 (call_symbol): Likewise.
19651 (call_value_reg): Likewise.
19652 (call_value_symbol): Likewise.
19654 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
19657 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
19659 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
19661 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
19662 compatible with CCGOCmode and with CCZmode.
19664 2017-05-14 Martin Sebor <msebor@redhat.com>
19666 PR middle-end/77671
19667 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
19668 (gimple_fold_builtin_snprintf): Same.
19669 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
19670 (gimple_fold_builtin_snprintf): Same.
19671 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
19672 of character types.
19673 (is_call_safe): New function.
19674 (try_substitute_return_value): Call it.
19675 (try_simplify_call): New function.
19676 (pass_sprintf_length::handle_gimple_call): Call it.
19678 2017-05-14 Martin Sebor <msebor@redhat.com>
19680 PR middle-end/80669
19681 * builtins.c (expand_builtin_stpncpy): Simplify.
19683 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
19685 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
19686 * config/i386/i386.h
19687 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
19688 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
19689 (struct machine_function): Add new members call_ms2sysv,
19690 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
19691 (struct machine_frame_state): New fields sp_realigned and
19692 sp_realigned_offset.
19693 * config/i386/i386.c
19694 (enum xlogue_stub): New enum.
19695 (enum xlogue_stub_sets): New enum.
19696 (class xlogue_layout): New class.
19697 (struct ix86_frame): New fields stack_realign_allocate_offset,
19698 stack_realign_offset and outlined_save_offset. Modify comments to
19699 detail stack layout when using out-of-line stubs.
19700 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
19701 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
19702 -mcall-ms2sysv-xlogues.
19703 (stub_managed_regs): New static variable.
19704 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
19705 registers managed by out-of-line stub.
19706 (disable_call_ms2sysv_xlogues): New function.
19707 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
19708 m->call_ms2sysv when appropriate and compute frame layout for
19710 (sp_valid_at, fp_valid_at): New inline functions.
19711 (choose_basereg): New function.
19712 (choose_baseaddr): Add align parameter, use choose_basereg and modify
19714 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
19715 Use align parameter of choose_baseaddr to generated aligned SSE movs
19717 (pro_epilogue_adjust_stack): Modify to track
19718 machine_frame_state::sp_realigned.
19719 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
19720 (ix86_nsaved_sseregs): Likewise.
19721 (ix86_emit_save_regs): Likewise.
19722 (ix86_emit_save_regs_using_mov): Likewise.
19723 (ix86_emit_save_sse_regs_using_mov): Likewise.
19724 (get_scratch_register_on_entry): Likewise.
19725 (gen_frame_set): New function.
19726 (gen_frame_load): Likewise.
19727 (gen_frame_store): Likewise.
19728 (emit_outlined_ms2sysv_save): Likewise.
19729 (emit_outlined_ms2sysv_restore): Likewise.
19730 (ix86_expand_prologue): Modify stack re-alignment code and call
19731 emit_outlined_ms2sysv_save when appropriate.
19732 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
19733 parameter rtx_insn *insn, which allows the function to be used to only
19734 generate the notes.
19735 (ix86_expand_epilogue): Modify validity checks of frame and stack
19736 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
19737 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
19738 * config/i386/predicates.md
19739 (save_multiple): New predicate.
19740 (restore_multiple): Likewise.
19741 * config/i386/sse.md
19742 (save_multiple<mode>): New pattern.
19743 (save_multiple_realign<mode>): Likewise.
19744 (restore_multiple<mode>): Likewise.
19745 (restore_multiple_and_return<mode>): Likewise.
19746 (restore_multiple_leave_return<mode>): Likewise.
19747 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
19749 2017-05-14 Julia Koval <julia.koval@intel.com>
19751 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
19752 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
19753 (__builtin_ia32_xsetbv): New builtins.
19754 * config/i386/i386.c (ix86_expand_special_args_builtin):
19756 (ix86_expand_builtin): Special expand for new intrinsics.
19757 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
19758 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
19759 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
19761 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19763 * cfganal.c (inverted_post_order_compute): Change argument type
19765 * cfganal.h (inverted_post_order_compute): Adjust prototype.
19766 * df-core.c (rest_of_handle_df_initialize): Adjust.
19767 (rest_of_handle_df_finish): Likewise.
19768 (df_analyze_1): Likewise.
19769 (df_analyze): Likewise.
19770 (loop_inverted_post_order_compute): Change argument to be a vec *.
19771 (df_analyze_loop): Adjust.
19772 (df_get_n_blocks): Likewise.
19773 (df_get_postorder): Likewise.
19774 * df.h (struct df_d): Change field to be a vec.
19775 * lcm.c (compute_laterin): Adjust.
19776 (compute_available): Likewise.
19777 * lra-lives.c (lra_create_live_ranges_1): Likewise.
19778 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
19779 * tree-ssa-pre.c (compute_antic): Likewise.
19781 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19783 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
19784 (depth_first_search::depth_first_search): Change structure init
19785 function to this constructor.
19786 (depth_first_search::add_bb): Rename function to this member.
19787 (depth_first_search::execute): Likewise.
19788 (flow_dfs_compute_reverse_finish): Adjust.
19790 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19792 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
19793 (longest_simple_path): Likewise.
19794 * shrink-wrap.c (spread_components): Likewise.
19795 (disqualify_problematic_components): Likewise.
19796 (emit_common_heads_for_components): Likewise.
19797 (emit_common_tails_for_components): Likewise.
19798 (insert_prologue_epilogue_for_components): Likewise.
19800 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19802 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
19805 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19807 * df-core.c (df_set_blocks): Start using auto_bitmap.
19808 (df_compact_blocks): Likewise.
19809 * df-problems.c (df_rd_confluence_n): Likewise.
19810 * df-scan.c (df_insn_rescan_all): Likewise.
19811 (df_process_deferred_rescans): Likewise.
19812 (df_update_entry_block_defs): Likewise.
19813 (df_update_exit_block_uses): Likewise.
19814 (df_entry_block_bitmap_verify): Likewise.
19815 (df_exit_block_bitmap_verify): Likewise.
19816 (df_scan_verify): Likewise.
19817 * lra-constraints.c (lra_constraints): Likewise.
19818 (undo_optional_reloads): Likewise.
19819 (lra_undo_inheritance): Likewise.
19820 * lra-remat.c (calculate_gen_cands): Likewise.
19821 (do_remat): Likewise.
19822 * lra-spills.c (assign_spill_hard_regs): Likewise.
19823 (spill_pseudos): Likewise.
19824 * tree-ssa-pre.c (bitmap_set_and): Likewise.
19825 (bitmap_set_subtract_values): Likewise.
19827 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19829 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
19830 management with auto_bitmap.
19831 (fix_inter_tick): Likewise.
19832 (fix_recovery_deps): Likewise.
19833 * ira.c (add_store_equivs): Likewise.
19834 (find_moveable_pseudos): Likewise.
19835 (split_live_ranges_for_shrink_wrap): Likewise.
19836 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
19837 (rtx_reuse_manager::seen_def_p): Likewise.
19838 (rtx_reuse_manager::set_seen_def): Likewise.
19839 * print-rtl.h (class rtx_reuse_manager): Likewise.
19841 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19843 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
19845 (migrate_btr_def): Likewise.
19846 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
19847 * df-core.c (loop_post_order_compute): Likewise.
19848 (loop_inverted_post_order_compute): Likewise.
19849 * hsa-common.h: Likewise.
19850 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
19851 * init-regs.c (initialize_uninitialized_regs): Likewise.
19852 * ipa-inline.c (resolve_noninline_speculation): Likewise.
19853 (inline_small_functions): Likewise.
19854 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
19855 * ira.c (combine_and_move_insns): Likewise.
19856 (build_insn_chain): Likewise.
19857 * loop-invariant.c (find_invariants): Likewise.
19858 * lower-subreg.c (propagate_pseudo_copies): Likewise.
19859 * predict.c (tree_predict_by_opcode): Likewise.
19860 (predict_paths_leading_to): Likewise.
19861 (predict_paths_leading_to_edge): Likewise.
19862 (estimate_loops_at_level): Likewise.
19863 (estimate_loops): Likewise.
19864 * shrink-wrap.c (try_shrink_wrapping): Likewise.
19865 (spread_components): Likewise.
19866 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
19867 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
19868 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
19869 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
19870 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
19871 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
19872 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
19873 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
19874 (mark_threaded_blocks): Likewise.
19875 (thread_through_all_blocks): Likewise.
19876 * tree-ssa.c (verify_ssa): Likewise.
19877 (execute_update_addresses_taken): Likewise.
19878 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
19880 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19882 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
19884 (post_order_compute): Likewise.
19885 (inverted_post_order_compute): Likewise.
19886 (pre_and_rev_post_order_compute_fn): Likewise.
19888 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19890 * genrecog.c (int_set::int_set): Explicitly construct our
19891 auto_vec base class.
19892 * vec.h (auto_vec::auto_vec): New constructor.
19894 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19896 * bitmap.h (class auto_bitmap): New constructor taking
19897 bitmap_obstack * argument.
19899 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19901 * bitmap.h (class auto_bitmap): Change type of m_bits to
19902 bitmap_head, and adjust ctor / dtor and member operators.
19904 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
19906 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
19907 when returned register mode doesn't match original mode.
19909 2017-05-12 Jeff Law <law@redhat.com>
19910 Jakub Jelinek <jakub@redhat.com>
19912 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
19913 we look for cc setter after the compare-elim changes.
19914 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
19915 within the vector to match what compare-elim now expects.
19916 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
19917 (xorsi3_flags, one_cmplsi2_flags): Likewise.
19919 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
19920 after the compare-elim changes.
19921 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
19922 the vector to match what compare-elim now expects.
19923 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
19924 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
19925 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
19926 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
19927 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
19929 * config/visium/visium.c (single_set_and_flags): Fix where
19930 we look for cc setter after the compare-elim changes.
19931 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
19932 with the vector to match what compare-elim now expects.
19933 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
19934 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
19935 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
19936 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
19937 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
19938 (neg<mode>2_insn_set_overflow): Likewise.
19940 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
19942 PR middle-end/79794
19943 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
19944 maybe_expand_insn call, set ops[0].target. If still set after call,
19945 set alt_rtl. Add extra arg to recursive calls.
19946 (extract_bit_field): Add alt_rtl argument. Pass to
19948 * expmed.h (extract_bit_field): Fix prototype.
19949 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
19950 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
19951 to extract_bit_field_calls.
19952 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
19953 Pass alt_rtl to extract_bit_field calls.
19954 * calls.c (store_unaligned_arguments_into_psuedos)
19955 load_register_parameters): Pass extra NULL to extract_bit_field calls.
19956 * optabs.c (maybe_legitimize_operand): Clear op->target when call
19958 * optabs.h (struct expand_operand): Add target bitfield.
19960 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19962 * compare-elim.c (try_eliminate_compare): Canonicalize
19963 operation with embedded compare to
19964 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
19965 (set (reg) (operation)].
19967 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
19969 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19972 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
19973 cost of adding a carry flag for ADC instruction.
19974 [case MINUS]: Ignore the cost of subtracting a carry flag
19975 for SBB instruction.
19977 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
19979 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
19981 * config/rs6000/bmiintrin.h: New file.
19982 * config/rs6000/bmi2intrin.h: New file.
19983 * config/rs6000/x86intrin.h: New file.
19985 2017-05-12 Jeff Law <law@redhat.com>
19987 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
19990 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
19992 PR middle-end/80707
19993 * tree-cfg.c: Remove cfg edges of unreachable case statements.
19995 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19997 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19998 early expansion of vector divide builtins.
19999 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
20000 builtins identified as having unsigned arguments.
20002 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20004 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
20005 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
20006 expansion of vector logical operations (and, andc, or, xor,
20009 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20011 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
20012 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
20014 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20016 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20017 early GIMPLE expansion of vector multiplies.
20019 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20021 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
20022 TARGET_HAVE_MOVT conditional.
20023 (movt splitter): Likewise.
20025 2017-05-12 Richard Biener <rguenther@suse.de>
20027 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
20028 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20029 Fold all stmts not inplace.
20031 2017-05-12 Richard Biener <rguenther@suse.de>
20033 PR tree-optimization/80713
20034 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
20035 inserted_exprs bit for not removed stmts.
20037 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
20039 PR middle-end/69921
20040 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
20041 parallelized" attribute for parallelized OpenACC kernels.
20042 * omp-offload.c (execute_oacc_device_lower): Use it.
20044 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
20045 Set "oacc kernels" attribute.
20046 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
20047 parameter. Adjust all users.
20048 (oacc_fn_attrib_kernels_p): Remove function.
20049 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
20050 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
20051 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
20052 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
20053 assert "oacc kernels" attribute is set.
20055 2017-05-11 Carl Love <cel@us.ibm.com>
20057 * config/rs6000/rs6000-c: Add support for built-in functions
20058 vector unsigned char vec_popcnt (vector signed char)
20059 vector unsigned char vec_popcnt (vector unsigned char)
20060 vector unsigned short vec_popcnt (vector signed short)
20061 vector unsigned short vec_popcnt (vector unsigned short)
20062 vector unsigned int vec_popcnt (vector signed int)
20063 vector unsigned int vec_popcnt (vector unsigned int)
20064 vector unsigned long long vec_popcnt (vector signed long long)
20065 vector unsigned long long vec_popcnt (vector unsigned long long)
20066 vector signed long long vec_slo (vector signed long long,
20067 vector signed char)
20068 vector signed long long vec_slo (vector signed long long,
20069 vector unsigned char)
20070 vector unsigned long long vec_slo (vector unsigned long long,
20071 vector signed char)
20072 vector unsigned long long vec_slo (vector unsigned long long,
20073 vector unsigned char)
20074 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
20075 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
20076 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
20077 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
20078 * doc/extend.texi: Update the built-in documentation file for the
20079 new built-in functions.
20081 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20083 * attribs.h (sorted_attr_string): Move machine independent
20084 functions for target clone support from the i386 port to common
20085 code. Rename ix86_function_versions to common_function_versions.
20086 Rename make_name to make_unique_name.
20087 (common_function_versions): Likewise.
20088 (make_unique_name): Likewise.
20089 (make_dispatcher_decl): Likewise.
20090 (is_function_default_version): Likewise.
20091 * attribs.c (attr_strcmp): Likewise.
20092 (sorted_attr_string): Likewise.
20093 (common_function_versions): Likewise.
20094 (make_unique_name): Likewise.
20095 (make_dispatcher_decl): Likewise.
20096 (is_function_default_version): Likewise.
20097 * config/i386/i386.c (attr_strcmp): Likewise.
20098 (sorted_attr_string): Likewise.
20099 (ix86_function_versions): Likewise.
20100 (make_name): Likewise.
20101 (make_dispatcher_decl): Likewise.
20102 (is_function_default_version): Likewise.
20103 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
20105 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20108 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20109 Account for direct move costs for vec_construct of integer
20112 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
20115 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
20116 (UNSPEC_STX_ATOMIC): Ditto.
20117 (loaddi_via_sse): New insn.
20118 (storedi_via_sse): Ditto.
20119 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
20120 Update corresponding peephole2 patterns.
20121 (atomic_storedi_fpu): Ditto.
20123 2017-05-11 Julia Koval <julia.koval@intel.com>
20125 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
20126 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
20128 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
20129 (__builtin_ia32_rsqrt14ss_mask): New builtins.
20130 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
20132 2017-05-11 Nathan Sidwell <nathan@acm.org>
20134 * graphite-poly.c: Include dumpfile.h.
20136 * dumpfle.h (dump_function): Declare here ...
20137 * tree-dump.h (dump_function): ... not here.
20138 * dumpfile.c: #include tree-cfg.h.
20139 (dump_function): Move here from ...
20140 * tree-dump.c (dump_function): ... here.
20141 * gimplify.c: #include splay-tree.h, not tree-dump.h.
20142 * graphite-poly.c: Don't include tree-dump.h.
20143 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
20144 * print-tree.c: Likewise.
20145 * stor-layout.c: Likewise.
20146 * tree-nested.c: Likewise.
20148 * dumpfile.c (dump_start): Use TDF_FLAGS.
20149 (dump_enable_all): Fix TDF_KIND check thinko.
20151 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
20153 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20154 array entries to represent two legal parameterizations of the
20155 overloaded __builtin_cmpb function, as represented by the
20156 P6_OV_BUILTIN_CMPB constant.
20157 (altivec_resolve_overloaded_builtin): Add special case handling
20158 for the __builtin_cmpb function, as represented by the
20159 P6_OV_BUILTIN_CMPB constant.
20160 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
20161 (BU_P6_64BIT_2): New macro.
20162 (BU_P6_OVERLOAD_2): New macro
20163 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
20164 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
20165 (CMPB): Add overload support to represent both 32-bit and 64-bit
20166 compare-bytes function.
20167 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
20168 support for TARGET_CMPB.
20169 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
20170 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20171 documentation of the __builtin_cmpb overloaded built-in function.
20173 2017-05-11 Richard Biener <rguenther@suse.de>
20175 PR tree-optimization/80705
20176 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
20177 bases are not vectorizable.
20179 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20181 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
20182 when counting register pressure.
20184 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20186 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
20187 (iv_ca_more_deps): Renamed to ...
20188 (iv_ca_compare_deps): ... this.
20189 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
20191 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20193 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
20195 (determine_group_iv_costs): ... here.
20196 (find_inv_vars_cb): Record inv var if it's not recorded before.
20198 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20200 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
20201 (get_shiftadd_cost): Ditto.
20203 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20205 * tree-ssa-address.c: Include header file.
20206 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
20208 (add_to_parts): Refactor.
20209 (addr_to_parts): New parameter. Update use of move_hint_to_base.
20210 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
20213 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20215 PR tree-optimization/53090
20216 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
20218 (extract_cond_operands): Detect condition with IV on both sides
20219 and return COMP_IV_EXPR_2.
20220 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
20221 (rewrite_use_compare): Simplify by removing call to function
20222 extract_cond_operands.
20224 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20226 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
20227 (extract_cond_operands): Detect condition comparing against non-
20228 invariant bound and return appropriate enum value.
20229 (find_interesting_uses_cond): Update use of extract_cond_operands.
20230 Handle its return value accordingly.
20231 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
20233 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20235 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
20236 nonlinear iv_use computation in loop invariant sensitive way.
20238 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20240 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
20241 (find_iv_candidates): Call relate_compare_use_with_all_cands.
20243 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20245 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
20246 (dump_cand): Support iv_cand.inv_exprs.
20247 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
20249 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
20252 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20254 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
20256 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
20257 as local function. Include necessary header files.
20258 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
20260 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20262 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
20264 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20266 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
20267 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
20268 RSHIFT_EXPR and BIT_NOT_EXPR.
20270 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20272 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
20273 (adjust_setup_cost): New parameter supporting round up adjustment.
20274 (struct address_cost_data): Delete.
20275 (force_expr_to_var_cost): Don't bound cost with spill_cost.
20276 (split_address_cost, ptr_difference_cost): Delete.
20277 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
20278 (struct ainc_cost_data): New struct.
20279 (get_address_cost_ainc): New function.
20280 (get_address_cost, get_computation_cost): Reimplement.
20281 (determine_group_iv_cost_address): Record inv_expr for all uses of
20283 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
20284 (iv_ca_has_deps): Reimplemented to ...
20285 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
20287 (iv_ca_extend): Call iv_ca_more_deps.
20289 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20291 * tree-ssa-address.c (struct mem_address): Move to header file.
20292 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
20293 * tree-ssa-address.h (struct mem_address): Move from C file.
20294 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
20296 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20298 * tree-affine.h (aff_combination_type): New interface.
20299 (aff_combination_zero_p): Remove static.
20300 (aff_combination_const_p): New interface.
20301 (aff_combination_singleton_var_p): New interfaces.
20303 2017-05-11 Richard Biener <rguenther@suse.de>
20305 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20306 Skip unreachable blocks and destinations.
20307 (eliminate): Move stmt removal and fixup ...
20308 (fini_eliminate): ... here. Skip inserted exprs.
20309 (pass_pre::execute): Move fini_pre after fini_eliminate.
20310 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
20311 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
20312 PRE to get rid of dead code that has invalid SSA form and
20313 split critical edges again.
20315 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20317 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20319 2017-05-11 Richard Biener <rguenther@suse.de>
20321 * passes.c (execute_function_todo): Verify loops if they are
20322 said to be up-to-date.
20323 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
20324 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
20326 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
20329 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
20330 handle calling assemble_external ourself.
20333 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
20334 modes with zero size. Enhance comment.
20336 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20338 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
20339 built-ins for vec_xl and vec_xst with short and char pointer
20342 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20344 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
20345 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
20346 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
20347 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
20348 (_mm_maskz_min_round_ss): New intrinsics.
20349 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
20350 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
20351 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
20352 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
20353 (__builtin_ia32_minss_mask_round): New builtins.
20354 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20355 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20356 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
20358 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
20359 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
20361 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
20364 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20366 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
20367 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
20368 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
20369 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
20370 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
20371 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
20372 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
20373 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20374 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20375 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
20376 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
20377 (__builtin_ia32_mulss_mask_round): New builtins.
20378 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20379 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20380 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
20382 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
20383 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20385 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20388 2017-05-10 Julia Koval <julia.koval@intel.com>
20390 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
20391 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
20392 (_mm256_setr_m128i): New intrinsics.
20394 2017-05-10 Julia Koval <julia.koval@intel.com>
20396 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
20397 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
20398 (_mm_maskz_rcp14_ss): New intrinsics.
20399 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
20400 (__builtin_ia32_rcp14ss_mask): New builtins.
20401 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
20403 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
20405 PR tree-optimization/51513
20406 * tree-cfg.c (gimple_seq_unreachable_p): New function.
20407 (assert_unreachable_fallthru_edge_p): Use it.
20408 (group_case_labels_stmt): Likewise.
20409 * tree-cfg.h: Prototype it.
20410 * stmt.c: Include cfghooks.h and tree-cfg.h.
20411 (emit_case_dispatch_table) <gap_label>: New local variable.
20412 Use it to fill dispatch table gaps.
20413 Test for default_label before updating probabilities.
20414 (expand_case) <default_label>: Remove unneeded initialization.
20415 Test for unreachable default case statement and remove its edge.
20416 Set default_label accordingly.
20417 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
20419 2017-05-10 Carl Love <cel@us.ibm.com>
20421 * config/rs6000/rs6000-c: Add support for built-in functions
20422 vector signed char vec_neg (vector signed char)
20423 vector signed short int vec_neg (vector short int)
20424 vector signed int vec_neg (vector signed int)
20425 vector signed long long vec_neg (vector signed long long)
20426 vector float vec_neg (vector float)
20427 vector double vec_neg (vector double)
20428 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
20430 * config/rs6000/altivec.h: Add define for vec_neg
20431 * doc/extend.texi: Update the built-in documentation for the
20432 new built-in functions.
20434 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20436 PR tree-optimization/77644
20437 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
20439 2017-05-10 Nathan Sidwell <nathan@acm.org>
20441 * dumpfile.h (TDI_lang_all): New.
20442 (TDF_KIND): New. Renumber others
20443 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
20445 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
20447 (get_dump_file_name): Adjust suffix generation.
20448 (dump_enable_all): Use TDF_KIND.
20449 * doc/invoke.texi (-fdump-lang-all): Document.
20451 * dumpfile.h: Tabify.
20453 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
20456 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
20457 Move member access before delete.
20459 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
20461 * tree-inline.c (expand_call_inline): Split block at stmt
20464 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
20467 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
20468 are now unused after splitting mov{sf,sd}_hardfloat.
20469 (f32_lr2): Likewise.
20470 (f32_lm): Likewise.
20471 (f32_lm2): Likewise.
20472 (f32_li): Likewise.
20473 (f32_li2): Likewise.
20474 (f32_lv): Likewise.
20475 (f32_sr): Likewise.
20476 (f32_sr2): Likewise.
20477 (f32_sm): Likewise.
20478 (f32_sm2): Likewise.
20479 (f32_si): Likewise.
20480 (f32_si2): Likewise.
20481 (f32_sv): Likewise.
20482 (f32_dm): Likewise.
20483 (f32_vsx): Likewise.
20484 (f32_av): Likewise.
20485 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
20486 For movsf, order stores so the VSX stores occur before the GPR
20487 store which encourages the register allocator to use a traditional
20488 FPR instead of a GPR. For movsd, order the stores so that the GPR
20489 store comes before the VSX stores to allow the power6 to work.
20490 This is due to the power6 not having a 32-bit integer store
20491 instruction from a FPR.
20492 (movsf_hardfloat): Likewise.
20493 (movsd_hardfloat): Likewise.
20495 2017-05-09 Martin Sebor <msebor@redhat.com>
20497 PR translation/80280
20498 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
20501 PR translation/80280
20502 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
20503 data member added in r247778.
20504 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
20506 2017-05-09 Nathan Sidwell <nathan@acm.org>
20508 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
20510 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
20513 2017-05-09 Marek Polacek <polacek@redhat.com>
20515 * doc/invoke.texi: Fix typo.
20517 2017-05-09 Richard Biener <rguenther@suse.de>
20519 * tree-vrp.c (vrp_val_is_max): Adjust comment.
20520 (vrp_val_is_min): Likewise.
20521 (set_value_range_to_value): Likewise.
20522 (set_value_range_to_nonnegative): Likewise.
20523 (gimple_assign_nonzero_p): Likewise.
20524 (gimple_stmt_nonzero_p): Likewise.
20525 (vrp_int_const_binop): Likewise. Remove unreachable case.
20526 (adjust_range_with_scev): Adjust comments.
20527 (compare_range_with_value): Likewise.
20528 (extract_range_from_phi_node): Likewise.
20529 (test_for_singularity): Likewise.
20531 2017-05-09 Richard Biener <rguenther@suse.de>
20533 * tree-vrp.c (get_single_symbol): Add assert that we don't
20534 get overflowed constants as invariant part.
20535 (compare_values_warnv): Add comment before the TREE_NO_WARNING
20536 checks. Use wi::cmp instead of recursing for integer constants.
20537 (compare_values): Just ignore whether we assumed undefined
20538 overflow instead of failing the compare.
20539 (extract_range_for_var_from_comparison_expr): Add comment before the
20540 TREE_NO_WARNING sets.
20541 (test_for_singularity): Likewise.
20542 (extract_range_from_comparison): Do not disable optimization
20543 when we assumed undefined overflow.
20544 (extract_range_basic): Remove init of unused var.
20546 2017-05-09 Richard Biener <rguenther@suse.de>
20548 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
20549 (extract_range_from_multiplicative_op_1): Adjust.
20550 (extract_range_from_binary_expr_1): Use int_const_binop.
20552 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20555 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
20556 rs6000_store_data_bypass_p in seven define_bypass directives and
20557 in several comments.
20558 * config/rs6000/rs6000-protos.h: Add prototype for
20559 rs6000_store_data_bypass_p function.
20560 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
20561 function implements slightly different (rs6000-specific) semantics
20562 than store_data_bypass_p, returning false rather than aborting
20563 with assertion error when arguments do not satisfy the
20564 requirements of store data bypass.
20565 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
20566 rs6000_store_data_bypass_p.
20568 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
20570 * config/xtensa/xtensa-protos.h
20571 (xtensa_initial_elimination_offset): New declaration.
20572 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20573 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
20574 macro definition, add case for FRAME_POINTER_REGNUM when
20575 FRAME_GROWS_DOWNWARD.
20576 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
20577 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
20578 xtensa_initial_elimination_offset.
20580 2017-05-08 Nathan Sidwell <nathan@acm.org>
20582 * doc/invoke.texi: Alphabetize -fdump options.
20584 2017-05-08 Martin Sebor <msebor@redhat.com>
20586 PR translation/80280
20587 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
20589 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
20591 * target.def (compute_frame_layout): New optional target hook.
20592 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
20593 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
20594 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
20596 * reload1.c (verify_initial_elim_offsets): Likewise.
20597 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
20598 (use_simple_return_p): Call arm_compute_frame_layout if needed.
20599 (arm_get_frame_offsets): Split up into this ...
20600 (arm_compute_frame_layout): ... and this function.
20602 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
20604 * config/aarch64/constraints.md (Usa): New constraint.
20605 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
20607 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
20609 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
20610 with_multilib_list after it has been checked.
20612 2017-05-08 Richard Biener <rguenther@suse.de>
20614 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
20615 (bitmap_set_subtract_values): Likewise.
20617 2017-05-08 Richard Biener <rguenther@suse.de>
20619 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
20620 (gimple_assign_nonzero): ... this and remove strict_overflow_p
20622 (gimple_stmt_nonzero_warnv_p): Rename to ...
20623 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
20625 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
20626 (extract_range_basic): Adjust, do not disable propagation on
20627 strict overflow sensitive simplification.
20628 (vrp_visit_cond_stmt): Likewise.
20630 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
20632 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
20633 body size unconditionally.
20635 2017-05-07 Jeff Law <law@redhat.com>
20638 2017-05-06 Jeff Law <law@redhat.com>
20639 PR tree-optimization/78496
20640 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
20643 PR tree-optimization/78496
20644 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
20645 (simplify_stmt_using_ranges): Call it.
20646 (vrp_dom_walker::before_dom_children): Extract equivalences
20647 from an ASSERT_EXPR with an equality comparison against a
20650 2017-05-06 Jeff Law <law@redhat.com>
20652 PR tree-optimization/78496
20653 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
20656 PR tree-optimization/78496
20657 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
20658 (simplify_stmt_using_ranges): Call it.
20659 (vrp_dom_walker::before_dom_children): Extract equivalences
20660 from an ASSERT_EXPR with an equality comparison against a
20663 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
20665 * lra-constraints.c (lra_copy_reg_equiv): New function.
20666 (split_reg): Use it to copy equivalence information from the
20667 original register to the spill register.
20669 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
20671 PR rtl-optimization/75964
20672 * simplify-rtx.c (simplify_const_relational_operation): Remove
20673 invalid handling of comparisons of integer ABS.
20675 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
20677 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
20678 initialize to zero.
20679 (init_regs): Remove declaration.
20680 (function_arg_advance_32): Initialize error_p as boolean variable.
20682 2017-05-05 Nathan Sidwell <nathan@acm.org>
20684 * store-motion.c (remove_reachable_equiv_notes): Reformat long
20685 lines. Use for (;;).
20687 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20689 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
20690 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
20691 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
20692 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
20693 VF=2 that require versioning.
20695 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20697 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
20700 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20702 * diagnostic.h (diagnostic_override_option_index): Convert from
20703 macro to inline function.
20705 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20707 * diagnostic.c (last_module_changed_p): New function.
20708 (set_last_module): New function.
20709 (diagnostic_report_current_module): Convert macro usage to
20710 the above functions.
20711 * diagnostic.h (diagnostic_context::last_module): Strengthen
20712 from const line_map * to const line_map_ordinary *.
20713 (diagnostic_last_module_changed): Delete macro.
20714 (diagnostic_set_last_module): Delete macro.
20716 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20718 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
20719 with diagnostic_report_diagnostic.
20720 (diagnostic_n_impl_richloc): Likewise.
20721 * diagnostic.h (report_diagnostic): Delete macro.
20722 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
20723 with diagnostic_report_diagnostic.
20724 * substring-locations.c (format_warning_va): Likewise.
20726 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20728 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
20729 save/restor of format_spec. Move option-printing code to...
20730 (print_option_information): ...this new function, and
20731 reimplement by simply printing to the pretty_printer,
20732 rather than appending to the format string.
20734 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20736 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
20737 handling logic into...
20738 (update_effective_level_from_pragmas): ...this new function.
20740 2017-05-04 Andrew Waterman <andrew@sifive.com>
20742 * config/riscv/riscv.opt (mstrict-align): New option.
20743 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
20744 (SLOW_UNALIGNED_ACCESS): Define.
20745 (riscv_slow_unaligned_access): Declare.
20746 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
20748 (riscv_slow_unaligned_access): New variable.
20749 (rocket_tune_info): Set slow_unaligned_access to true.
20750 (optimize_size_tune_info): Set slow_unaligned_access to false.
20751 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
20752 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
20753 (riscv_option_override): Set riscv_slow_unaligned_access.
20754 * doc/invoke.texi: Add -mstrict-align to RISC-V.
20756 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
20758 * config/riscv/riscv.md: Unify indentation.
20760 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20765 * config/rs6000/rs6000.md (u code attribute): Add FIX and
20767 (extendsi<mode>2): Add support for doing sign extension via
20768 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
20769 don't have ISA 3.0 instructions.
20770 (extendsi<mode>2 splitter): Likewise.
20771 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
20772 generate the normal insns since SImode can now go in vector
20773 registers. Disallow the special UNSPECs needed for previous
20774 machines to hide SImode being used. Add new insns
20775 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
20776 (fix_trunc<mode>si2_stfiwx): Likewise.
20777 (fix_trunc<mode>si2_internal): Likewise.
20778 (fixuns_trunc<mode>si2): Likewise.
20779 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20780 (fctiw<u>z_<mode>_smallint): Likewise.
20781 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
20782 of floating point to 32-bit integer from doing a direct move to
20783 the GPR registers to do a store.
20784 (fctiwz_<mode>): Break long line.
20786 2017-05-05 Bin Cheng <bin.cheng@arm.com>
20788 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
20789 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
20790 (addr_list, addr_offset_valid_p): New.
20791 (split_address_groups): Check offset validity with above function.
20792 (gt-tree-ssa-loop-ivopts.h): Include header file.
20794 2017-05-05 Nathan Sidwell <nathan@acm.org>
20796 * config.gcc (arm*-*-*): Add missing 'fi'.
20798 2017-05-05 Steve Ellcey <sellcey@cavium.com>
20800 * doc/invoke.texi (-fopt-info): Explicitly say order of options
20801 included in -fopt-info does not matter.
20802 * doc/optinfo.texi (-fopt-info): Fix description of default
20803 behavour. Explicitly say order of options included in -fopt-info
20806 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20808 * config.gcc: Allow combinations of aprofile and rmprofile values for
20809 --with-multilib-list.
20810 * config/arm/t-multilib: New file.
20811 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
20812 variables. Remove setting of ISA and floating-point ABI in
20813 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
20814 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
20815 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
20816 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
20818 * config/arm/t-rmprofile: Likewise except for the matches changes.
20819 * doc/install.texi (--with-multilib-list): Document the combination of
20820 aprofile and rmprofile values and warn about pitfalls in doing that.
20822 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20824 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
20825 (movdi_aarch64): Likewise.
20827 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20829 PR tree-optimization/80632
20830 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
20832 (build_arrays): Initialize it for virtual phis.
20833 (fix_phi_nodes): Use it for virtual phis.
20835 PR tree-optimization/80558
20836 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
20837 [x, y] op z into [x op, y op z] for op & or | if conditions
20840 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
20841 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20844 * config/arm/arm.md (use_literal_pool): Remove.
20845 (64-bit immediate split): No longer takes cost into consideration
20846 if arm_disable_literal_pool is enabled.
20847 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
20848 used when arm_disable_literal_pool is enabled.
20849 (arm_max_const_double_inline_cost): Remove use of
20850 arm_disable_literal_pool.
20851 (push_minipool_fix): Add assert.
20852 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
20853 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
20854 (no_literal_pool_sf_immediate): New.
20856 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20858 PR tree-optimization/80613
20859 * tree-ssa-dce.c (propagate_necessity): Remove cases for
20860 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
20862 2017-05-05 Richard Biener <rguenther@suse.de>
20864 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
20866 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
20868 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
20869 of this flag from insn conditions due to removal from r247495.
20871 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20873 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
20875 (arm_early_store_addr_dep_ptr): Likewise.
20876 * config/arm/aarch-common-protos.h
20877 (arm_early_load_addr_dep_ptr): Add prototype.
20878 (arm_early_store_addr_dep_ptr): Likewise.
20879 * config/arm/cortex-a53.md: Add new bypasses.
20881 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20883 * tree.c (next_type_uid): Change type to unsigned.
20884 (type_hash_canon): Decrement back next_type_uid if
20885 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
20886 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
20889 2017-05-04 Martin Sebor <msebor@redhat.com>
20891 * builtins.c: Fix a trivial typo in a comment.
20893 PR middle-end/79234
20894 * builtins.c (check_sizes): Adjust to handle reading past the end.
20895 Avoid printing excessive upper bound of ranges. Use %E to print
20896 tree nodes instead of converting them to %wu.
20897 (expand_builtin_memchr): New function.
20898 (compute_dest_size): Rename...
20899 (compute_objsize): ...to this.
20900 (expand_builtin_memcpy): Adjust.
20901 (expand_builtin_mempcpy): Adjust.
20902 (expand_builtin_strcat): Adjust.
20903 (expand_builtin_strcpy): Adjust.
20904 (check_strncat_sizes): Adjust.
20905 (expand_builtin_strncat): Adjust.
20906 (expand_builtin_strncpy): Adjust and simplify.
20907 (expand_builtin_memset): Adjust.
20908 (expand_builtin_bzero): Adjust.
20909 (expand_builtin_memcmp): Adjust.
20910 (expand_builtin): Handle memcmp.
20911 (maybe_emit_chk_warning): Check strncat just once.
20913 2017-05-04 Martin Sebor <msebor@redhat.com>
20915 PR preprocessor/79214
20916 PR middle-end/79222
20917 PR middle-end/79223
20918 * builtins.c (check_sizes): Add inlining context and issue
20919 warnings even when -Wno-system-headers is set.
20920 (check_strncat_sizes): Same.
20921 (expand_builtin_strncat): Same.
20922 (expand_builtin_memmove): New function.
20923 (expand_builtin_stpncpy): Same.
20924 (expand_builtin): Handle memmove and stpncpy.
20926 2017-05-04 Bin Cheng <bin.cheng@arm.com>
20928 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
20929 which is not used any more.
20931 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20933 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
20935 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20937 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
20938 (cortexa53_tunings): Likewise.
20939 (cortexa57_tunings): Likewise.
20940 (cortexa72_tunings): Likewise.
20941 (cortexa73_tunings): Likewise.
20943 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20945 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
20946 Set loop alignment to 8.
20948 2017-05-04 Martin Sebor <msebor@redhat.com>
20950 PR translation/80280
20951 * builtins.c (expand_builtin_object_size): Add missing quoting to
20952 %D and like directives.
20953 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
20954 (hsa_type_for_tree_type): Same.
20955 (verify_function_arguments): Same.
20956 * symtab.c (symbol_table::change_decl_assembler_name): Same.
20957 * varasm.c (get_section): Same.
20960 2017-05-04 Martin Sebor <msebor@redhat.com>
20962 PR translation/80280
20963 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
20965 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20967 * config/aarch64/aarch64.c (generic_addrcost_table):
20968 Change HI/TI mode setting.
20970 2017-05-04 Martin Jambor <mjambor@suse.cz>
20972 PR tree-optimization/80622
20973 * tree-sra.c (comes_initialized_p): New function.
20974 (build_accesses_from_assign): Only set write lazily when
20975 comes_initialized_p is false.
20976 (analyze_access_subtree): Use comes_initialized_p.
20977 (propagate_subaccesses_across_link): Assert !comes_initialized_p
20978 instead of testing for PARM_DECL.
20980 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20982 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
20983 constraint on operand 0 to allow more general addressing modes.
20984 Adjust output template.
20985 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
20987 * config/aarch64/aarch64-protos.h
20988 (aarch64_address_valid_for_prefetch_p): Declare prototype.
20989 * config/aarch64/constraints.md (Dp): New address constraint.
20990 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
20993 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
20995 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
20996 update use of estimate_ipcp_clone_size_and_time.
20997 (estimate_local_effects): Update use of
20998 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
20999 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
21000 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
21001 Return nonspecialized time.
21003 2017-05-04 Richard Biener <rguenther@suse.de>
21005 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
21006 for the last VUSE which def dominates the PHI. Directly call
21008 (get_continuation_for_phi_1): Remove.
21010 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
21012 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
21013 to explain the use of truncating division. Cap the number of
21014 iterations to the maximum given by nb_iterations_upper_bound,
21017 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21019 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
21020 * configure: Regenerate.
21021 * config.in: Regenerate.
21022 * config/i386/driver-mingw32.c: new file.
21023 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
21024 * config.host: Link driver-mingw32.o on MinGW host.
21025 * doc/install.texi: Document new --enable-mingw-wildcard configure
21028 2017-05-04 Marek Polacek <polacek@redhat.com>
21030 PR tree-optimization/80612
21031 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
21033 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21034 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
21036 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
21037 (movt splitter): Likewise.
21038 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
21039 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
21040 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
21041 block for Thumb-1 with MOVT.
21042 (thumb2_legitimate_address_p): Move code block ...
21043 (can_avoid_literal_pool_for_label_p): ... into this new function.
21044 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
21046 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
21047 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
21048 "M-profile targets with the MOVT instruction".
21050 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21052 * config/arm/arm-builtins.c (arm_init_builtins): Rename
21053 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
21054 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
21056 2017-05-04 Martin Liska <mliska@suse.cz>
21058 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
21059 variable cond_code.
21061 2017-05-04 Richard Biener <rguenther@suse.de>
21063 * tree.c (array_at_struct_end_p): Handle arrays at struct
21064 end with flexarrays more conservatively. Refactor and treat
21065 arrays of arrays or aggregates more strict. Fix
21066 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
21067 * tree.c (array_at_struct_end_p): Adjust prototype.
21068 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
21069 * gimple-fold.c (get_range_strlen): Likewise.
21070 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
21072 2017-05-04 Richard Biener <rguenther@suse.de>
21074 PR tree-optimization/31130
21075 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
21077 (supports_overflow_infinity): Likewise.
21078 (is_negative_overflow_infinity): Likewise.
21079 (is_positive_overflow_infinity): Likewise.
21080 (is_overflow_infinity): Likewise.
21081 (stmt_overflow_infinity): Likewise.
21082 (overflow_infinity_range_p): Likewise.
21083 (usable_range_p): Remove as always returning true.
21084 (make_overflow_infinity): Remove.
21085 (negative_overflow_infinity): Likewise.
21086 (positive_overflow_infinity): Likewise.
21087 (avoid_overflow_infinity): Likewise.
21088 (set_value_range): Adjust accordingly.
21089 (set_value_range_to_nonnegative): Likewise, remove now unused
21090 overflow_infinity arg.
21091 (vrp_operand_equal_p): Adjust.
21092 (update_value_range): Likewise.
21093 (range_int_cst_singleton_p): Likewise.
21094 (operand_less_p): Likewise.
21095 (compare_values_warnv): Likewise.
21096 (extract_range_for_var_from_comparison_expr): Likewise.
21097 (vrp_int_const_binop): Likewise.
21098 (zero_nonzero_bits_from_vr): Likewise.
21099 (extract_range_from_multiplicative_op_1): Likewise.
21100 (extract_range_from_binary_expr_1): Likewise.
21101 (extract_range_from_unary_expr): Likewise.
21102 (extract_range_from_comparison): Likewise.
21103 (extract_range_basic): Likewise.
21104 (adjust_range_with_scev): Likewise.
21105 (compare_ranges): Likewise.
21106 (compare_range_with_value): Likewise.
21107 (dump_value_range): Likewise.
21108 (test_for_singularity): Likewise, remove strict_overflow_p parameter
21110 (simplify_cond_using_ranges): Adjust.
21112 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21114 * brig-builtins.def: Added a builtin for class_f64.
21115 * builtin-types.def: Added a builtin type needed by class_f64.
21117 2017-05-03 Jason Merrill <jason@redhat.com>
21119 * timevar.def: Add TV_CONSTEXPR.
21121 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21123 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
21125 2017-05-03 Martin Jambor <mjambor@suse.cz>
21127 * ipa-prop.c (ipa_update_after_lto_read): Removed.
21128 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
21129 * ipa-cp.c (ipcp_propagate_stage): Do not call
21130 ipa_update_after_lto_read.
21131 * ipa-inline.c (ipa_inline): Likewise.
21133 2017-05-03 Martin Jambor <mjambor@suse.cz>
21135 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
21136 tag. Added a default constructor and a destructor.
21137 (ipa_edge_args_sum_t): New class;
21138 (ipa_edge_args_sum): Declare.
21139 (ipa_edge_args_vector): Remove declaration.
21140 (IPA_EDGE_REF): Use ipa_edge_args_sum.
21141 (ipa_free_edge_args_substructures): Remove declaration.
21142 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
21143 (ipa_edge_args_info_available_for_edge_p): Likewise.
21144 * ipa-prop.c (ipa_edge_args_vector): Removed.
21145 (edge_removal_hook_holder): Likewise.
21146 (edge_duplication_hook_holder): Likewise.
21147 (ipa_edge_args_sum): New variable.
21148 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
21149 ipa_edge_args_vector.
21150 (ipa_free_edge_args_substructures): Likewise.
21151 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
21152 ipa_edge_args_vector.
21153 (ipa_edge_removal_hook): Turned into method
21154 ipa_edge_args_sum_t::remove.
21155 (ipa_edge_duplication_hook): Turned into method
21156 ipa_edge_args_sum_t::duplicate.
21157 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
21158 registering edge hooks.
21159 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
21160 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
21161 ipa_edge_args_sum instead of ipa_edge_args_vector.
21162 * ipa-profile.c (ipa_profile): Likewise.
21164 2017-05-03 Martin Jambor <mjambor@suse.cz>
21166 * symbol-summary.h (function_summary): New method exists.
21167 (function_summary::symtab_removal): Deallocate through release.
21168 (call_summary): New class.
21169 (gt_ggc_mx): New overload.
21170 (gt_pch_nx): Likewise.
21171 (gt_pch_nx): Likewise.
21173 2017-05-03 Jeff Law <law@redhat.com>
21175 PR tree-optimization/78496
21176 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
21177 from simplify_cond_using_ranges. Split off code to walk
21178 backwards through casts into ...
21179 (simplify_cond_using_ranges_2): New function.
21180 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
21181 (execute_vrp): After identifying jump threads, call
21182 simplify_cond_using_ranges_2.
21184 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
21187 * ipa-inline.h (inline_summary): Add ctor.
21188 (create_ggc): Do not use ggc_cleared_alloc.
21190 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
21191 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21193 * gcc.c (handle_braces): Support escaping in switch matching
21195 * doc/invoke.texi (Spec Files): Document it.
21196 Remove superfluous @code markup in items.
21198 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21200 * diagnostic-show-locus.c (struct column_range): New struct.
21201 (get_affected_columns): New function.
21202 (get_printed_columns): New function.
21203 (struct correction): New struct.
21204 (correction::ensure_capacity): New function.
21205 (correction::ensure_terminated): New function.
21206 (struct line_corrections): New struct.
21207 (line_corrections::~line_corrections): New dtor.
21208 (line_corrections::add_hint): New function.
21209 (layout::print_trailing_fixits): Reimplement in terms of the new
21211 (selftest::test_overlapped_fixit_printing): New function.
21212 (selftest::diagnostic_show_locus_c_tests): Call it.
21214 2017-05-03 Nathan Sidwell <nathan@acm.org>
21216 Canonicalize canonical type hashing
21217 * tree.h (type_hash_canon_hash): Declare.
21218 * tree.c (type_hash_list, attribute_hash_list): Move into
21219 type_hash_canon_hash.
21220 (build_type_attribute_qual_variant): Break out hash code calc into
21221 type_hash_canon_hash.
21222 (type_hash_canon_hash): New. Generic type hash computation.
21223 (build_range_type_1, build_array_type_1, build_function_type,
21224 build_method_type_directly, build_offset_type, build_complex_type,
21225 make_vector_type): Call it.
21227 2017-05-03 Richard Biener <rguenther@suse.de>
21229 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
21230 When all DRs have unknown misaligned do not always peel
21231 when there is a store but apply the same costing model as if
21232 there were only loads.
21234 2017-05-03 Richard Biener <rguenther@suse.de>
21237 PR tree-optimization/80492
21238 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21239 compare_base_decls returning dont-know properly.
21241 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
21243 * config/arm/iterators.md (CCSI): New mode iterator.
21244 (arch): New mode attribute.
21245 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
21246 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
21247 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
21248 code iterator for success result mode.
21249 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
21250 the corresponding new insn generators.
21252 2017-05-03 Bin Cheng <bin.cheng@arm.com>
21255 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21256 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21258 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
21260 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
21261 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
21262 (DDR_A): Wrap DDR argument in brackets.
21263 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
21264 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
21265 (DDR_REVERSED_P): Likewise.
21267 2017-05-03 Jakub Jelinek <jakub@redhat.com>
21269 PR tree-optimization/79472
21270 * tree-switch-conversion.c (struct switch_conv_info): Add
21271 contiguous_range and default_case_nonstandard fields.
21272 (collect_switch_conv_info): Compute contiguous_range and
21273 default_case_nonstandard fields, don't clear final_bb if
21274 contiguous_range and only the default case doesn't have the required
21276 (check_all_empty_except_final): Set default_case_nonstandard instead
21277 of failing if contiguous_range and the default case doesn't have empty
21279 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
21280 and only the default case doesn't have the required constants. Skip
21282 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
21283 if default_case_nonstandard.
21284 (build_constructors): Build constant 1 just once. Assert that default
21285 values aren't inserted in between cases if contiguous_range. Skip
21287 (build_arrays): Skip virtual phis.
21288 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
21289 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
21290 Handle virtual phis.
21291 (gen_inbound_check): Handle default_case_nonstandard case.
21292 (process_switch): Adjust check_final_bb caller. Call
21293 gather_default_values with the first non-default case instead of
21294 default case if default_case_nonstandard.
21296 2017-05-02 Nathan Sidwell <nathan@acm.org>
21298 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
21299 check. Fix formatting.
21301 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
21303 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
21304 errors when comparing specialized and unspecialized times.
21306 2017-05-02 David Malcolm <dmalcolm@redhat.com>
21308 * diagnostic-show-locus.c
21309 (layout::should_print_annotation_line_p): Make private.
21310 (layout::print_annotation_line): Make private.
21311 (layout::annotation_line_showed_range_p): Make private.
21312 (layout::show_ruler): Make private.
21313 (layout::print_source_line): Make private. Pass in line and
21314 line_width, rather than calling location_get_source_line. Drop
21316 (layout::print_leading_fixits): New method.
21317 (layout::print_any_fixits): Rename to...
21318 (layout::print_trailing_fixits): ...this, and make private.
21319 Don't print newline fixits.
21320 (diagnostic_show_locus): Move logic for printing one row into...
21321 (layout::print_line): ...this new function. Move the
21322 location_get_source_line call and error-handling from
21323 print_source_line to here. Call print_leading_fixits, and rename
21324 print_any_fixits to print_trailing_fixits.
21325 (selftest::test_fixit_insert_containing_newline): Update now that
21326 newlines are partially supported.
21327 (selftest::test_fixit_insert_containing_newline_2): New test.
21328 (selftest::test_fixit_replace_containing_newline): Update comments.
21329 (selftest::diagnostic_show_locus_c_tests): Call the new test.
21330 * edit-context.c (class added_line): New class.
21331 (class edited_line): Describe newline handling in comment.
21332 (edited_line::actually_edited_p): New method.
21333 (edited_line::print_content): Delete redundant decl.
21334 (edited_line::m_predecessors): New field.
21335 (edited_file::print_content): Call edited_line::print_content.
21336 (edited_file::print_diff): Update to support newlines.
21337 (edited_file::print_diff_hunk): Likewise.
21338 (edited_file::print_run_of_changed_lines): New function.
21339 (edited_file::print_diff_line): Convert to...
21340 (print_diff_line): ...this.
21341 (edited_file::get_effective_line_count): New function.
21342 (edited_line::edited_line): Initialize new field m_predecessors.
21343 (edited_line::~edited_line): Clean up m_predecessors.
21344 (edited_line::apply_fixit): Handle newlines.
21345 (edited_line::get_effective_line_count): New function.
21346 (edited_line::print_content): New function.
21347 (edited_line::print_diff_lines): New function.
21348 (selftest::test_applying_fixits_insert_containing_newline): New
21350 (selftest::test_applying_fixits_replace_containing_newline): New
21352 (selftest::insert_line): New function.
21353 (selftest::test_applying_fixits_multiple_lines): Add example of
21355 (selftest::edit_context_c_tests): Call the new tests.
21357 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21359 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
21360 parameter cand. Update dump information.
21361 (get_computation_cost): Update uses.
21363 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21365 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
21366 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
21367 (get_computation_at, rewrite_use_address): Update use of
21368 get_computation_aff.
21370 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21372 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
21373 (get_computation): Delete.
21374 (get_computation_cost): Implement like get_computation_cost_at.
21375 Use get_computation_at.
21376 (get_computation_cost_at): Delete.
21377 (rewrite_use_nonlinear_expr): Use get_computation_at.
21378 (rewrite_use_compare, remove_unused_ivs): Ditto.
21380 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21382 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
21384 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21386 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
21387 (ivopts_global_cost_for_size): Rename parameter and update uses.
21388 (iv_ca_recount_cost): Update uses.
21389 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
21390 candidates seperately in n_invs and n_cands.
21391 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
21393 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21395 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
21396 (find_inv_vars_cb): New.
21397 (find_depends): Renamed to ...
21398 (find_inv_vars): ... this.
21399 (add_candidate_1, force_var_cost): Call find_inv_vars.
21400 (split_address_cost, determine_group_iv_cost_cond): Ditto.
21402 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21404 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
21405 inv_vars. Add inv_exprs.
21406 (struct iv_cand): Rename depends_on to inv_vars.
21407 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
21408 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
21409 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
21410 (dump_cand): Dump inv_vars.
21411 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
21412 (record_invariant, find_depends, add_candidate_1): Ditto.
21413 (set_group_iv_cost, force_var_cost): Ditto.
21414 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
21415 (get_computation_cost_at, get_computation_cost): Ditto.
21416 (determine_group_iv_cost_generic): Ditto.
21417 (determine_group_iv_cost_address): Ditto.
21418 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
21419 (determine_group_iv_costs): Ditto.
21420 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
21421 (iv_ca_set_remove_invariants): Renamed to ...
21422 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
21423 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
21424 (iv_ca_set_add_invariants): Renamed to ...
21425 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
21426 (iv_ca_set_cp): Use iv_ca_set_add_invs.
21427 (iv_ca_has_deps): Support inv_vars and inv_exprs.
21428 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
21429 (create_new_ivs): Remove useless dump.
21431 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21433 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
21435 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
21436 (iv_ca_set_no_cp, create_new_iv): Ditto.
21438 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21440 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21442 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21444 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
21445 function tree_check2.
21447 2017-05-02 Martin Liska <mliska@suse.cz>
21449 * doc/gcov.texi: Add missing preposition.
21450 * gcov.c (function_info::function_info): Properly fill up
21451 all member variables.
21453 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21455 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
21457 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21459 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
21461 2017-05-02 Martin Liska <mliska@suse.cz>
21464 * lto-streamer-in.c (lto_read_tree_1): Remove
21465 LTO_STREAMER_DEBUG.
21466 * lto-streamer.c (struct tree_hash_entry): Likewise.
21467 (struct tree_entry_hasher): Likewise.
21468 (tree_entry_hasher::hash): Likewise.
21469 (tree_entry_hasher::equal): Likewise.
21470 (lto_streamer_init): Likewise.
21471 (lto_orig_address_map): Likewise.
21472 (lto_orig_address_get): Likewise.
21473 (lto_orig_address_remove): Likewise.
21474 * lto-streamer.h: Likewise.
21475 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
21476 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
21478 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
21480 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
21481 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
21482 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
21483 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
21484 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
21485 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
21486 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
21487 (mm_maskz_sub_ss): New intrinsics.
21488 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21489 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21490 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
21491 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
21492 (__builtin_ia32_subss_mask_round): New builtins.
21493 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21494 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21495 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
21497 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
21498 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21500 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21503 2017-05-02 Martin Jambor <mjambor@suse.cz>
21505 PR tree-optimization/78687
21506 * tree-sra.c (access): New field parent.
21507 (process_subtree_disqualification): New function.
21508 (disqualify_candidate): Call it.
21509 (build_accesses_from_assign): Reset write flag if creating an
21511 (build_access_subtree): Fill in parent field and also prpagate
21512 down grp_write flag.
21513 (create_artificial_child_access): New parameter set_grp_write, set
21514 grp_write to its value.
21515 (propagate_subaccesses_across_link): Also propagate grp_write flag
21517 (propagate_all_subaccesses): Push the closest parent back to work
21518 queue if add_access_to_work_queue returned true.
21520 2017-05-02 Richard Biener <rguenther@suse.de>
21522 * common.opt (fstrict-overflow): Alias negative to fwrapv.
21523 * doc/invoke.texi (fstrict-overflow): Remove all traces of
21524 -fstrict-overflow documentation.
21525 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
21526 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
21527 flag_strict_overflow.
21528 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
21529 * lto-opts.c (lto_write_options): Do not stream it.
21530 * lto-wrapper.c (merge_and_complain): Do not handle it.
21531 * opts.c (default_options_table): Do not set -fstrict-overflow.
21532 (finish_options): Likewise do not clear it when sanitizing.
21533 * simplify-rtx.c (simplify_const_relational_operation): Do not
21534 test flag_strict_overflow.
21536 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21538 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
21539 using enabled attribute.
21540 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
21541 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
21542 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
21543 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
21544 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
21545 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
21546 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
21547 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
21548 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
21549 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
21551 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21553 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
21555 2017-05-02 Richard Biener <rguenther@suse.de>
21557 PR tree-optimization/80591
21559 2017-04-10 Richard Biener <rguenther@suse.de>
21561 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21564 2017-05-02 Richard Biener <rguenther@suse.de>
21566 PR tree-optimization/80549
21567 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
21568 (cleanup_tree_cfg_noloop): Create forwarders to known loop
21569 headers if they do not have a preheader.
21571 2017-05-02 Martin Liska <mliska@suse.cz>
21574 * common.opt: Fix typo.
21575 * doc/invoke.texi: Likewise.
21577 2017-05-01 Jan Beulich <jbeulich@suse.com>
21579 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
21580 swapping, add (x,x,m,x,n) alternative.
21582 2017-05-01 Nathan Sidwell <nathan@acm.org>
21584 * calls.c (combine_pending_stack_adjustment_and_call): Remove
21585 unnecessary unadjusted_alignment check.
21587 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21590 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
21592 * gimplify.c (gimplify_cilk_detach): New function.
21593 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
21594 * tree-core.h: Document EXPR_CILK_SPAWN.
21595 * tree.h (EXPR_CILK_SPAWN): Define.
21597 2017-05-01 David Malcolm <dmalcolm@redhat.com>
21599 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
21600 to use new fixit_hint representation, using the "replace" logic.
21601 (get_line_span_for_fixit_hint): Likewise.
21602 (layout::print_any_fixits): Likewise.
21603 (selftest::test_one_liner_many_fixits): Rename to...
21604 (selftest::test_one_liner_many_fixits_1): ...this, and update
21605 comment and expected output to reflect that the multiple fix-it
21606 hints are now consolidated into one insertion.
21607 (selftest::test_one_liner_many_fixits_2): New test.
21608 (selftest::test_diagnostic_show_locus_one_liner): Update for
21610 (selftest::test_fixit_consolidation): Update for fix-it API
21612 * diagnostic.c (print_parseable_fixits): Likewise.
21613 * edit-context.c (edited_line::m_line_events): Convert from
21614 auto_vec <line_event *> to auto_vec <line_event>.
21615 (class line_event): Convert from abstract base class to a concrete
21616 class, taking over the role of replace_event.
21617 (class insert_event): Delete.
21618 (class replace_event): Rename to class line_event. Convert to
21620 (edit_context::add_fixits): Reimplement.
21621 (edit_context::apply_insert): Delete.
21622 (edit_context::apply_replace): Rename to...
21623 (edit_context::apply_fixit): ...this. Convert to half-open range.
21624 (edited_file::apply_insert): Delete.
21625 (edited_file::apply_replace): Rename to...
21626 (edited_file::apply_fixit): ...this.
21627 (edited_line::~edited_line): Drop deletion of events.
21628 (edited_line::apply_insert): Delete.
21629 (edited_line::apply_replace): Rename to...
21630 (edited_line::apply_fixit): ...this. Convert to half-open range.
21631 Update for change to type of m_line_events.
21632 * edit-context.h (edit_context::apply_insert): Delete.
21633 (edit_context::apply_replace): Rename to...
21634 (edit_context::apply_fixit): ...this.
21636 2017-05-01 Martin Sebor <msebor@redhat.com>
21638 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
21641 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
21644 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
21645 __get_cpuid_max returns 0.
21646 (__get_cpuid_count): Ditto.
21648 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
21650 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
21651 replacement expression is another instance of one of its arguments.
21653 2017-05-01 Jakub Jelinek <jakub@redhat.com>
21656 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
21657 check for stack push/pop autoinc.
21658 * config/i386/i386.c (ix86_agi_dependent): Return false
21659 if the only reason why modified_in_p returned true is that
21660 addr is SP based and set_insn is a push or pop.
21662 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
21664 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
21667 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
21670 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
21671 (account_size_time): Use two predicates - exec_pred and
21673 (evaluate_conditions_for_known_args): Compute both clause and
21675 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
21676 (inline_summary_t::duplicate): Update.
21677 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
21679 (compute_inline_parameters): Likewise.
21680 (estimate_edge_size_and_time): Update caluclation of time.
21681 (estimate_node_size_and_time): Compute both time and nonspecialized
21683 (estimate_ipcp_clone_size_and_time): Update.
21684 (inline_merge_summary): Update.
21685 (do_estimate_edge_time): Update.
21686 (do_estimate_edge_size): Update.
21687 (do_estimate_edge_hints): Update.
21688 (inline_read_section, inline_write_summary): Stream both new predicates.
21689 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
21691 (compute_inlined_call_time): Cleanup.
21692 (big_speedup_p): Update.
21693 (edge_badness): Update.
21694 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
21695 (size_time_entry): Replace predicate by exec_predicate and
21696 nonconst_predicate.
21697 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
21698 (estimate_edge_time): Return also nonspec_time.
21699 (reset_edge_growth_cache): Update.
21701 2017-04-29 Jakub Jelinek <jakub@redhat.com>
21703 PR rtl-optimization/80491
21704 * ifcvt.c (noce_process_if_block): When looking for x setter
21705 with missing else_bb, don't check only the insn right before
21706 cond_earliest, but look for the last insn that x is modified in
21707 within the same bb.
21709 PR rtl-optimization/80491
21710 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
21712 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
21714 PR tree-optimization/80487
21715 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
21717 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21719 PR tree-optimization/79697
21720 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
21721 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
21722 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
21724 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
21725 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
21727 2017-04-28 Martin Sebor <msebor@redhat.com>
21729 PR tree-optimization/80523
21730 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
21731 (init_target_to_host_charmap, target_to_host, target_strtol10): New
21733 (maybe_warn, format_directive, parse_directive): Use new functions.
21734 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
21736 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
21738 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
21740 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
21742 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
21743 target_header_dir): Set correctly.
21744 * configure: Regenerated.
21745 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21746 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21747 instead of SYSTEM_HEADER_DIR.
21749 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
21751 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
21752 (estimate_local_effects): Likewise.
21753 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
21754 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
21755 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
21756 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
21757 do_estimate_edge_time, estimate_edge_time): Likewise.
21758 * ipa-inline-analysis.c (estimate_node_size_and_time,
21759 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
21760 (estimate_time_after_inlining): Remove.
21762 2017-04-28 Martin Liska <mliska@suse.cz>
21764 * doc/gcov.texi: Enhance documentation of gcov.
21766 2017-04-28 Martin Liska <mliska@suse.cz>
21768 * doc/gcov.texi: Sort options in alphabetic order.
21769 * doc/gcov-dump.texi: Likewise.
21770 * doc/gcov-tool.texi: Likewise.
21771 * gcov.c (print_usage): Likewise.
21772 * gcov-dump.c (print_usage): Likewise.
21773 * gcov-tool.c (print_merge_usage_message): Likewise.
21774 (print_rewrite_usage_message): Likewise.
21775 (print_overlap_usage_message): Likewise.
21777 2017-04-28 Martin Liska <mliska@suse.cz>
21779 PR gcov-profile/53915
21780 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
21782 2017-04-28 Martin Liska <mliska@suse.cz>
21784 PR gcov-profile/79891
21785 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
21786 is marked by compiler as living on a line.
21787 (get_cycles_count): Remove usage of the union.
21788 (output_intermediate_file): Likewise.
21789 (find_source): Fix GNU coding style.
21790 (accumulate_line_counts): Remove old non-all block mode.
21791 (output_lines): Remove usage of the union.
21792 * profile.c (output_location): Include all BBs, even if
21793 belonging to a same line (and file) as a previous BB.
21795 2017-04-28 Martin Liska <mliska@suse.cz>
21797 * gcov.c (process_args): Handle new argument 'w'.
21798 (read_graph_file): Assign ID to BBs.
21799 (output_branch_count): Display BB # if verbose flag is set.
21800 (output_lines): Likewise for arcs.
21801 (print_usage): Add '--verbose' option help.
21802 * doc/gcov.texi: Document --verbose (-w) option.
21804 2017-04-28 Martin Liska <mliska@suse.cz>
21806 * gcov.c (struct block_location_info): New struct.
21807 (process_file): Fill up the new structure.
21808 (read_graph_file): Replace usage of encoding by the newly added
21810 (add_line_counts): Likewise.
21811 (accumulate_line_counts): Remove usage of the union.
21812 (function_info::function_info): New function.
21813 (function_info::~function_info): Likewise.
21814 (process_file): Call delete instead of release_function.
21815 (release_function): Release the function.
21816 (release_structures): Call delete instead of release_function.
21817 (solve_flow_graph): Replace usage of num_blocks.
21818 (find_exception_blocks): Likewise.
21819 (output_lines): Fix GNU coding style.
21821 2017-04-28 Martin Liska <mliska@suse.cz>
21824 * coverage.c (coverage_remove_note_file): New function.
21825 * coverage.h: Declare the function.
21826 * toplev.c (finalize): Clean if an error has been seen.
21828 2017-04-28 Martin Liska <mliska@suse.cz>
21830 PR gcov-profile/80031
21831 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
21832 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
21833 * gcov.c (read_graph_file): Read just number of blocks.
21834 * profile.c (branch_prob): Do not stream 0 flags per a basic
21837 2017-04-28 Martin Liska <mliska@suse.cz>
21839 * gcov-dump.c (tag_*): Add new argument to declarations.
21840 (dump_gcov_file): Likewise.
21841 (tag_blocks): Add and use new argument depth.
21842 (tag_arcs): Likewise.
21843 (tag_lines): Likewise.
21844 (tag_counters): Likewise.
21845 (tag_summary): Likewise.
21846 (dump_working_sets): Use depth to do a proper indentation.
21848 2017-04-28 Jakub Jelinek <jakub@redhat.com>
21851 * cgraph.h (symtab_node::debug_symtab): No longer inline.
21852 * symtab.c (symtab_node::debug_symtab): Move definition here.
21854 2017-04-28 Richard Biener <rguenther@suse.de>
21856 * lto-streamer.h (LTO_major_version): Bump to 7.
21858 2017-04-28 Richard Biener <rguenther@suse.de>
21860 * tree-vrp.c (assert_info): New struct.
21861 (add_assert_info): New helper.
21862 (register_edge_assert_for_2): Refactor to add asserts to a vector
21864 (register_edge_assert_for_1): Likewise.
21865 (register_edge_assert_for): Likewise.
21866 (finish_register_edge_assert_for): New helper actually registering
21867 asserts where live on edge.
21868 (find_conditional_asserts): Adjust.
21869 (find_switch_asserts): Likewise.
21870 (evrp_dom_walker::try_find_new_range): Generalize.
21871 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
21873 2017-04-27 Marek Polacek <polacek@redhat.com>
21876 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
21877 arg10 and arg11 to itype.
21879 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
21881 * doc/extend.texi (Object Size Checking): Improve grammar.
21883 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
21886 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
21887 that the logic for permitting reciprocal estimates matches that
21890 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21893 * tree.c (type_cache_hasher::equal): Only compare
21894 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
21895 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
21896 non-aggregate element types.
21897 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
21898 about the flag on ARRAY_TYPEs in the comment, formatting fix.
21900 2017-04-27 Richard Biener <rguenther@suse.de>
21902 PR middle-end/80533
21903 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
21904 stripping ARRAY_REFs from MEM_EXPR make sure we're not
21905 keeping a reference to a trailing array.
21907 2017-04-27 Richard Biener <rguenther@suse.de>
21909 PR middle-end/80539
21910 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
21911 being in loop-closed SSA form conservatively.
21912 (chrec_fold_multiply_poly_poly): Likewise.
21914 2017-04-27 Tamar Christina <tamar.christina@arm.com>
21916 PR middle-end/79665
21917 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
21918 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
21920 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21923 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
21924 (aarch64_function_arg_alignment): Return unsigned int again, but still
21925 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
21926 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21927 Don't emit -Wpsabi note.
21928 (aarch64_function_arg_boundary): Likewise.
21929 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21932 2017-04-26 Nathan Sidwell <nathan@acm.org>
21934 * tree.h (crc32_unsigned_n): Declare.
21935 (crc32_unsigned, crc32_unsigned): Make inline.
21936 * tree.c (crc32_unsigned_bits): Replace with ...
21937 (crc32_unsigned_n): ... this.
21938 (crc32_unsigned, crc32_byte): Remove.
21939 (crc32_string): Remove unnecessary braces.
21941 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21943 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
21944 * ipa-inline-analysis.c (MAX_TIME): Remove.
21945 (account_size_time): Use sreal for time.
21946 (dump_inline_summary): Update.
21947 (estimate_function_body_sizes): Update.
21948 (estimate_edge_size_and_time): Update.
21949 (estimate_calls_size_and_time): Update.
21950 (estimate_node_size_and_time): Update.
21951 (inline_merge_summary): Update.
21952 (inline_update_overall_summary): Update.
21953 (estimate_time_after_inlining): Update.
21954 (inline_read_section): Update.
21955 (inline_write_summary): Update.
21956 * ipa-inline.c (compute_uninlined_call_time): Update.
21957 (compute_inlined_call_time): Update.
21958 (recursive_inlining): Update.
21959 (inline_small_functions): Update.
21960 (dump_overall_stats): Update.
21961 * ipa-inline.h: Include sreal.h.
21962 (size_time_entry): Turn time to sreal.
21963 (inline_summary): Turn self_time nad time to sreal.
21965 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21967 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
21969 (sreal::stream_out, sreal::stream_in): New.
21970 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
21972 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21974 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
21977 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
21980 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
21982 (dimode_scalar_chain::compute_convert_gain): Ditto.
21983 (dimode_scalar_chain::make_vector_copies): Ditto.
21984 (dimode_scalar_chain::convert_reg): Ditto.
21985 (dimode_scalar_chain::convert_insn): Ditto.
21986 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
21987 (VI248_AVX512BW_1): New mode iterator.
21988 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
21989 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
21992 2017-04-25 Martin Sebor <msebor@redhat.com>
21994 PR tree-optimization/80497
21995 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
21996 constants are representable in HOST_WIDE_INT.
21997 (parse_directive): Ditto.
21999 2017-04-25 Martin Sebor <msebor@redhat.com>
22002 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
22003 (new_zero_array): Adjust signature.
22004 (dom_info::dom_init): Used unsigned rather that size_t.
22005 (dom_info::dom_info): Same.
22007 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22008 Jakub Jelinek <jakub@redhat.com>
22011 * config/arm/arm.c: Include gimple.h.
22012 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22013 returns negative, increment ncrn only if it returned positive.
22014 (arm_needs_doubleword_align): Return int instead of bool,
22015 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
22016 members, but if there is any such non-FIELD_DECL
22017 > PARM_BOUNDARY aligned decl, return -1 instead of false.
22018 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22019 returns negative, increment nregs only if it returned positive.
22020 (arm_setup_incoming_varargs): Likewise.
22021 (arm_function_arg_boundary): Emit -Wpsabi note if
22022 arm_needs_doubleword_align returns negative, return
22023 DOUBLEWORD_ALIGNMENT only if it returned positive.
22025 2017-04-25 Marek Polacek <polacek@redhat.com>
22028 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22029 first argument to type.
22031 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
22034 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
22035 type checks to test for compatibility instead of equality.
22037 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22038 Jakub Jelinek <jakub@redhat.com>
22041 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
22043 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
22044 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
22045 the alignment computation, but return their maximum in warn_alignment.
22046 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22047 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
22049 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
22050 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22053 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22055 * config/arc/simdext.md (dmpyh): Fix typo.
22057 2017-04-25 Richard Biener <rguenther@suse.de>
22059 PR tree-optimization/80492
22060 * alias.c (compare_base_decls): Handle registers with asm
22061 specification conservatively.
22062 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22063 compare_base_decls returning dont-know properly.
22065 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22067 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
22068 (legitimate_offset_address_p): New function.
22069 (arc_legitimate_address_p): Use above function.
22071 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22073 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
22075 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22077 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
22078 ACCH registers whenever they are available.
22080 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22082 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
22083 double regs fix when not used.
22085 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22087 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
22089 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
22090 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
22092 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22094 * config/arc/arc.c (arc_output_addsi): Check for h-register class
22095 when emitting short ADD instructions.
22097 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22099 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
22101 (cmpsi_cc_c_insn): Likewise.
22102 (cbranchsi4_scratch): Compute proper instruction length using
22103 compact_hreg_operand.
22104 * config/arc/predicates.md (compact_hreg_operand): New predicate.
22106 2017-04-25 Richard Biener <rguenther@suse.de>
22108 PR middle-end/80509
22109 * passes.c (pass_manager::pass_manager): Initialize
22110 m_name_to_pass_map.
22112 2017-04-25 Richard Biener <rguenther@suse.de>
22114 PR tree-optimization/79201
22115 * tree-ssa-sink.c (statement_sink_location): Handle calls.
22117 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22120 * config/s390/vector.md: Split MEM->GPR vector moves for
22121 non-s_operand addresses.
22123 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22126 * config/s390/predicates.md (reload_const_wide_int_operand): New
22128 * config/s390/s390.md ("movti"): Remove d/P alternative.
22129 ("movti_bigconst"): New pattern definition.
22131 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22134 * s390-protos.h (s390_expand_cs_hqi): Removed.
22135 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
22136 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
22137 modes as well as CCZ1mode and CCZmode.
22138 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
22139 signature of s390_emit_compare_and_swap.
22140 (s390_expand_cs_hqi): Likewise, make static.
22141 (s390_expand_cs_tdsi): Generate an explicit compare before trying
22142 compare-and-swap, in some cases.
22143 (s390_expand_cs): Wrapper function.
22144 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
22146 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
22147 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
22148 patterns for small and large integers. Forbid symref memory operands.
22149 Move expander to s390.c. Require cc register.
22150 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
22151 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
22152 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
22153 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
22154 symref memory operands. Remove CC mode and call s390_match_ccmode
22156 ("atomic_exchange<mode>"): Allow and implement all integer modes.
22158 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22160 * config/s390/s390.md (define_peephole2): New peephole to help
22161 combining the load-and-test pattern with volatile memory.
22163 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22165 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
22166 with CCZmode for TARGET_Z196.
22168 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22170 PR rtl-optimization/80501
22171 * combine.c (make_compound_operation_int): Set subreg_code to SET
22172 even for AND with mask of the sign bit of mode.
22174 PR rtl-optimization/80500
22175 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
22176 sum's initial value.
22178 2017-04-25 Julian Brown <julian@codesourcery.com>
22179 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22181 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
22183 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
22185 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
22187 2017-04-25 Julian Brown <julian@codesourcery.com>
22188 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22190 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
22191 (thunderx2t99_sha): New Reservation.
22193 2017-04-25 Julian Brown <julian@codesourcery.com>
22194 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22196 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
22197 type for 1-element load.
22199 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
22201 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
22203 2017-04-24 Martin Jambor <mjambor@suse.cz>
22205 PR tree-optimization/80293
22206 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
22207 char arrays not totally scalarizable if it is false.
22208 (analyze_all_variable_accesses): Pass correct value in the new
22209 parameter. Add a statistics counter.
22211 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
22213 PR middle-end/79931
22214 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
22216 2017-04-24 Richard Biener <rguenther@suse.de>
22218 PR tree-optimization/80494
22219 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
22220 out for complex types.
22222 2017-04-24 Richard Biener <rguenther@suse.de>
22224 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
22225 * tree-ssa-sccvn.c (print_scc): Print SCC size.
22226 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
22227 (DFS): Adjust and never fail.
22228 (sccvn_dom_walker::fail): Remove.
22229 (sccvn_dom_walker::before_dom_children): Adjust.
22230 (run_scc_vn): Likewise and never fail.
22231 * tree-ssa-pre.c (pass_pre::execute): Adjust.
22232 (pass_fre::execute): Likewise.
22234 2017-04-24 Richard Biener <rguenther@suse.de>
22236 PR tree-optimization/79725
22237 * tree-ssa-sink.c (statement_sink_location): Return whether
22238 failure reason was zero uses. Move that check later.
22239 (sink_code_in_bb): Deal with zero uses by removing the stmt
22242 2017-04-24 Richard Biener <rguenther@suse.de>
22245 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
22246 pointer-based references.
22248 2017-04-24 Richard Biener <rguenther@suse.de>
22251 * pass_manager.h (pass_manager::operator new): Remove.
22252 (pass_manager::operator delete): Likewise.
22253 * passes.c (pass_manager::operator new): Remove.
22254 (pass_manager::operator delete): Likewise.
22255 (pass_manager::pass_manager): Zero individual pass members.
22257 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
22260 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
22261 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
22262 Check "XEXP (src, 1)" operand here.
22263 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
22264 Check "XEXP (src, 1)" operand here.
22265 (dimode_scalar_chain::make_vector_copies): Detect count register
22266 of a shift instruction. Zero extend count register from QImode
22267 to DImode to satisfy vector shift pattern count operand predicate.
22268 Substitute vector shift count operand with a DImode copy.
22269 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
22272 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
22274 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
22275 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22276 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
22277 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22278 (UNSPEC_NOREX_MEM): Remove definition.
22280 2017-04-21 Richard Biener <rguenther@suse.de>
22282 PR tree-optimization/79547
22283 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22284 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
22285 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
22286 without any constraints.
22288 2017-04-21 Richard Biener <rguenther@suse.de>
22290 PR tree-optimization/78847
22291 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
22293 2017-04-21 Richard Biener <rguenther@suse.de>
22295 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
22296 (build_distinct_type_copy): Likewise.
22297 (build_variant_type_copy): Likewise.
22298 * tree.c (build_qualified_type): Pass down mem-stat info.
22299 (build_distinct_type_copy): Likewise.
22300 (build_variant_type_copy): Likewise.
22302 2017-04-21 Richard Biener <rguenther@suse.de>
22304 PR tree-optimization/80237
22305 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
22307 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
22308 for a simplified result.
22310 2016-04-21 Richard Biener <rguenther@suse.de>
22312 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
22313 sth as strict as a simple_iv but a chrec without symbols and an
22314 operand defined in the loop we are peeling (and not some subloop).
22315 (propagate_constants_for_unrolling): Propagate all constants.
22317 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22320 * config/i386/i386.c (print_reg): Remove assert for disalowed
22321 regno values, call output_operand_lossage instead.
22323 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22326 * config/i386/constraints.md (Yc): New register constraint.
22327 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
22328 Use Yc constraint for alternative 2 of operand 0. Remove
22329 preferred_for_speed attribute.
22331 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22333 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
22334 lastprivate clauses in SIMT case.
22336 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
22338 * doc/invoke.texi (-Wextra-semi): Document new warning option.
22340 2017-04-20 Richard Biener <rguenther@suse.de>
22342 PR tree-optimization/57796
22343 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
22344 as N scalar stores.
22345 (vect_model_load_cost): Cost gathers as N scalar loads.
22347 2017-04-20 Richard Biener <rguenther@suse.de>
22349 * ggc-page.c (ggc_allocated_p): Rename to ...
22350 (safe_lookup_page_table_entry): ... this and return the lookup
22352 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
22354 2017-04-20 Richard Biener <rguenther@suse.de>
22356 PR tree-optimization/80453
22357 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
22358 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
22359 from the conditions.
22360 (vn_phi_eq): Pass them down.
22361 (vn_phi_lookup): Record them.
22362 (vn_phi_insert): Likewise.
22364 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
22366 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
22367 uninitialized variable warning to avoid buffer overrun.
22369 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22372 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
22373 is suppressed for '{ 0 }' in C.
22375 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22377 * BASE-VER: Set to 8.0.0.
22379 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
22381 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
22382 priority .init_array and .fini_array section with SECTION_NOTYPE
22385 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22387 PR middle-end/80423
22388 * tree.h (build_array_type): Add typeless_storage default argument.
22389 * tree.c (type_cache_hasher::equal): Also compare
22390 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
22391 (build_array_type): Add typeless_storage argument, set
22392 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
22394 (build_nonshared_array_type): Adjust build_array_type_1 caller.
22395 (build_array_type): Likewise. Add typeless_storage argument.
22397 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22398 Jakub Jelinek <jakub@redhat.com>
22400 PR tree-optimization/80426
22401 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
22402 operation on symbolic operands, also compute the overflow for the
22403 invariant part when the operation degenerates into a negation.
22405 2017-04-19 Jakub Jelinek <jakub@redhat.com>
22408 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
22409 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
22412 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
22414 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
22417 * config/avr/avr.c (tree.h): Include it.
22418 (cgraph.h): Include it.
22419 (avr_encode_section_info): Don't warn for uninitialized progmem
22420 variable if it's just an alias.
22422 2017-04-19 Richard Biener <rguenther@suse.de>
22425 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
22426 when needed by AutoPGO.
22428 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
22431 * doc/lto.texi: Remove an extra 'that'.
22433 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
22435 PR rtl-optimization/80429
22436 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
22437 are only used in debug insns.
22439 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22440 Vladimir Makarov <vmakarov@redhat.com>
22442 * config/sparc/predicates.md (input_operand): Add comment. Return
22443 true for any memory operand when LRA is in progress.
22444 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
22446 2017-04-18 Jeff Law <law@redhat.com>
22449 * mips.md ({return,simple_return}_internal): Do not overwrite
22452 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22454 PR tree-optimization/80443
22455 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
22456 instead of adding 1, subtract -1 and similarly instead of subtracting
22459 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
22461 PR rtl-optimization/80357
22462 * haifa-sched.c (tmp_bitmap): New variable.
22463 (model_recompute): Handle duplicate use records.
22464 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
22465 (free_global_sched_pressure_data): Free it.
22467 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
22470 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22471 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22472 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22473 instead of SYSTEM_HEADER_DIR.
22475 2017-04-18 Jeff Law <law@redhat.com>
22477 PR middle-end/80422
22478 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
22479 predecessors after walking up the insn chain.
22481 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22484 * dwarf2out.c (modified_type_die): Try harder not to emit internal
22485 sizetype type into debug info.
22487 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22490 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
22491 unneeded test for TARGET_UPPER_REGS_SF.
22492 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
22494 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22497 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
22498 instead of gsi_after_labels.
22500 2017-04-18 Jeff Law <law@redhat.com>
22502 * regcprop.c (maybe_mode_change): Avoid creating copies of the
22506 2017-04-13 Jeff Law <law@redhat.com>
22507 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22508 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22510 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
22513 * config/avr/avr.c (intl.h): Include it.
22514 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
22516 2017-04-18 Martin Liska <mliska@suse.cz>
22518 PR gcov-profile/78783
22519 * gcov-tool.c (gcov_output_files): Validate that destination
22520 file is either removed by the tool or by a user.
22522 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22523 Guy Benyei <guybe@mellanox.com>
22525 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
22526 block, and do not negate it, the stored id is already negative.
22528 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22530 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
22532 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22535 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
22536 masks of options that should be turned off if the VSX vector
22537 options are turned off.
22538 (OTHER_P8_VECTOR_MASKS): Likewise.
22539 (OTHER_VSX_VECTOR_MASKS): Likewise.
22540 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
22541 rs6000_disable_incompatible_switches to validate no type switches
22543 (rs6000_incompatible_switch): New function to disallow turning on
22544 other vector options if -mno-vsx, -mno-power8-vector, or
22545 -mno-power9-vector are specified.
22547 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22549 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
22551 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22553 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
22554 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
22555 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
22556 (ARG_POINTER_CFA_OFFSET): Likewise.
22558 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22560 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
22561 conditions to take advantage of various optimizations.
22563 2017-04-13 Jeff Law <law@redhat.com>
22565 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22566 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22567 (zero_extendsidi2_dext): Likewise.
22569 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22572 * fold-const.c (fold_ternary_loc): Revert
22573 use op0 instead of fold_convert_loc (loc, type, arg0) part of
22576 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
22578 PR rtl-optimization/80343
22579 * lra-remat.c (update_scratch_ops): Assign original hard reg to
22580 new scratch pseudo.
22582 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
22585 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
22586 to ubsan_encode_value.
22588 2017-04-13 Jeff Law <law@redhat.com>
22590 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
22591 appearing in DEBUG_INSNs.
22593 2017-04-13 Martin Liska <mliska@suse.cz>
22595 PR gcov-profile/80413
22596 * gcov-io.c (gcov_write_string): Copy to buffer just when
22597 allocated size is greater than zero.
22599 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22602 * dwarf2out.c (decls_for_scope): Ignore declarations of
22603 current_function_decl in BLOCK_NONLOCALIZED_VARS.
22605 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
22608 * ipa-visibility.c (non_local_p): Fix typos.
22609 (localize_node): When localizing symbol in same comdat group,
22610 dissolve the group only when we know external symbols are going
22612 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
22614 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22616 PR tree-optimization/79390
22617 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
22618 order does not result in usable sequence, retry with reversed operand
22624 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
22625 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
22626 op0 instead of fold_convert_loc (loc, type, arg0).
22628 2017-04-12 Jeff Law <law@redhat.com>
22630 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
22631 has a delay slot in the generated code.
22633 * config/cris/cris.md (cris_preferred_reload_class): Return
22634 GENNONACR_REGS rather than GENERAL_REGS.
22636 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22639 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
22640 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
22641 signedness of the result type.
22643 2017-04-12 Richard Biener <rguenther@suse.de>
22644 Jeff Law <law@redhat.com>
22646 PR tree-optimization/80359
22647 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
22648 trim stores to TARGET_MEM_REFs.
22650 2017-04-12 Richard Biener <rguenther@suse.de>
22652 PR tree-optimization/79390
22653 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
22654 threading case even more.
22656 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
22659 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
22660 for quad_address_p for TImode, instead of just not indexed_address.
22662 2017-04-12 Richard Biener <rguenther@suse.de>
22663 Bernd Edlinger <bernd.edlinger@hotmail.de>
22665 PR middle-end/79671
22666 * alias.c (component_uses_parent_alias_set_from): Handle
22667 TYPE_TYPELESS_STORAGE.
22668 (get_alias_set): Likewise.
22669 * tree-core.h (tree_type_common): Add typeless_storage flag.
22670 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
22671 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
22672 for types containing members with TYPE_TYPELESS_STORAGE.
22673 (place_field): Likewise.
22674 (layout_type): Likewise for ARRAY_TYPE.
22675 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
22676 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
22677 TYPE_TYPELESS_STORAGE.
22678 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
22680 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22683 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22684 first argument to type.
22686 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22690 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
22691 CONST0_RTX (mode) rather than const0_rtx where appropriate.
22692 (rs6000_expand_binop_builtin): Likewise.
22693 (rs6000_expand_ternop_builtin): Likewise; also add missing
22694 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
22695 vshasigma built-ins.
22696 * doc/extend.texi: Document that vec_xxpermdi's third argument
22697 must be a constant.
22699 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
22701 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
22702 Use shift_const cost parameter when calculating gain of STV shifts.
22704 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
22706 PR rtl-optimization/70478
22707 * lra-constraints.c (process_alt_operands): Check memory for
22708 disfavoring memory insn operand.
22710 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22712 PR middle-end/80100
22713 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
22714 left shift in unsigned HOST_WIDE_INT type.
22716 PR rtl-optimization/80385
22717 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
22718 (not (neg X)) into (plus X -1) for complex or non-integral modes.
22721 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
22722 if they have any depend clauses.
22724 2017-04-11 Martin Liska <mliska@suse.cz>
22727 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
22728 * ipa-split.c (split_function): Create a local comdat symbol
22729 if caller is in a comdat group.
22731 2017-04-11 Martin Liska <mliska@suse.cz>
22734 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
22737 2017-04-11 Martin Sebor <msebor@redhat.com>
22739 PR middle-end/80364
22740 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
22741 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
22743 (directive::set_width, directive::set_precision, format_character):
22745 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
22748 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
22751 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
22752 conflict, set target->arch_name instead of target->cpu_name.
22754 2017-04-11 Richard Biener <rguenther@suse.de>
22756 PR tree-optimization/80374
22757 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
22758 build_zero_cst, remove fold_convertible_p check again.
22760 2017-04-11 Martin Liska <mliska@suse.cz>
22763 * ubsan.c (instrument_object_size): Do not instrument register
22766 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22769 * config/i386/i386-builtin-types.def
22770 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
22771 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
22772 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
22773 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
22774 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
22775 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
22776 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
22777 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
22778 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
22779 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
22780 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
22781 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
22782 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
22783 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
22784 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
22785 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
22786 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
22787 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
22788 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
22789 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
22790 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
22791 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
22792 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
22793 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
22794 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
22795 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
22796 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
22797 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
22798 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
22799 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
22800 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
22801 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
22802 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
22803 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
22804 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
22805 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
22806 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
22807 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
22808 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
22809 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
22810 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
22811 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
22812 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
22813 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
22814 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
22815 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
22817 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
22818 flag to second_arg_count, handle 4 argument function type _COUNT
22819 aliases, handle second_arg_count on second argument rather than last.
22821 2017-04-10 Jeff Law <law@redhat.com>
22823 PR tree-optimization/80374
22824 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
22825 record anything if we can not convert integer_zero_node to the
22828 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22831 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22832 Enhance special handling given to the TARGET_P9_MINMAX option in
22833 relation to certain other options.
22835 2017-04-10 Bin Cheng <bin.cheng@arm.com>
22837 PR tree-optimization/80153
22838 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
22839 remove POINTER_PLUS_EXPR's base part directly, rather than through
22842 2017-04-10 Richard Biener <rguenther@suse.de>
22843 Bin Cheng <bin.cheng@arm.com>
22845 PR tree-optimization/80153
22846 * tree-affine.c (aff_combination_to_tree): Get base pointer from
22847 the first element of pointer type aff_tree. Build result expr in
22849 (add_elt_to_tree): Convert to type unconditionally. Remove other
22850 fold_convert calls.
22851 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
22852 (rewrite_use_nonlinear_expr): Check invariant using iv information.
22854 2017-04-10 Richard Biener <rguenther@suse.de>
22856 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22859 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
22861 PR rtl-optimization/70478
22862 * lra-constraints.c (curr_small_class_check): New.
22863 (update_and_check_small_class_inputs): New.
22864 (process_alt_operands): Update curr_small_class_check. Disfavor
22865 alternative insn memory operands. Check available regs for small
22868 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22871 * config/mips/mips.opt (-mvirt): Update description.
22872 * doc/invoke.texi (-mvirt): Likewise.
22874 2017-04-10 Richard Biener <rguenther@suse.de>
22876 PR middle-end/80362
22877 * fold-const.c (fold_binary_loc): Look at unstripped ops when
22878 looking for NEGATE_EXPR in -A / -B to A / B folding.
22880 2017-04-10 Martin Liska <mliska@suse.cz>
22882 PR gcov-profile/80224
22883 * gcov.c (print_usage): Fix usage string.
22884 (get_gcov_intermediate_filename): Remove.
22885 (output_gcov_file): Use both for normal and intermediate format.
22886 (generate_results): Do not initialize special file for
22887 intermediate format.
22889 2017-04-10 Richard Biener <rguenther@suse.de>
22891 PR tree-optimization/80304
22892 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
22895 2017-04-10 Nathan Sidwell <nathan@acm.org>
22898 * config/rs6000/rs6000.c (rs6000_vector_type): New.
22899 (rs6000_init_builtins): Use it.
22901 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22903 * config/arm/arm.md (<mrc>): Add mode to SET source.
22904 (<mrrc>): Likewise.
22906 2017-04-10 Richard Biener <rguenther@suse.de>
22908 PR middle-end/80344
22909 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
22911 2017-04-10 Jakub Jelinek <jakub@redhat.com>
22914 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
22915 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
22916 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
22917 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
22918 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
22919 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
22920 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
22921 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
22922 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
22923 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
22924 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
22925 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
22926 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
22927 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
22928 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
22929 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
22930 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
22931 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
22932 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
22933 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
22934 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
22935 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
22936 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
22938 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
22940 PR rtl-optimization/70478
22941 * lra-constraints.c: Reverse the last patch.
22943 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
22945 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
22946 Add comment for WCHAR_T.
22948 2017-04-08 Martin Liska <mliska@suse.cz>
22951 2017-04-07 Martin Liska <mliska@suse.cz>
22954 * ipa-split.c (split_function): Add function part to a same comdat
22957 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22960 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
22962 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
22964 * rs6000/rs6000.c (vec_load_pendulum): Rename...
22965 (vec_pairing): ...to this.
22966 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
22967 (rs6000_sched_init): Adjust for name change.
22968 (struct rs6000_sched_context): Likewise.
22969 (rs6000_init_sched_context): Likewise.
22970 (rs6000_set_sched_context): Likewise.
22972 2017-04-07 Jakub Jelinek <jakub@redhat.com>
22978 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
22980 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
22981 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
22982 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
22984 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
22986 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
22988 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22990 PR rtl-optimization/70703
22991 * ira-color.c (update_conflict_hard_regno_costs): Use
22992 int64_t instead of HOST_WIDE_INT.
22994 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22996 PR rtl-optimization/70478
22997 * lra-constraints.c (process_alt_operands): Disfavor alternative
22998 insn memory operands.
23000 2017-04-07 Jeff Law <law@redhat.com>
23002 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
23003 CALL and NOTE_INSN_CALL_ARG_LOCATION.
23005 2017-04-07 Martin Liska <mliska@suse.cz>
23008 * config/aarch64/aarch64.c (aarch64_process_target_attr):
23009 Show error message instead of an ICE.
23011 2017-04-07 Martin Liska <mliska@suse.cz>
23014 * ipa-split.c (split_function): Add function part to a same comdat
23017 2017-04-07 Richard Biener <rguenther@suse.de>
23019 PR middle-end/80341
23020 * tree.c (get_unwidened): Also handle ! for_type case for
23022 * convert.c (do_narrow): Split out from ...
23023 (convert_to_integer_1): ... here. Do not pass final truncation
23024 type to get_unwidened for TRUNC_DIV_EXPR.
23026 2017-04-07 Richard Biener <rguenther@suse.de>
23028 * tree-affine.c (wide_int_ext_for_comb): Take type rather
23030 (aff_combination_const): Adjust.
23031 (aff_combination_scale): Likewise.
23032 (aff_combination_add_elt): Likewise.
23033 (aff_combination_add_cst): Likewise.
23034 (aff_combination_convert): Likewise.
23035 (add_elt_to_tree): Likewise. Remove unused argument.
23036 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
23038 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
23040 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
23042 * config/arm/arm.c (arm_default_short_enums): Use
23043 ARM_DEFAULT_SHORT_ENUMS.
23044 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
23046 2017-04-06 Jakub Jelinek <jakub@redhat.com>
23049 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
23050 members with redundant out-of-class redeclaration.
23052 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23055 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
23056 * config/i386/i386.md (*zero_extendsidi2):
23057 Add (?*x,*x) and (?*v,*v) alternatives.
23059 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23062 * config/i386/i386.c (ix86_expand_builtin)
23063 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
23064 mode from insn data. Convert operands to insn operand mode.
23065 Copy operands that don't satisfy insn predicate to a register.
23067 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
23069 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
23072 2017-04-06 Richard Biener <rguenther@suse.de>
23074 PR tree-optimization/80334
23075 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
23076 preserve alignment of accesses.
23078 2017-04-06 Richard Biener <rguenther@suse.de>
23080 PR tree-optimization/80262
23081 * tree-sra.c (build_ref_for_offset): Preserve address-space
23083 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
23084 Drop useless address-space information on MEM_REF offsets.
23086 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
23088 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
23090 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23092 PR rtl-optimization/70703
23093 * ira-color.c (update_conflict_hard_regno_costs): Use
23094 HOST_WIDE_INT instead of long.
23096 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
23099 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
23100 not defined for x86_64 target. Add -mmmx target option when __SSE2__
23102 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
23103 for x86_64 target. Handle -m3dnowa option.
23105 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23107 PR rtl-optimization/70703
23108 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
23109 (update_conflict_hard_regno_costs): Use long instead of unsigned
23110 arithmetic for cost calculation.
23112 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23113 Bernd Edlinger <bernd.edlinger@hotmail.de>
23116 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
23119 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
23122 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
23123 ptr_mode with Pmode throughout.
23124 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
23125 into probe_stack_range and use DImode.
23127 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
23130 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
23131 call_eh_return is true.
23133 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23135 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
23136 Initialize last_match_fntype_index.
23138 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23141 * tree-nvr.c: Include internal-fn.h.
23142 (pass_return_slot::execute): Ignore internal calls without
23145 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23146 Richard Biener <rguenther@suse.de>
23149 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
23150 captures used multiple times, except for the last use.
23151 * generic-match-head.c: Include gimplify.h.
23153 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23155 PR tree-optimization/79390
23156 * target.h (struct noce_if_info): Declare.
23157 * targhooks.h (default_noce_conversion_profitable_p): Declare.
23158 * target.def (noce_conversion_profitable_p): New target hook.
23159 * ifcvt.h (struct noce_if_info): New type, moved from ...
23160 * ifcvt.c (struct noce_if_info): ... here.
23161 (noce_conversion_profitable_p): Renamed to ...
23162 (default_noce_conversion_profitable_p): ... this. No longer
23164 (noce_try_store_flag_constants, noce_try_addcc,
23165 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
23166 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
23167 instead of noce_conversion_profitable_p.
23168 * config/i386/i386.c: Include ifcvt.h.
23169 (ix86_option_override_internal): Don't override
23170 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
23171 (ix86_noce_conversion_profitable_p): New function.
23172 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
23173 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
23174 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
23175 * doc/tm.texi: Regenerated.
23177 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23179 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
23182 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23185 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
23186 instructions for small multiply cores.
23188 2017-04-04 Jeff Law <law@redhat.com>
23190 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
23192 (mips_expand_vec_perm_const): Initialize elements in orig_perm
23193 that are not set by the loop over the elements.
23195 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23198 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
23199 int mode, convert_modes it to mode as unsigned, otherwise use
23200 lowpart_subreg to mode rather than SImode.
23201 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
23202 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
23203 Use DImode instead of SImode for the shift count operand.
23204 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
23207 2017-04-04 Richard Biener <rguenther@suse.de>
23209 PR middle-end/80281
23210 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
23211 arithmetic done for the negate or the plus. Simplify.
23212 (A - (-B) -> A + B): Likewise.
23213 * fold-const.c (split_tree): Make sure to not negate pointers.
23215 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
23217 PR rtl-optimization/60818
23218 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
23219 a compare of comparisons with the thing compared if this results
23220 in a different machine mode.
23222 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
23224 * alias.c (base_alias_check): Fix typo in comment.
23225 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
23226 * cgraphunit.c (symbol_table::compile): Likewise.
23227 * collect2.c (maybe_run_lto_and_relink): Likewise.
23228 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
23229 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
23230 * config/avr/avr.c (avr_map_op_t): Likewise.
23231 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
23232 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
23233 * config/epiphany/epiphany.md (movcc): Likewise.
23234 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
23235 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
23237 * config/mips/mips.c (mips_save_restore_reg): Likewise.
23238 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
23239 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
23240 * config/sh/sh.c (sh_rtx_costs): Likewise.
23241 * fold-const.c (fold_truth_andor): Likewise.
23242 * genautomata.c (collapse_flag): Likewise.
23243 * gengtype.h (struct type::u::s): Likewise.
23244 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
23245 * input.c (FORMAT_AMOUNT): Likewise.
23246 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
23247 (known_aggs_to_agg_replacement_list): Likewise.
23248 * ipa-inline-analysis.c: Likewise.
23249 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
23250 * ipa-polymorphic-call.c
23251 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
23252 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
23253 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
23255 * modulo-sched.c (apply_reg_moves): Likewise.
23256 * omp-expand.c (build_omp_regions_1): Likewise.
23257 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
23258 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
23259 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
23260 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
23261 * value-prof.c: Likewise.
23262 * var-tracking.c (val_reset): Likewise.
23264 2017-04-03 Richard Biener <rguenther@suse.de>
23266 PR tree-optimization/80275
23267 * fold-const.c (split_address_to_core_and_offset): Handle
23270 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
23272 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
23273 descriptors is at least equal to that of functions.
23275 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23277 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
23279 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23282 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
23283 (mov<IMOD4:mode>): New expander.
23284 (*mov<IMOD4:mode>_internal): New insn and split pattern.
23286 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
23288 PR rtl-optimization/79405
23289 * fwprop.c (propagations_left): New variable.
23290 (forward_propagate_into): Decrement it.
23291 (fwprop_init): Initialize it.
23292 (fw_prop): If the variable has reached zero, stop propagating.
23293 (fwprop_addr): Ditto.
23295 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23298 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
23299 a FUNCTION_DECL, pass it as decl instead of origin to
23302 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
23304 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
23307 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
23310 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
23311 TARGET_VSX_SMALL_INTEGER.
23313 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23315 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23316 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
23318 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23320 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
23321 extraction from odd-numbered MSA register.
23323 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23325 PR middle-end/80173
23326 * expmed.c (store_bit_field_1): Don't attempt to create
23327 a word subreg out of hard registers wider than word if they
23328 have HARD_REGNO_NREGS of 1 for their mode.
23330 PR middle-end/80163
23331 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
23332 conversions to integer types wider than word and pointer.
23335 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
23336 (rtx_equal_for_cselib_p): Pass 0 to it.
23337 * cselib.c (cselib_hasher::equal): Likewise.
23338 (rtx_equal_for_cselib_1): Add depth argument. If depth
23339 is 128, don't look up VALUE locs and punt. Increment
23340 depth in recursive calls when walking VALUE locs.
23342 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
23344 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
23345 (make_gcov_file_name): Use the canonical path name for generating
23347 (read_line): Fix handling of files with ascii null bytes.
23349 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
23351 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
23352 to initialise a vector register instead
23353 of using a const_int.
23355 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23357 PR translation/80189
23358 * gimplify.c (omp_default_clause): Use %qs instead of %s in
23359 diagnostic messages.
23361 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
23364 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
23365 (dfp_diex_<mode>): Update mode of operand 1.
23366 * doc/extend.texi (dxex, dxexq): Document change to return type.
23367 (diex, diexq): Document change to argument type.
23369 2017-03-30 Martin Jambor <mjambor@suse.cz>
23372 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
23373 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
23374 it reflects the signature changes performed at the callee side.
23375 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
23376 to cgraph_build_function_type_skip_args.
23377 (build_function_decl_skip_args): Adjust call to the above function.
23379 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23382 * config/i386/sse.md
23383 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
23384 register as dest whenever it is a MEM not rtx_equal_p to the
23385 corresponding dup operand, and when forcing into reg move the
23386 reg into the memory afterwards.
23387 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
23388 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
23389 for the force_reg mode.
23390 (avx512vl_vextractf128<mode>): Use register as dest either
23391 always when a MEM, or when it is a MEM not rtx_equal_p to the
23392 corresponding dup operand, or even not when it is a CONST_VECTOR
23393 depending on the mode and lo vs. hi.
23394 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
23396 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
23397 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
23398 Likewise. Require that operands[2] is even.
23399 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
23400 Remove extraneous parens. Require that operands[2] is a multiple
23402 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
23403 operands[0] is a MEM if <mask_applied>, the predicates/constraints
23404 disallow memory then.
23406 2017-03-30 Richard Biener <rguenther@suse.de>
23408 PR tree-optimization/77498
23409 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
23410 to non-constants over backedges.
23412 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
23414 PR rtl-optimization/80233
23415 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
23416 as last_combined_insn. Do not test for BARRIER_P separately.
23418 2017-03-29 Andreas Schwab <schwab@suse.de>
23421 * calls.c (prepare_call_address): Convert funexp to Pmode before
23422 copying to temp reg.
23424 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23426 PR tree-optimization/80158
23427 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
23428 Handle possible future case of more than one alternate
23430 (replace_rhs_if_not_dup): Likewise.
23431 (replace_one_candidate): Likewise.
23433 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
23435 PR rtl-optimization/80193
23436 * ira.c (ira): Do not check allocation for LRA.
23438 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
23440 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
23441 (nvptx_output_simt_exit): Declare.
23442 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
23443 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
23444 (init_softstack_frame): Move initialization of crtl->is_leaf to...
23445 (nvptx_declare_function_name): ...here. Emit declaration of local
23446 memory space buffer for omp_simt_enter insn.
23447 (nvptx_output_unisimt_switch): New.
23448 (nvptx_output_softstack_switch): New.
23449 (nvptx_output_simt_enter): New.
23450 (nvptx_output_simt_exit): New.
23451 * config/nvptx/nvptx.h (struct machine_function): New fields
23452 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
23453 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
23454 (UNSPECV_SIMT_EXIT): Ditto.
23455 (omp_simt_enter_insn): New insn.
23456 (omp_simt_enter): New expansion.
23457 (omp_simt_exit): New insn.
23458 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
23460 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
23461 (expand_GOMP_SIMT_ENTER_ALLOC): New.
23462 (expand_GOMP_SIMT_EXIT): New.
23463 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
23464 (GOMP_SIMT_ENTER_ALLOC): Ditto.
23465 (GOMP_SIMT_EXIT): Ditto.
23466 * target-insns.def (omp_simt_enter): New insn.
23467 (omp_simt_exit): Ditto.
23468 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
23470 (lower_rec_simd_input_clauses): Implement SIMT privatization.
23471 (lower_rec_input_clauses): Likewise.
23472 (lower_lastprivate_clauses): Handle SIMT privatization.
23474 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
23475 (ompdevlow_adjust_simt_enter): New.
23476 (find_simtpriv_var_op): New.
23477 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
23478 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
23480 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
23481 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
23482 (copy_decl_for_dup_finish): Ditto.
23484 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
23486 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
23489 * config/i386/i386.c (ix86_option_override_internal): Always
23490 allow -mpreferred-stack-boundary=3 for 64-bit targets.
23492 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23494 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
23496 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23498 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
23499 mark new edge's irreducible flag accordign to it.
23500 (vect_do_peeling): Check loop preheader edge's irreducible flag
23501 and pass it to function slpeel_add_loop_guard.
23503 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
23505 PR tree-optimization/80218
23506 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23507 Update block frequencies and counts.
23509 2017-03-28 Richard Biener <rguenther@suse.de>
23511 PR tree-optimization/78644
23512 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
23513 of a simplification result we may not use it at all.
23515 2017-03-28 Richard Biener <rguenther@suse.de>
23518 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
23519 without arguments, generate default definition of a SSA name.
23521 2017-03-28 Richard Biener <rguenther@suse.de>
23523 PR middle-end/80222
23524 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
23525 TYPE_REF_CAN_ALIAS_ALL references.
23526 * fold-const.c (fold_indirect_ref_1): Likewise.
23528 2017-03-28 Martin Liska <mliska@suse.cz>
23531 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
23532 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
23534 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23535 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
23537 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
23538 (EXTRA_SPECS): Define.
23539 (SUBTARGET_EXTRA_SPECS): Likewise.
23540 (SUBTARGET_CPP_SPEC): Likewise.
23541 * config/arc/elf.h (EXTRA_SPECS): Renamed to
23542 SUBTARGET_EXTRA_SPECS.
23543 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
23545 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23547 * config/arc/simdext.md (vst64_insn): Update pattern.
23548 (vld32wh_insn): Likewise.
23549 (vld32wl_insn): Likewise.
23550 (vld64_insn): Likewise.
23551 (vld32_insn): Likewise.
23553 2017-03-28 Marek Polacek <polacek@redhat.com>
23556 * fold-const.c (fold_comparison): Use protected_set_expr_location
23557 instead of SET_EXPR_LOCATION.
23559 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
23561 * tree.c (add_expr): Avoid name lookup warning.
23563 2017-03-27 Jeff Law <law@redhat.com>
23565 PR tree-optimization/80216
23566 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
23567 function name. Limit recursion depth.
23568 (record_temporary_equivalences): Corresponding changes.
23570 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
23572 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
23575 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23578 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
23580 * cfgcleanup.c (reg_note_cfa_p): New array.
23581 (insns_have_identical_cfa_notes): New function.
23582 (old_insns_match_p): Don't cross-jump in between /f
23583 and non-/f instructions. If both i1 and i2 are frame related,
23584 verify all CFA notes, their order and content.
23586 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
23589 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
23590 HImode and SImode with zero extend to DImode to one insn.
23591 (bswap<mode>2_extenddi): Likewise.
23592 (bswapsi2_extenddi): Likewise.
23593 (bswaphi2_extendsi): Likewise.
23594 (bswaphi2): Combine bswap HImode and SImode into one insn.
23595 Separate memory insns from swapping register.
23596 (bswapsi2): Likewise.
23597 (bswap<mode>2): Likewise.
23598 (bswaphi2_internal): Delete, no longer used.
23599 (bswapsi2_internal): Likewise.
23600 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
23601 store, and gpr<-gpr swap insns.
23602 (bswap<mode>2_store): Likewise.
23603 (bswaphi2_reg): Register only splitter, combine with the splitter.
23604 (bswaphi2 splitter): Likewise.
23605 (bswapsi2_reg): Likewise.
23606 (bswapsi2 splitter): Likewise.
23607 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
23608 the insns into load, store, and register/register insns.
23609 (bswapdi2_ldbrx): Likewise.
23610 (bswapdi2_load): Likewise.
23611 (bswapdi2_store): Likewise.
23612 (bswapdi2_reg): Likewise.
23614 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
23616 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
23617 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
23619 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
23622 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
23624 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23625 special handling for target option conflicts between dform
23626 options (-mpower9-dform, -mpower9-dform-vector,
23627 -mpower9-dform-scalar) and -mno-direct-move.
23629 2017-03-27 Richard Biener <rguenther@suse.de>
23631 PR tree-optimization/80181
23632 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
23634 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23636 * config/arc/predicates.md (move_double_src_operand): Replace the
23637 call to move_double_src_operand with a call to address_operand.
23639 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23641 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
23642 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
23643 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
23645 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23647 * config/arc/predicates.md (long_immediate_loadstore_operand):
23648 Consider scaled addresses cases.
23650 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23652 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
23653 restored when in interrupt.
23654 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
23655 doesn't have delay slot.
23657 2017-03-27 Richard Biener <rguenther@suse.de>
23660 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
23661 inlined thunk clones.
23663 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23666 * asan.c (instrument_derefs): Copy over last operand from
23667 original COMPONENT_REF to the new COMPONENT_REF with
23668 DECL_BIT_FIELD_REPRESENTATIVE.
23669 * ubsan.c (instrument_object_size): Likewise.
23671 2017-03-27 Richard Biener <rguenther@suse.de>
23673 PR tree-optimization/80170
23674 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
23675 sure DR/SCEV didnt fold in constants we do not see when looking
23676 at the reference base alignment.
23678 2017-03-27 Richard Biener <rguenther@suse.de>
23680 PR middle-end/80171
23681 * gimple-fold.c (fold_ctor_reference): Properly guard against
23682 NULL return value from canonicalize_constructor_val.
23684 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
23687 * config/i386/i386.c (ix86_expand_builtin)
23688 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
23689 flags reg setting and flags reg using instructions.
23690 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
23691 clobbering instructions to zero extend op2.
23693 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
23695 * doc/install.texi (Configuration) <--with-aix-soname>:
23696 Update link to AIX ld.
23698 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
23700 PR rtl-optimization/80160
23701 PR rtl-optimization/80159
23702 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
23703 reg_alternate_class into account.
23705 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
23708 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
23709 to consider in curr_insn_transform.
23711 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23713 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
23714 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
23715 and emit_mode_inner.
23717 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23719 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
23720 argument to the overloaded builtin variants. Use the new flag to
23721 deprecate certain builtin variants.
23722 * config/s390/s390-builtin-types.def: Add new builtin types.
23723 * config/s390/s390-builtins.h: Support new flags field for
23724 overloaded builtins.
23725 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
23726 (s390_macro_to_expand): Enable vector float data type.
23727 (s390_cpu_cpp_builtins_internal): Indicate support of the new
23728 builtins by incrementing the __VEC__ version number.
23729 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
23731 (s390_resolve_overloaded_builtin): Emit error messages depending
23732 on the builtin flags.
23733 * config/s390/s390.c (s390_expand_builtin): Support additional
23734 flags argument. Change error message to match the messages
23735 emitted in s390-c.c.
23736 * config/s390/s390.md: New UNSPEC_* constants.
23737 (op_type): Add new instruction types.
23738 * config/s390/vecintrin.h: Add new builtins and test data class
23740 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
23741 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
23742 (VEC_INEXACT, VEC_NOINEXACT): New constants.
23743 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
23744 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
23745 ("vec_mergel<mode>"): V_HW -> VEC_HW.
23747 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
23748 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
23749 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
23750 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
23752 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
23753 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
23754 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
23755 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
23757 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
23758 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
23759 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
23760 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
23761 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
23762 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
23763 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
23765 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
23766 ("vec_scatter_element<V_HW_4:mode>_DI")
23767 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
23768 ("vec_fpint<mode>", "vflls")
23769 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
23770 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
23771 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
23772 ("*vec_cmphe<mode>_cc"): ... these.
23774 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
23775 mode constant instead of magic value.
23777 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23779 * config/s390/s390.c (s390_expand_vec_compare): Support other
23780 vector floating point modes than just V2DF.
23781 (s390_expand_vcond): Likewise.
23782 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
23783 (s390_cannot_change_mode_class): Prevent mode changes between TF
23784 and V1TF in vector registers.
23785 * config/s390/s390.md (DF, SF): New mode attributes.
23786 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
23787 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
23788 SFmode support for VRs.
23789 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
23791 (VFT, VF_HW): New mode iterators.
23792 (vw, sdx): New mode attributes.
23793 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
23794 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
23795 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
23796 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
23797 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
23798 also the new vector floating point modes. Renaming to ...
23800 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
23801 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
23802 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
23803 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
23804 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
23805 ("vec_unordered<mode>"): ... these.
23807 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
23808 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
23809 ("*vec_extendv2df"): New insn definitions.
23811 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23813 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
23814 ("mulditi3_2", "*muldi3_sign"): New patterns.
23815 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
23816 rename the pattern definition.
23818 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23820 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
23822 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
23824 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23826 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
23827 instruction if possible.
23828 * config/s390/vector.md (vec_halfnumelts): New mode
23830 ("*vec_vllezlf<mode>"): New pattern.
23832 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23834 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
23835 ("popcountv4si2", "popcountv2di2"): Rename to ...
23836 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
23837 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
23839 ("popcount<mode>2_vxe"): New pattern.
23841 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23843 * common/config/s390/s390-common.c (processor_flags_table): Add
23845 * config.gcc: Add arch12.
23846 * config/s390/driver-native.c (s390_host_detect_local_cpu):
23847 Default to arch12 for unknown CPU model numbers.
23848 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
23849 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
23850 PROCESSOR_max sanity check.
23851 * config/s390/s390-opts.h (enum processor_type): Add
23853 * config/s390/s390.c (processor_table): Add arch12.
23854 (s390_expand_builtin): Add check for B_VXE flag.
23855 (s390_issue_rate): Add PROCESSOR_ARCH12.
23856 (s390_get_sched_attrmask): Likewise.
23857 (s390_get_unit_mask): Likewise.
23858 (s390_sched_score): Enable z13 scheduling for arch12.
23859 (s390_sched_reorder): Likewise.
23860 (s390_sched_variable_issue): Likewise.
23861 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
23863 (s390_tune_attr): Use z13 scheduling also for arch12.
23864 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
23865 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
23866 (TARGET_VXE_P): New macros.
23867 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
23868 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
23869 * config/s390/s390.opt: Add arch12 as processor_type.
23871 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23873 * config/s390/s390.md
23874 ("fixuns_truncdddi2", "fixuns_trunctddi2")
23875 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
23876 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
23878 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
23879 Rename expanders to ...
23881 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
23882 ("fixuns_truncdddi2_emu"): ... these.
23884 ("fixuns_trunc<mode>si2_emu"): New expander.
23886 ("*fixuns_truncdfdi2_z13"): Rename to ...
23887 ("*fixuns_truncdfdi2_vx"): ... this.
23889 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23891 * config/s390/2964.md: Remove the single element vector compare
23892 instructions which are no longer used.
23893 * config/s390/s390.c (s390_select_ccmode): Remove handling of
23895 (s390_canonicalize_comparison): Remove handling of DFmode
23897 (s390_expand_vec_compare_scalar): Remove function.
23898 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
23899 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
23901 ("*cmp<mode>_ccs"): Add wfcdb instruction.
23903 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23905 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
23907 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
23908 will anyway by matched by mov<mode>_64dfp.
23910 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23912 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
23913 vlef/vstef. Add missing operand to vleif.
23915 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23917 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
23918 pair for all vector types with 64 bit elements.
23919 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
23920 * config/s390/vector.md (V_HW_64): ... here.
23921 (V_128_NOSINGLE): New mode iterator.
23922 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
23923 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
23924 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
23925 ("*vec_load_pairv2di"): Change to ...
23926 ("*vec_load_pair<mode>"): ... this one.
23928 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23930 * config/s390/constraints.md: Add comments.
23931 (jKK): Reject element sizes > 8 bytes.
23932 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
23934 * config/s390/s390.md: Add the s_operand checks formerly in
23935 s390_split_ok_p to various splitters where they are still
23937 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
23938 for 128 bit vectors. Plus two splitters.
23940 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23942 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
23945 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23948 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
23949 error if the boundary argument is not constant.
23951 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23953 PR rtl-optimization/80112
23954 * loop-doloop.c (doloop_condition_get): Don't check condition
23955 if cmp isn't SET with IF_THEN_ELSE src.
23957 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23959 PR tree-optimization/80158
23960 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
23961 replacing a candidate statement, also replace it for the
23962 candidate's alternate interpretation.
23963 (replace_rhs_if_not_dup): Likewise.
23964 (replace_one_candidate): Likewise.
23966 2017-03-24 Richard Biener <rguenther@suse.de>
23968 PR tree-optimization/80167
23969 * graphite-isl-ast-to-gimple.c
23970 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
23972 (translate_isl_ast_to_gimple::get_rename): Likewise.
23974 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23976 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
23977 handling of certain combinations of target options, including the
23978 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
23979 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
23981 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23984 * config/arm/arm.md (*load_multiple): Add reload_completed to
23985 matching condition.
23987 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23988 Richard Biener <rguenth@suse.de>
23990 PR tree-optimization/79908
23991 PR tree-optimization/80136
23992 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23993 been cast away, gimplify_and_add suffices.
23995 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
23997 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
23999 2017-03-23 Richard Biener <rguenther@suse.de>
24001 PR tree-optimization/80032
24002 * gimplify.c (gimple_push_cleanup): Forced unconditional
24003 cleanups still have to go to the conditional_cleanups
24006 2017-03-22 Jakub Jelinek <jakub@redhat.com>
24008 PR tree-optimization/80072
24009 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
24011 (next_operand_entry_id): Change type to unsigned int.
24012 (sort_by_operand_rank): Make sure to return the right return value
24013 even if unsigned fields are bigger than INT_MAX.
24014 (struct oecount): Change cnt and id type to unsigned int.
24015 (oecount_hasher::equal): Formatting fix.
24016 (oecount_cmp): Make sure to return the right return value
24017 even if unsigned fields are bigger than INT_MAX.
24018 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
24021 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
24022 TREE_READONLY on result if writing it more than once.
24025 * doc/invoke.texi (-fsanitize=thread): Document that with
24026 -fnon-call-exceptions atomics are not able to throw
24030 * tsan.c: Include tree-eh.h.
24031 (instrument_builtin_call): Call maybe_clean_eh_stmt or
24032 maybe_clean_or_replace_eh_stmt where needed.
24033 (instrument_memory_accesses): Add cfg_changed argument.
24034 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
24035 if it returned true.
24036 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
24038 PR rtl-optimization/63191
24039 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
24040 wrapper function, moved the whole old content into ...
24041 (ix86_delegitimize_address_1): ... this. New inline function.
24042 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
24043 true as last argument instead of ix86_delegitimize_address.
24045 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24047 * config/aarch64/aarch64.c (generic_branch_cost): Copy
24048 cortexa57_branch_cost.
24050 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24052 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
24054 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24057 * doc/md.texi (Constraints): Document wA constraint.
24058 * config/rs6000/constraints.md (wA): New.
24059 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
24060 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
24061 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
24062 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
24064 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
24067 * gimplify.c (is_oacc_declared): New function.
24068 (oacc_default_clause): Use it to set default flags for acc declared
24069 variables inside parallel regions.
24070 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
24071 declared variables.
24072 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
24073 declare attribute to any decl as necessary.
24075 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24078 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
24079 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
24080 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
24081 (arm_arch_lpae): This.
24082 * config/arm/arm.c (arm_arch7ve): Rename into ...
24083 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
24084 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
24087 2017-03-22 Martin Liska <mliska@suse.cz>
24090 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
24091 error message instead of an ICE.
24093 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24095 * doc/extend.texi (6.11 Additional Floating Types): Revise.
24097 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
24099 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24101 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24104 2017-03-21 Martin Sebor <msebor@redhat.com>
24106 * doc/extend.texi: Use "cannot" instead of "can't."
24107 * doc/hostconfig.texi: Same.
24108 * doc/install.texi: Same.
24109 * doc/invoke.texi: Same.
24110 * doc/loop.texi: Same.
24111 * doc/md.texi: Same.
24112 * doc/objc.texi: Same.
24113 * doc/rtl.texi: Same.
24114 * doc/tm.texi: Same.
24115 * doc/tm.texi.in: Same.
24116 * doc/trouble.texi: Same.
24118 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
24121 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
24122 (collect_checksum_attributes): Set it.
24123 (die_checksum_ordered): Use it.
24125 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24127 PR tree-optimization/79908
24128 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
24129 change: For a VA_ARG whose LHS has been cast away, use
24130 force_gimple_operand to construct the side effects.
24132 2017-03-21 David Malcolm <dmalcolm@redhat.com>
24134 PR translation/80001
24135 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
24136 more amenable to translation.
24137 (oacc_loop_auto_partitions): Likewise.
24139 2017-03-21 Marek Polacek <polacek@redhat.com>
24140 Martin Sebor <msebor@redhat.com>
24142 PR tree-optimization/80109
24143 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
24144 on INTEGRAL_TYPE_P.
24146 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24147 Segher Boessenkool <segher@kernel.crashing.org>
24150 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
24151 check reg_used_between_p between insn and one of succ or succ2
24152 depending on if succ is artificial insn not inserted into insn
24155 2017-03-21 Martin Liska <mliska@suse.cz>
24157 PR gcov-profile/80081
24158 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
24159 * doc/gcc.texi: Include gcov-dump stuff.
24160 * doc/gcov-dump.texi: New file.
24162 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
24164 PR rtl-optimization/79150
24165 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
24166 conditional jump, if the jump is the last insn of the loop.
24168 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24169 Richard Biener <rguenth@suse.de>
24171 PR tree-optimization/79908
24172 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24173 been cast away, use force_gimple_operand to construct the side
24176 2017-03-21 Martin Liska <mliska@suse.cz>
24178 PR libfortran/79956
24179 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
24182 2017-03-21 Brad Spengler <spender@grsecurity.net>
24185 * plugin.c (htab_hash_plugin): New function.
24186 (add_new_plugin): Use it and adjust.
24187 (parse_plugin_arg_opt): Adjust.
24188 (init_one_plugin): Likewise.
24190 2017-03-21 Richard Biener <rguenther@suse.de>
24192 PR tree-optimization/80032
24193 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
24194 if set force the cleanup to happen unconditionally.
24195 (gimplify_target_expr): Push inserted clobbers with force_uncond
24196 to avoid them being removed by control-dependent DCE.
24198 2017-03-21 Richard Biener <rguenther@suse.de>
24200 PR tree-optimization/80122
24201 * tree-inline.c (copy_bb): Do not expans va-arg packs or
24202 va_arg_pack_len when the inlined call stmt requires pack
24204 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
24206 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24209 * tsan.c (instrument_builtin_call): If the memory model argument
24210 is not a constant, assume it is valid.
24213 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
24216 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24218 PR rtl-optimization/79910
24219 * combine.c (can_combine_p): Do not allow combining an I0 or I1
24220 if its dest is used by an insn before I2 (other than the combined
24221 insns themselves, which are properly handled already).
24223 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24226 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24228 * combine.c (record_used_regs): New static function.
24229 (try_combine): Handle situations where there is an additional
24230 instruction between I2 and I3 which needs to have a LOG_LINK
24234 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24236 * combine.c (try_combine): Delete redundant i1 test. Call
24237 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24239 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24242 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
24243 alternatives 13/14.
24245 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24247 PR tree-optimization/80054
24248 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
24249 the optimization if a PHI or any of its arguments is not dominated
24250 by the candidate's basis. Use gphi* rather than gimple* as
24252 (replace_profitable_candidates): Clean up a gimple* variable that
24253 should be a gphi* variable.
24255 2017-03-20 Martin Sebor <msebor@redhat.com>
24258 * doc/extend.texi (attribute constructor): Document present limitation.
24260 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
24263 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
24264 __POWER9_VECTOR__ #ifdef control, change template definition to
24265 use Power9-specific built-in function.
24266 (vec_any_eq): Likewise.
24267 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
24268 to control outcomes from this test.
24269 (vector_ae_<mode>p): For VEC_F modes, likewise.
24271 2017-03-20 Ian Lance Taylor <iant@google.com>
24273 * config/i386/i386.c (ix86_function_regparm): Save an extra
24274 register for -fsplit-stack with DECL_STATIC_CHAIN.
24276 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24279 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
24280 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
24282 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24284 * config/riscv/riscv.c (riscv_print_operand): Use "fence
24286 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
24289 2017-03-20 Marek Polacek <polacek@redhat.com>
24292 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
24294 2017-03-20 Richard Biener <rguenther@suse.de>
24296 PR tree-optimization/80113
24297 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
24298 allocate extra SSA name for PHI def.
24299 (add_close_phis_to_outer_loops): Likewise.
24300 (add_close_phis_to_merge_points): Likewise.
24301 (copy_loop_close_phi_args): Likewise.
24302 (copy_cond_phi_nodes): Likewise.
24304 2017-03-20 Martin Liska <mliska@suse.cz>
24306 PR middle-end/79753
24307 * tree-chkp.c (chkp_build_returned_bound): Do not build
24308 returned bounds for a LHS that's not a BOUNDED_P type.
24310 2017-03-20 Martin Liska <mliska@suse.cz>
24314 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
24315 COMPLEX_CST and VECTOR_CST.
24317 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24320 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
24321 target operand. A new splitter adds the clobber statement in case
24322 the target operand is dead anyway.
24324 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
24326 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
24327 to age-old versions of binutils and glibc.
24329 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
24331 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
24333 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24335 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
24337 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24339 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
24340 requirement for binutils 2.13.
24342 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24344 * combine.c (try_combine): Delete redundant i1 test. Call
24345 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24347 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
24349 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
24350 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
24352 <riscv64-*-elf>: Re-arrange section
24353 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
24354 <riscv32-*-linux>: Likewise.
24355 <riscv64-*-elf>: Likewise
24356 <riscv64-*-linux>: Likewise.
24358 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
24361 * aarch64.opt(verbose-cost-dump): Fix typo.
24363 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
24366 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
24367 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
24369 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24371 * reload.c (find_reloads): When reloading a nonoffsettable address,
24372 use RELOAD_OTHER for it and its address reloads.
24374 PR rtl-optimization/79910
24375 * combine.c (record_used_regs): New static function.
24376 (try_combine): Handle situations where there is an additional
24377 instruction between I2 and I3 which needs to have a LOG_LINK
24380 2017-03-17 Jeff Law <law@redhat.com>
24382 PR tree-optimization/71437
24383 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
24384 conditional in the hash table first.
24385 (vrp_dom_walker::before_dom_children): Extract condition from
24386 ASSERT_EXPR. Record condition, its inverion and any implied
24387 conditions as well.
24389 2017-03-17 Marek Polacek <polacek@redhat.com>
24390 Markus Trippelsdorf <markus@trippelsdorf.de>
24392 PR tree-optimization/80079
24393 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
24396 2017-03-17 Richard Biener <rguenther@suse.de>
24398 PR middle-end/80075
24399 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
24400 Properly verify the LHS before the RHS possibly claims to be
24402 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
24405 2017-03-17 Martin Jambor <mjambor@suse.cz>
24407 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
24408 (List of -O2 options): Likewise.
24409 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
24412 2017-03-17 Tom de Vries <tom@codesourcery.com>
24414 * gcov-dump.c (print_usage): Print bug_report_url.
24416 2017-03-17 Richard Biener <rguenther@suse.de>
24418 PR middle-end/80050
24419 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
24420 (parser::peek): Likewise.
24422 2017-03-17 Richard Biener <rguenther@suse.de>
24424 PR tree-optimization/80048
24425 * sese.c (free_sese_info): Properly release rename_map and
24426 copied_bb_map elements.
24428 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
24430 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
24431 Add linked-list forward and backlinks. Insert on
24432 construction, remove on destruction.
24433 (class pass_store_merging): Add m_stores_head field.
24434 (pass_store_merging::terminate_and_process_all_chains):
24435 Iterate over m_stores_head list.
24436 (pass_store_merging::terminate_all_aliasing_chains):
24438 (pass_store_merging::execute): Check for debug stmts first.
24439 Push new chains onto the m_stores_head stack.
24441 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24444 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
24445 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
24446 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
24448 2017-03-16 Jeff Law <law@redhat.com>
24450 PR tree-optimization/71437
24451 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
24452 member function. Implementation moved into after_dom_children
24453 member function and into the threader's thread_outgoing_edges
24455 (dom_opt_dom_walker::after_dom_children): Simplify by moving
24456 some code into new thread_outgoing_edges.
24457 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
24458 definition. Simplify marker handling (do it here). Assume we always
24459 have the available expression and the const/copies tables.
24460 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
24462 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
24463 * tree-vrp.c (equiv_stack): No longer file scoped.
24464 (vrp_dom_walker): New class.
24465 (vrp_dom_walker::before_dom_children): New member function.
24466 (vrp_dom_walker::after_dom_children): Likewise.
24467 (identify_jump_threads): Setup domwalker. Use it rather than
24468 walking edges in a random order by hand. Simplify setup/finalization.
24469 (finalize_jump_threads): Remove.
24470 (vrp_finalize): Do not call identify_jump_threads here.
24471 (execute_vrp): Do it here instead and call thread_through_all_blocks
24474 PR tree-optimization/71437
24475 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
24477 (simplify_stmt_for_jump_threading): Add basic_block argument. All
24479 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
24480 (dom_opt_dom_walker::thread_across_edge): Remove
24481 handle_dominating_asserts argument. All callers changed.
24482 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
24483 changes. Remove calls to lhs_of_dominating_assert. Other
24484 uses of handle_dominating_asserts turn into unconditional code
24485 (simplify_control_stmt_condition_1): Likewise.
24486 (simplify_control_stmt_condition): Likewise.
24487 (thread_through_normal_block, thread_across_edge): Likewise.
24488 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
24489 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
24490 object if it is not an SSA_NAME.
24491 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
24492 before calling into the VRP specific simplifiers.
24493 (identify_jump_threads): Remove handle_dominating_asserts
24496 2017-03-16 Jakub Jelinek <jakub@redhat.com>
24499 * tree-diagnostic.c (default_tree_printer): No longer static.
24500 * tree-diagnostic.h (default_tree_printer): New prototype.
24502 2017-03-16 Tamar Christina <tamar.christina@arm.com>
24504 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
24505 Change ins into fmov.
24507 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24509 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
24510 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
24511 Use h_con constraint for operand 1.
24512 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
24513 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
24515 2017-03-15 Jeff Law <law@redhat.com>
24517 PR tree-optimization/71437
24518 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
24519 (record_temporary_equivalences): Use it.
24521 PR tree-optimization/71437
24522 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
24523 tree-ssa-scopedtables.
24524 (lookup_avail_expr, build_and_record_new_cond): Likewise.
24525 (record_conditions, record_cond, vuse_eq): Likewise.
24526 (record_edge_info): Adjust to API tweak of record_conditions.
24527 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
24528 (record_temporary_equivalences, optimize_stmt): Likewise.
24529 (eliminate_redundant_computations): Likewise.
24530 (record_equivalences_from_stmt): Likewise.
24531 * tree-ssa-scopedtables.c: Include options.h and params.h.
24532 (vuse_eq): New function, moved from tree-ssa-dom.c
24533 (build_and_record_new_cond): Likewise.
24534 (record_conditions): Likewise. Accept vector of conditions rather
24535 than edge_equivalence structure for first argument.
24536 for the first argument.
24537 (avail_exprs_stack::lookup_avail_expr): New member function, moved
24538 from tree-ssa-dom.c.
24539 (avail_exprs_stack::record_cond): Likewise.
24540 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
24541 from tree-ssa-dom.c.
24542 (avail_exprs_stack): Add new member functions lookup_avail_expr
24544 (record_conditions): Declare.
24546 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
24549 * lra-constraints.c (process_alt_operands): Increase reject for
24550 reloading an input/output operand.
24552 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
24555 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
24556 insns to convert from signed/unsigned char/short to IEEE 128-bit
24558 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
24560 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
24563 * config/i386/i386.c (ix86_vector_duplicate_value): Create
24564 subreg of inner mode for values already in registers.
24566 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
24568 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
24569 iteration reg is used after the loop.
24571 2017-03-14 Martin Sebor <msebor@redhat.com>
24573 PR tree-optimization/79800
24574 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
24575 precision in negative-positive range.
24576 (format_floating): Call non-const overload with adjusted precision.
24578 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24581 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
24584 2017-03-14 Martin Sebor <msebor@redhat.com>
24586 PR middle-end/80020
24587 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
24588 * builtins.def (aligned_alloc): Use it.
24591 * Makefile.in (GTFILES): Add calls.c.
24592 * calls.c: Include "gt-calls.h".
24594 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
24596 PR rtl-optimization/79728
24597 * regs.h (struct target_regs): New field
24598 x_contains_allocatable_regs_of_mode.
24599 (contains_allocatable_regs_of_mode): New macro.
24600 * reginfo.c (init_reg_sets_1): Initialize it, and change
24601 contains_reg_of_mode so it includes global regs as well.
24602 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
24603 rather than contains_regs_of_mode.
24605 2017-03-14 Martin Liska <mliska@suse.cz>
24607 * doc/invoke.texi: Document options that can't be combined with
24608 -fcheck-pointer-bounds.
24610 2017-03-14 Martin Liska <mliska@suse.cz>
24612 PR middle-end/79831
24613 * doc/invoke.texi (-Wchkp): Document the option.
24615 2017-03-14 Martin Liska <mliska@suse.cz>
24617 * Makefile.in: Install gcov-dump.
24619 2017-03-14 Martin Liska <mliska@suse.cz>
24621 * multiple_target.c (expand_target_clones): Bail out for
24622 an invalid attribute.
24624 2017-03-14 Richard Biener <rguenther@suse.de>
24626 * alias.c (struct alias_set_entry): Pack properly.
24627 * cfgloop.h (struct loop): Likewise.
24628 * cse.c (struct set): Likewise.
24629 * ipa-utils.c (struct searchc_env): Likewise.
24630 * loop-invariant.c (struct invariant): Likewise.
24631 * lra-remat.c (struct cand): Likewise.
24632 * recog.c (struct change_t): Likewise.
24633 * rtl.h (struct address_info): Likewise.
24634 * symbol-summary.h (function_summary): Likewise.
24635 * tree-loop-distribution.c (struct partition): Likewise.
24636 * tree-object-size.c (struct object_size_info): Likewise.
24637 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
24638 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
24639 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
24640 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
24641 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
24642 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
24643 (struct _stmt_vec_info): Likewise.
24645 2017-03-14 Martin Liska <mliska@suse.cz>
24648 * multiple_target.c (create_dispatcher_calls): Check that
24649 a target can create a function dispatcher.
24651 2017-03-14 Martin Liska <mliska@suse.cz>
24654 * multiple_target.c (expand_target_clones): Drop local.local
24655 flag for default implementation.
24657 2017-03-14 Richard Biener <rguenther@suse.de>
24659 PR tree-optimization/80030
24660 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
24662 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
24664 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
24665 gcc_fallthrough() instead of __attribute__((fallthrough));
24667 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24669 * doc/gcc.texi: Remove "up" link to (DIR).
24670 * doc/gccint.texi: Ditto.
24672 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24674 * doc/install.texi (Specific) <avr>: Remove reference to
24677 2017-03-13 Jeff Law <law@redhat.com>
24679 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
24680 attribute rather than comments.
24682 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
24683 match_scratch operand is highest.
24685 2017-03-13 Martin Liska <mliska@suse.cz>
24687 PR middle-end/78339
24688 * ipa-pure-const.c (warn_function_noreturn): If the declarations
24689 is a CHKP clone, use original declaration.
24691 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24693 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
24694 (arc_conditional_register_usage): Use a different allocation order
24695 when optimizing for size.
24696 * common/config/arc/arc-common.c (arc_option_optimization_table):
24697 Section anchors default on when optimizing for size.
24699 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24701 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
24703 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24705 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
24706 * config/arc/arc.md (cpu_facility): Add cd variant.
24707 (*movqi_insn): Add code density variant.
24708 (*movhi_insn): Likewise.
24709 (*movqi_insn): Likewise.
24710 (*addsi3_mixed): Likewise.
24711 (subsi3_insn): Likewise.
24713 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24715 * config/arc/arc.md (movsi_cond_exec): Update constraint.
24717 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24719 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
24720 expressions with MINUS and UNARY ops.
24722 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24725 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
24727 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
24728 between vec_select and vector argument.
24729 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
24730 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
24731 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
24732 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
24733 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
24734 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
24736 2017-03-13 Richard Biener <rguenther@suse.de>
24739 * params.def (vect-max-peeling-for-alignment): Fix typo.
24741 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24743 * doc/install.texi (Specific) <mips-*-*>: Remove description of
24744 issue that only occurred with binutils below 2.18.
24746 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24748 * doc/install.texi (Specific) <cris-axis-elf>: No longer
24749 refer to binutils 2.11/2.12 minimum.
24751 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24753 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
24754 ftp.kernel.org and simplify binutils requirement.
24756 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
24758 * doc/invoke.texi (Warning Options): Fix spelling of link-time
24760 (Optimize Options): Ditto. Also remove redundancy.
24762 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24764 PR translation/79848
24765 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
24767 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
24768 to G_ to avoid double translation.
24770 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24772 PR translation/79923
24773 * auto-profile.c (get_combined_location): Convert leading
24774 character of diagnostics to lower case and remove trailing period.
24775 (read_profile): Likewise for various diagnostics.
24776 * config/arm/arm.c (arm_option_override): Remove trailing period
24777 from various diagnostics.
24778 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
24779 (msp430_expand_delay_cycles): Likewise.
24781 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24784 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
24785 full command-line argument, rather than just "str".
24786 (aarch64_validate_march): Likewise.
24787 (aarch64_validate_mtune): Likewise.
24789 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
24791 PR rtl-optimization/78911
24792 * lra-assigns.c (must_not_spill_p): New function.
24793 (spill_for): Use it.
24795 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24797 PR tree-optimization/79981
24798 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
24799 ATOMIC_COMPARE_EXCHANGE ifn result.
24800 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
24801 IFN_ATOMIC_COMPARE_EXCHANGE.
24803 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24806 * opts.c (parse_sanitizer_options): Add missing question mark to
24807 "did you mean" message.
24809 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24811 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
24813 (VMULEUH_UNS): Likewise.
24814 (VMULOUB_UNS): Likewise.
24815 (VMULOUH_UNS): Likewise.
24816 * config/rs6000/rs6000.c (builtin_function_type): Remove
24817 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
24819 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24822 * read-rtl-function.c (function_reader::read_rtx_operand): Update
24823 x with result of extra_parsing_for_operand_code_0.
24824 (function_reader::extra_parsing_for_operand_code_0): Convert
24825 return type from void to rtx, returning x. When reading
24826 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
24827 larger size containing struct block_symbol.
24829 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
24831 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
24832 -mfloat128-hardware without -m64.
24834 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
24837 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
24838 entries to the case statement that marks unsigned arguments to
24839 overloaded functions.
24841 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24843 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
24844 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
24846 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
24849 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
24850 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
24852 2017-03-10 Martin Liska <mliska@suse.cz>
24856 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
24857 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
24860 2017-03-10 Olivier Hainque <hainque@adacore.com>
24862 * tree-switch-conversion (array_value_type): Start by resetting
24863 candidate type to it's main variant.
24865 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24867 PR rtl-optimization/79909
24868 * combine.c (try_combine): Use simplify_replace_rtx on individual
24869 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
24870 of the whole CALL_INSN_FUNCTION_USAGE.
24872 PR tree-optimization/79972
24873 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
24874 get_range_info on SSA_NAMEs. Formatting fixes.
24876 2017-03-10 Richard Biener <rguenther@suse.de>
24877 Jakub Jelinek <jakub@redhat.com>
24879 PR tree-optimization/77975
24880 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
24881 edge to be constant.
24882 (get_val_for): For constant x return it. Formatting fix.
24883 (loop_niter_by_eval): Avoid pointless looping if the next iteration
24884 would use the same bases as the current one.
24886 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24888 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
24889 instead of vec_select for V1TImode.
24890 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
24892 (VSX_LE_128): Add V1TI to this mode iterator.
24893 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
24894 (*vsx_le_perm_store_<mode>): Likewise.
24895 (pre-reload splitter for VSX stores): Likewise.
24896 (post-reload splitter for VSX stores): Likewise.
24897 (*vsx_xxpermdi2_le_<mode>): Likewise.
24898 (*vsx_lxvd2x2_le_<mode>): Likewise.
24899 (*vsx_stxvd2x2_le_<mode>): Likewise.
24901 2017-03-09 Michael Eager <eager@eagercon.com>
24903 Correct failures with --enable-checking=yes,rtl.
24905 * config/microblaze/microblaze.c (microblaze_expand_shift):
24906 Replace GET_CODE test with CONST_INT_P and INTVAL test with
24907 test for const0_rtx.
24908 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
24909 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
24911 2017-03-09 Richard Biener <rguenther@suse.de>
24913 PR tree-optimization/79977
24914 * graphite-scop-detection.c (scop_detection::merge_sese):
24915 Handle the case of extra exits to blocks dominating the entry.
24917 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
24919 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24922 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
24924 PR rtl-optimization/79949
24925 * lra-constraints.c (process_alt_operands): Check memory when
24926 trying to predict a cycle. Print about the overall increase.
24928 2017-03-09 Richard Biener <rguenther@suse.de>
24930 PR middle-end/79971
24931 * gimple-expr.c (useless_type_conversion_p): Preserve
24932 TYPE_SATURATING for fixed-point types.
24934 2017-03-09 Richard Biener <rguenther@suse.de>
24937 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
24938 alignment of BLKmode params.
24940 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24943 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
24944 (VALL_NO_V2Q): Likewise.
24946 * config/aarch64/aarch64-simd.md
24947 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
24949 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
24950 VALL_NO_V2Q mode iterator.
24951 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
24953 2017-03-09 Martin Liska <mliska@suse.cz>
24955 PR tree-optimization/79631
24956 * tree-chkp-opt.c (chkp_is_constant_addr): Call
24957 tree_int_cst_sign_bit just for INTEGER constants.
24959 2017-03-09 Martin Liska <mliska@suse.cz>
24963 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
24966 2017-03-09 Marek Polacek <polacek@redhat.com>
24969 * tree.c (inchash::add_expr): Handle TREE_VEC.
24971 2017-03-09 Martin Liska <mliska@suse.cz>
24974 (chkp_narrow_size_and_offset): New function.
24975 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
24976 (void chkp_parse_bit_field_ref): New function.
24977 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
24978 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
24980 2017-03-09 Martin Liska <mliska@suse.cz>
24983 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
24984 (chkp_find_bounds_1): Remove gcc_unreachable.
24986 2017-03-09 Jakub Jelinek <jakub@redhat.com>
24989 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
24990 BUILT_IN_SYNC*, determine the access type from the size suffix and
24991 always build a MEM_REF with that type. Handle forgotten
24992 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
24995 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
24996 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
24997 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
24998 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
24999 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
25000 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
25001 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
25002 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
25003 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
25004 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
25005 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
25006 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
25007 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
25008 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
25009 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
25010 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
25011 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
25012 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
25013 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
25014 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
25015 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
25016 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
25017 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
25018 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
25019 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
25020 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
25021 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
25022 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
25023 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
25024 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
25025 definitions outside of __OPTIMIZE__ guarded section.
25028 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
25029 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
25030 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
25031 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
25034 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25036 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
25037 ("vfenez<mode>"): Add missing constraints.
25039 2017-03-08 Martin Sebor <msebor@redhat.com>
25042 * config/nds32/nds32.c (nds32_option_override):
25043 Fix misspelled diagnostic.
25045 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25048 * gimplify.c (gimplify_omp_for): Replace index var in outer
25049 taskloop statement with an artificial variable and add
25050 OMP_CLAUSE_PRIVATE clause for it.
25052 2017-03-08 Richard Biener <rguenther@suse.de>
25054 PR tree-optimization/79955
25055 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
25056 for accesses that are completely outside of the variable.
25058 2017-03-08 Andrew Haley <aph@redhat.com>
25060 PR tree-optimization/79943
25061 * tree-ssa-loop-split.c (compute_new_first_bound): When
25062 calculating the new upper bound, (END-BEG) should be added, not
25065 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25067 * config/avr/avr.md (setmemhi): Make sure match_dup
25068 operand number comes before match_scratch.
25070 2017-03-08 Richard Biener <rguenther@suse.de>
25072 PR tree-optimization/79920
25073 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
25074 with ncopies == 1 to ...
25075 (vect_transform_slp_perm_load): ... here. Properly compute
25076 all element loads by iterating VF times over the group. Do
25077 not handle ncopies (computed in a broken way) in
25078 vect_create_mask_and_perm.
25080 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25083 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
25084 is a uniform vector, use uniform_vector_p return value instead of
25085 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
25087 2017-03-07 Marek Polacek <polacek@redhat.com>
25089 PR middle-end/79809
25090 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
25091 (alloca_call_type): Likewise.
25093 2017-03-07 Martin Liska <mliska@suse.cz>
25095 * gcov.c (process_args): Put comment to correct location.
25097 2017-03-07 Martin Liska <mliska@suse.cz>
25099 PR middle-end/68270
25100 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
25101 Use array_at_struct_end_p instead of DECL_CHAIN (field).
25102 (chkp_narrow_bounds_for_field): Likewise.
25103 (chkp_parse_array_and_component_ref): Pass one more argument to
25106 2017-03-07 Richard Biener <rguenther@suse.de>
25108 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
25111 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25113 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
25114 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
25116 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25119 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
25120 to end of description.
25121 (PARAM_MAX_STORES_TO_MERGE): Likewise.
25123 2017-03-07 Jakub Jelinek <jakub@redhat.com>
25125 PR rtl-optimization/79901
25126 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
25128 (*avx512f_<code><mode>3<mask_name>): ... this.
25129 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
25130 iterator instead of VI8_AVX2_AVX512BW.
25132 PR rtl-optimization/79901
25133 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
25134 min/max expander, expand it using expand_vec_cond_expr.
25137 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
25140 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25143 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
25144 to void * for PCH reasons.
25145 * dwarf2out.c (output_loc_operands, output_die): Cast
25146 v.val_vec.array to unsigned char *.
25148 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
25151 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
25154 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
25156 PR rtl-optimization/79571
25157 * lra-constraints.c (process_alt_operands): Calculate static
25158 reject and subtract it from overall when only addresses will be
25161 2017-03-06 Julia Koval <julia.koval@intel.com>
25164 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
25165 incoming stack boundary to 128 for 64-bit targets.
25167 2017-03-06 Richard Biener <rguenther@suse.de>
25169 PR tree-optimization/79894
25170 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
25171 to NULL after folding it.
25173 2017-03-06 Richard Biener <rguenther@suse.de>
25175 PR tree-optimization/79824
25176 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
25177 check disabling peeling for gaps.
25179 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
25181 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25182 attributes): Document gettimeofday.
25184 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25186 * config/s390/s390.c (s390_option_override_internal): Set
25187 PARAM_MIN_VECT_LOOP_BOUND
25189 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25191 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
25192 * config/s390/s390.md: Likewise.
25194 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25197 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
25198 (<avx2_avx512>_perm<mode>): Rename to ...
25199 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
25201 (<avx512>_perm<mode>_mask): Rename to ...
25202 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
25204 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
25205 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
25206 instead of VI8F_256_512.
25207 (avx512f_perm<mode>): New define_expand.
25208 (avx512f_perm<mode>_mask): Likewise.
25209 (avx512f_perm<mode>_1<mask_name>): New define_insn.
25210 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
25212 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25214 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
25215 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
25217 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
25219 2017-03-06 Martin Liska <mliska@suse.cz>
25222 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
25223 when having a SSA NAME w/o VAR_DECL assigned to it.
25225 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25227 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
25228 msa_dpsub_<su>_d): Fix MODE for vec_select.
25230 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25232 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
25234 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
25236 2017-03-06 Richard Biener <rguenther@suse.de>
25238 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
25239 * plugin.c (register_plugin_info): Likewise.
25240 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
25242 2017-03-05 Jakub Jelinek <jakub@redhat.com>
25244 * config/i386/sse.md (sse_storehps, sse_storelps,
25245 avx_<castmode><avxsizesuffix>_<castmode>,
25246 avx512f_<castmode><avxsizesuffix>_<castmode>,
25247 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
25248 in condition that at least one operand is not a MEM.
25250 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25252 PR middle-end/79805
25253 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
25254 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
25256 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
25257 gimple_call_nothrow_p flag based on whether original builtin can throw.
25258 If it can, emit following stmts on the fallthrough edge.
25259 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
25260 don't create new bb if inserting just debug stmts on the edge, try to
25261 insert them on the fallthru bb or just reset debug stmts.
25263 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
25266 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
25267 restore recog_data (including the operand rtxes inside it) around
25268 the call to get_insn_template.
25270 2017-03-03 Martin Sebor <msebor@redhat.com>
25272 PR tree-optimization/79699
25273 * context.c (context::~context): Free MPFR caches to avoid
25274 a memory leak on program exit.
25276 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25278 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
25279 Use wide_int::ulow () instead of .elt (0).
25281 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25283 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
25284 (*pushxf): Limit oF constraint to 32bit targets and add oC
25285 constraint for 64bit targets.
25286 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
25287 (*pushdf): Change rmF constraint to rmC.
25289 2017-03-03 Martin Liska <mliska@suse.cz>
25291 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
25292 Remove unused variable.
25294 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25297 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
25298 is a memory operand, increase num_memory.
25299 (ix86_expand_args_builtin): Likewise.
25301 2017-03-03 Jan Hubicka <jh@suse.cz>
25304 * ipa-devirt.c (maybe_record_node): Properly handle
25305 __cxa_pure_virtual visibility.
25307 2017-03-03 Martin Liska <mliska@suse.cz>
25309 PR tree-optimization/79803
25310 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
25312 (pass_loop_prefetch::execute): Disabled optimization if an
25313 assumption about L1 cache size is not met.
25315 2017-03-03 Martin Liska <mliska@suse.cz>
25317 PR rtl-optimization/79574
25318 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
25319 (hash_scan_set): Likewise.
25320 (dump_hash_table): Likewise.
25321 (hoist_code): Likewise.
25323 2017-03-03 Richard Biener <rguenther@suse.de>
25325 * fixed-value.c (fixed_from_string): Restore use of elt (1)
25326 in place of uhigh ().
25327 (fixed_convert_from_real): Likewise.
25329 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25332 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
25334 2017-03-03 Richard Biener <rguenther@suse.de>
25336 PR middle-end/79818
25337 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
25338 TYPE_OVERFLOW_UNDEFINED check.
25340 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25342 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
25344 (vector_ae_<mode>_p): Likewise.
25345 (vector_nez_<mode>_p): Likewise.
25346 (vector_ne_v2di_p): Likewise.
25347 (vector_ae_v2di_p): Likewise.
25348 (vector_ne_<mode>_p): Likewise.
25349 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
25351 (vsx_tsqrt<mode>2_fe): Likewise.
25353 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
25356 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
25358 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25360 PR rtl-optimization/79780
25361 * cprop.c (one_cprop_pass): When second and further conditional trap
25362 in a single basic block is turned into an unconditional trap, turn it
25363 into a deleted note to avoid RTL verification failures.
25365 2017-03-02 Richard Biener <rguenther@suse.de>
25367 * fold-const.c (const_binop): Use ulow () instead of elt (0).
25369 2017-03-02 Richard Biener <rguenther@suse.de>
25371 PR tree-optimization/79345
25373 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
25374 param and abort the walk, returning -1 if it is hit.
25375 (walk_aliased_vdefs): Take a limit param and pass it on.
25376 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
25377 defaulting to 0 and return a signed int.
25378 * tree-ssa-uninit.c (struct check_defs_data): New struct.
25379 (check_defs): New helper.
25380 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
25381 about uninitialized memory.
25382 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
25383 bogus uninitialized warning.
25384 (fixed_convert_from_real): Likewise.
25386 2017-03-02 Bin Cheng <bin.cheng@arm.com>
25388 PR tree-optimization/66768
25389 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
25390 iv_use if base object can't be determined.
25392 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25394 PR tree-optimization/79345
25395 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
25396 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
25397 (get_pattern_stats): Initialize it.
25398 * genemit.c (gen_expand): Verify match_scratch numbers come after
25399 match_operand/match_dup numbers.
25400 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
25401 match_scratch numbers.
25402 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
25404 * config/s390/s390.md (trunctdsd2): Likewise.
25406 2017-03-02 Richard Biener <rguenther@suse.de>
25408 * wide-int.h (wide_int_storage::operator=): Implement in terms
25411 2017-03-02 Richard Biener <rguenther@suse.de>
25413 PR tree-optimization/79777
25414 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
25415 the to insert expression to sth existing.
25417 2017-03-01 Martin Sebor <msebor@redhat.com>
25419 PR middle-end/79692
25420 * gimple-ssa-sprintf.c
25421 (directive::known_width_and_precision): New function.
25422 (format_integer): Use it.
25423 (get_mpfr_format_length): Consider the full range of precision
25424 when computing %g output with the # flag. Set the likely byte
25425 count to 3 rather than 1 when precision is indeterminate.
25426 (format_floating): Correct the lower bound of precision.
25428 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25430 * doc/invoke.texi: Document default code model for 64-bit Linux.
25432 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25435 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
25436 udiv rather than div since input pattern is unsigned.
25438 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
25440 * config/i386/i386.c (print_reg): Warn for values of
25441 unsupported size in integer register.
25443 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
25446 * config/rs6000/predicates.md (current_file_function_operand): Do
25447 not allow self calls to be local if the function is replaceable.
25449 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
25452 * config/rs6000/altivec.h (vec_ctz and others): Change the
25453 preprocessor macro that controls conditional compilation from
25454 _ARCH_PWR9 to __POWER9_VECTOR__.
25455 (vec_all_ne): Change parameterization of __altivec_scalar_pred
25456 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
25457 control (instead of _ARCH_PWR9 control) so that template
25458 definition uses power9-specific function.
25459 (vec_any_eq): Likewise.
25460 (vec_all_ne): Change macro definition to use a power9-specific
25461 expansion under #ifdef __POWER9_VECTOR__ control (instead of
25462 _ARCH_PWR9 control).
25463 (vec_any_eq) Likewise.
25464 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
25465 expansion for CMPNEF to remove support for xvcmpnesp instruction.
25466 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
25467 support for xvcmpnedp instruction.
25468 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
25469 macro expansion so that Power9 implementation of vec_all_ne does
25470 not use the AltiVec predicate framework.
25471 (VCMPNEH_P): Likewise.
25472 (VCMPNEW_P): Likewise.
25473 (VCMPNED_P): Likewise.
25474 (VCMPNEFP_P): Likewise.
25475 (VCMPNEDP_P): Likewise.
25476 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
25477 implementation of vec_any_eq to not use AltiVec predicate
25479 (VCMPAEH_P): Likewise.
25480 (VCMPAEW_P): Likewise.
25481 (VCMPAED_P): Likewise.
25482 (VCMPAEFP_P): Likewise.
25483 (VCMPAEDP_P): Likewise.
25484 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
25485 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
25486 not use the AltiVec predicate framework.
25487 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
25488 of vec_any_eq to not use AltiVec predicate framework.
25489 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25490 support for predefined __POWER9_VECTOR__ macro to indicate that
25491 Power9 instruction selection is enabled.
25492 (altivec_overloaded_builtins): Remove extraneous
25493 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
25494 function argument types RS6000_BTI_bool_V16QI and
25495 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
25496 entry for overloaded function argument types RS6000_BTI_bool_V4SI
25497 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
25498 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
25499 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
25500 Power9 for implementations of vec_cmpne. Change the signature for
25501 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
25502 (representing vec_all_ne) to remove the previously described first
25503 argument of type RS6000_BTI_INTSI, as this was an artifact of
25504 reliance on the AltiVec predicate framework, which is no longer
25505 used in the implementation of these functions. Add
25506 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
25507 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
25508 since, unlike the AltiVec predicate framework implementation, we
25509 do not share function descriptors between vec_alle and vec_anyeq.
25510 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
25511 set of modes that receive special treatment even when
25512 TARGET_P9_VECTOR is true. The special treatment emits code that
25513 does not depend on Power9 instructions.
25514 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
25515 define_expand to not rely on AltiVec predicate framework.
25516 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25518 (vector_ne_v2di_p): Change this define_expand to not rely on
25519 AltiVec predicate framework.
25520 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
25522 (vector_ne_<mode>_p): Change this define_expand to not rely on
25523 AltiVec predicate framework.
25524 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25526 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
25527 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
25528 define_insn pattern.
25529 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
25530 define_insn pattern because the xvcmpne<VSs>. instruction is not
25532 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
25533 instruction is not supported.
25535 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25537 * config/nvptx/nvptx.c: Include intl.h.
25539 2017-03-01 Martin Jambor <mjambor@suse.cz>
25542 * ipa-prop.h (ipa_bits): Removed field known.
25543 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
25544 to pointers. Adjusted their comments to warn about their sharing.
25545 (ipcp_transformation_summary): Change bits to a vector of pointers.
25546 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
25547 (ipa_get_ipa_bits_for_value): Declare.
25548 * tree-vrp.h (value_range): Mark as GTY((for_user)).
25549 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
25550 (ipa_bits_hash_table): Likewise.
25551 (ipa_vr_ggc_hash_traits): Likewise.
25552 (ipa_vr_hash_table): Likewise.
25553 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
25554 being pointers and vr_known being removed.
25555 (ipa_set_jf_unknown): Likewise.
25556 (ipa_get_ipa_bits_for_value): New function.
25557 (ipa_set_jfunc_bits): Likewise.
25558 (ipa_get_value_range): New overloaded functions.
25559 (ipa_set_jfunc_vr): Likewise.
25560 (ipa_compute_jump_functions_for_edge): Use the above functions to
25561 construct bits and vr parts of jump functions.
25562 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
25563 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25565 (ipcp_grow_transformations_if_necessary): Also allocate
25566 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25568 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
25569 them. Fix too long lines.
25570 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
25571 vr_known being removed.
25572 (ipa_read_jump_function): Use new setter functions to construct bits
25573 and vr parts of jump functions or set them to NULL.
25574 (write_ipcp_transformation_info): Adjust for bits being pointers.
25575 (read_ipcp_transformation_info): Likewise.
25576 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
25578 Include gt-ipa-prop.h.
25579 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
25581 (ipcp_store_bits_results): Likewise.
25582 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
25583 Do not write to existing jump functions but use a temporary instead.
25585 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25588 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
25589 attempt to use its first operand as BIT_FIELD_REF base.
25591 2017-03-01 Richard Biener <rguenther@suse.de>
25593 PR middle-end/79721
25594 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
25595 interpolating formula in wrapping arithmetic.
25596 (chrec_apply): Convert chrec_evaluate return value to wanted type.
25598 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25600 PR tree-optimization/79734
25601 * tree-vect-generic.c (expand_vector_condition): Optimize
25602 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
25603 Handle VEC_COND_EXPR where comparison has different inner width from
25604 type's inner width.
25606 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
25608 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
25609 markup, and similar issues. Remove @opindex entries for things
25610 that aren't options. Add missing -mmpy-option entries.
25612 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25614 PR tree-optimization/79737
25615 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
25616 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
25617 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
25618 instead of byte_size. Formatting fix.
25619 (shift_bytes_in_array_right): Formatting fix.
25621 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
25624 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
25625 condition on optimize for the leaf function test.
25627 2017-02-28 Martin Liska <mliska@suse.cz>
25630 * read-rtl-function.c (function_reader::handle_unknown_directive):
25631 Bail out when one uses -flto.
25633 2017-02-28 Martin Liska <mliska@suse.cz>
25635 * common.opt: Replace space with tabular for options of <number>
25637 * config/i386/i386.opt: Show <number> value for
25638 -mlarge-data-threshold.
25639 * opts.c (print_filtered_help): Do not display number in hexadecimal
25642 2017-02-28 Martin Liska <mliska@suse.cz>
25644 * common.opt: Fix --help=option -Q for options which are of
25647 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
25649 * config/i386/i386.c (print_reg): Error out for values
25650 of 8-bit size in invalid integer register.
25652 2017-02-28 Martin Sebor <msebor@redhat.com>
25654 PR tree-optimization/79691
25655 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
25657 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25660 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
25661 gcc_unreachable with output_operand_lossage.
25663 2017-02-28 Richard Biener <rguenther@suse.de>
25665 PR tree-optimization/79740
25666 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
25668 (visit_nary_op): Insert the nary into the hashtable if we
25669 pattern-matched sth.
25670 * tree-ssa-pre.c (eliminate_insert): Robustify.
25672 2017-02-28 Richard Biener <rguenther@suse.de>
25674 PR middle-end/79731
25675 * fold-const.c (decode_field_reference): Reject out-of-bound
25678 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25680 * config/i386/i386.c: Include intl.h.
25681 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
25682 instead of just cond ? "..." : "...".
25683 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
25684 * coverage.c (read_counts_file): Likewise.
25685 * omp-offload.c: Include intl.h.
25686 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
25687 of just cond ? "..." : "...".
25688 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
25689 of just cond ? "..." : "...".
25691 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
25694 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
25696 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
25697 'tune for' CPU name.
25698 * config/arm/arm-cpu-data.h: Regenerated.
25700 2017-02-28 Richard Biener <rguenther@suse.de>
25702 PR tree-optimization/79732
25703 * tree-inline.c (expand_call_inline): Do not shadow var.
25705 2017-02-28 Richard Biener <rguenther@suse.de>
25707 PR tree-optimization/79723
25708 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
25709 address-space properly.
25711 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
25713 * doc/optinfo.texi (Optimization groups): Fix option used for
25715 * doc/invoke.texi (-fopt-info): Document "omp".
25716 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
25717 (OPTGROUP_ALL): Add OPTGROUP_OMP.
25718 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
25719 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
25720 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
25722 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
25724 * dumpfile.c (optgroup_options): Instead of "openmp", associate
25725 OPTGROUP_OMP with "omp".
25727 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
25730 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
25731 for arithmetic shift of unsigned V2DI.
25733 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
25735 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
25736 arc/linux.h headers.
25737 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
25738 (LINK_SPEC): Likewise.
25739 (ARC_TLS_EXTRA_START_SPEC): Likewise.
25740 (EXTRA_SPECS): Likewise.
25741 (STARTFILE_SPEC): Likewise.
25742 (ENDFILE_SPEC): Likewise.
25743 (LIB_SPEC): Likewise.
25744 (TARGET_SDATA_DEFAULT): Likewise.
25745 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
25746 (MULTILIB_DEFAULTS): Likewise.
25747 (DWARF2_UNWIND_INFO): Likewise.
25748 * config/arc/big.h: New file.
25749 * config/arc/elf.h: Likewise.
25750 * config/arc/linux.h: Likewise.
25751 * config/arc/t-uClibc: Remove.
25753 2017-02-27 Bin Cheng <bin.cheng@arm.com>
25755 PR tree-optimization/77536
25756 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
25757 (tree_transform_and_unroll_loop): Use above function to compute the
25758 estimated niter of unrolled loop and use it when scaling profile.
25759 Also use count info rather than frequency if it's non-zero.
25760 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
25761 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
25762 (vect_transform_loop): Call above function.
25764 2017-02-27 Richard Biener <rguenther@suse.de>
25766 PR tree-optimization/45397
25767 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
25768 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
25769 (visit_nary_op): Add pattern matching for CSEing sign-changed
25770 or truncated operations with wider ones.
25772 2017-02-27 Richard Biener <rguenther@suse.de>
25774 PR tree-optimization/79690
25775 * tree-vect-stmts.c (vectorizable_store): Use vector type
25776 built from the DR with address-space.
25778 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
25780 * doc/invoke.texi (Optimize Options): Refine the description
25781 of asan-use-after-return.
25783 2017-02-25 Alan Modra <amodra@gmail.com>
25785 PR rtl-optimization/79584
25786 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
25787 base, not ad->base_term, the reg within base. Remove assertion
25788 that ad->base == ad->base_term. Replace gen_int_mode using
25789 bogus mode with const0_rtx.
25791 2017-02-25 Jakub Jelinek <jakub@redhat.com>
25793 PR middle-end/79396
25794 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
25795 FMA_EXPR like tcc_binary or tcc_unary.
25797 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
25800 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
25802 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
25803 (output_loc_operands): Handle DW_OP_call_ref and
25804 DW_OP_GNU_variable_value.
25805 (struct variable_value_struct): New type.
25806 (struct variable_value_hasher): Likewise.
25807 (variable_value_hash): New variable.
25808 (string_types): Remove.
25809 (copy_loc_descr): New function.
25810 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
25811 (prepend_loc_descr_to_each): New function.
25812 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
25813 instead of add_loc_descr_to_each if the first argument is single
25814 location list and the second has multiple.
25815 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
25816 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
25817 when looking for variable value which doesn't have other location info.
25818 (loc_list_from_tree): Formatting fix.
25819 (gen_array_type_die): Simplify DW_AT_string_length handling.
25820 (adjust_string_types): Remove.
25821 (gen_subprogram_die): Don't call adjust_string_types nor test/set
25822 string_types. Call resolve_variable_values.
25823 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
25824 (resolve_addr_in_expr): Likewise. Add A argument.
25825 (copy_deref_exprloc): Remove deref argument. Adjust for the
25826 original expression being DW_OP_GNU_variable_value with optionally
25827 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
25828 optionally after it.
25829 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
25830 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
25831 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
25832 (variable_value_hasher::hash, variable_value_hasher::equal): New
25834 (resolve_variable_value_in_expr, resolve_variable_value,
25835 resolve_variable_values, note_variable_value_in_expr,
25836 note_variable_value): New functions.
25837 (dwarf2out_early_finish): Call note_variable_value on all toplevel
25840 2017-02-24 Jakub Jelinek <jakub@redhat.com>
25843 * opts.h (handle_generated_option): Add GENERATED_P argument.
25844 * opts-common.c (handle_option): Adjust function comment.
25845 (handle_generated_option): Add GENERATED_P argument, pass it to
25847 (control_warning_option): Pass false to handle_generated_option
25849 * opts.c (maybe_default_option): Pass true to handle_generated_option
25851 * optc-gen.awk: Likewise.
25853 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25855 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
25856 a REG, look at the REG it is a SUBREG of.
25857 (splitter for cmpeqsi_t): Ditto.
25859 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25861 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
25862 the special USEs with the pattern of the insn, not the insn itself.
25864 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
25867 * doc/invoke.texi: Document -mload-store-pairs.
25869 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25870 Sandra Loosemore <sandra@codesourcery.com>
25872 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
25873 argument isn't a CONST_INT.
25874 (nios2_alternate_compare_const): Assert op is a CONST_INT.
25875 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
25876 (nios2_validate_compare): Bypass alternate compare logic if *op2
25877 is not a CONST_INT.
25878 (ldstwm_operation_p): Return false if first_base is not a REG or
25879 if first_offset is not a CONST_INT.
25881 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25883 * config/cris/cris.md: Use correct operand in a define_peephole2.
25885 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25887 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
25889 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25891 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
25892 this_insn if it is an INSN or JUMP_INSN.
25893 (force_offsettable): Look at base, not at addr.
25894 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
25895 on things that aren't necessarily CONST_INTs.
25897 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
25899 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
25900 -mfpmath=sse is the default also for x86-32 targets with SSE2
25901 instruction set when @option{-ffast-math} is enabled
25903 2017-02-24 Jeff Law <law@redhat.com>
25905 PR rtl-optimizatoin/79286
25906 * ira.c (update_equiv_regs): Drop may_trap_p exception to
25909 2017-02-24 Richard Biener <rguenther@suse.de>
25911 PR tree-optimization/79389
25912 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
25915 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
25917 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
25918 function comment to reflect reality.
25919 (loop_exits_before_overflow): Fix typo in function description.
25921 2017-02-24 Richard Biener <rguenther@suse.de>
25923 PR tree-optimization/79389
25924 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
25925 properly that a threading opportunity exists. Detect conditional
25926 copy/constant propagation opportunities.
25928 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
25930 * config/visium/visium.md (type): Add trap.
25931 (b): New mode attribute.
25932 (*btst): Rename into...
25933 (*btst<mode>): ...this and adjust.
25934 (*cbranchsi4_btst_insn): Rename into...
25935 (*cbranch<mode>4_btst_insn): ...this and adjust.
25936 (trap): New define_insn.
25938 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25940 PR tree-optimization/79389
25941 * ifcvt.c (struct noce_if_info): Add rev_cond field.
25942 (noce_reversed_cond_code): New function.
25943 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
25944 reversed_comparison_code. Formatting fix.
25945 (noce_try_store_flag): Test rev_cond != NULL in addition to
25946 reversed_comparison_code.
25947 (noce_try_store_flag_constants): Likewise.
25948 (noce_try_store_flag_mask): Likewise.
25949 (noce_try_addcc): Use rev_cond if non-NULL instead of
25950 reversed_comparison_code.
25951 (noce_try_cmove_arith): Likewise. Formatting fixes.
25952 (noce_try_minmax, noce_try_abs): Clear rev_cond.
25953 (noce_find_if_block): Initialize rev_cond.
25954 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
25955 instead of false as last argument never attempt to reverse it
25958 2017-02-23 Bin Cheng <bin.cheng@arm.com>
25960 PR tree-optimization/79663
25961 * tree-predcom.c (combine_chains): Process refs in reverse order
25962 only for ZERO length chains, and add explaining comment.
25964 2017-02-23 Jeff Law <law@redhat.com>
25966 PR tree-optimization/79578
25967 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
25968 in call to operand_equal_p.
25970 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
25973 * config/i386/cpuid.h: Fix another undefined behavior.
25975 2017-02-23 Richard Biener <rguenther@suse.de>
25977 PR tree-optimization/79683
25978 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
25979 vector types for data-refs.
25981 2017-02-23 Martin Liska <mliska@suse.cz>
25983 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
25985 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25987 PR middle-end/79665
25988 * internal-fn.c (get_range_pos_neg): Moved to ...
25989 * tree.c (get_range_pos_neg): ... here. No longer static.
25990 * tree.h (get_range_pos_neg): New prototype.
25991 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
25992 are known to be in between 0 and signed maximum inclusive, try to
25993 expand both unsigned and signed divmod and use the cheaper one from
25996 2017-02-22 Jeff Law <law@redhat.com>
25998 PR tree-optimization/79578
25999 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
26000 to compare base operands.
26002 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
26005 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
26006 gpc_reg_operand instead of fpr_reg_operand.
26008 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
26010 * config/mips/mips.c (mips_return_in_memory): Force FP
26011 vector types to be returned in memory for o32 ABI.
26013 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26015 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
26016 instead of DW_TAG_member for static data member declarations and don't
26017 set no_linkage_name for static inline data members.
26018 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
26021 2017-02-22 Martin Liska <mliska@suse.cz>
26023 * doc/invoke.texi: Replace inequality signs with square brackets
26026 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26028 PR tree-optimization/68644
26029 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
26031 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
26034 * lra-constraints.c (simplify_operand_subreg): Handle
26035 WORD_REGISTER_OPERATIONS targets.
26037 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26040 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
26041 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
26042 elimination by swapping fld*.
26044 2017-02-22 Richard Biener <rguenther@suse.de>
26046 PR tree-optimization/79673
26047 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
26048 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
26049 irrelevant address-space qualifiers and avoiding a
26050 ADDR_SPACE_CONVERT_EXPR from fold_convert.
26052 2017-02-22 Richard Biener <rguenther@suse.de>
26054 PR tree-optimization/79666
26055 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
26056 to not symbolically negate if that may introduce undefined
26059 2017-02-22 Martin Liska <mliska@suse.cz>
26062 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
26063 * data-streamer-out.c (streamer_write_gcov_count_stream):
26065 * value-prof.c (stream_out_histogram_value): Make assert more
26066 precise based on type of counter.
26068 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
26071 * config/i386/i386.md (standard_x87sse_constant_load splitter):
26072 Use nonimmediate_operand instead of memory_operand for operand 1.
26073 (float-extend standard_x87sse_constant_load splitter): Ditto.
26075 2017-02-21 Jeff Law <law@redhat.com>
26077 PR tree-optimization/79621
26078 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
26079 blocks with edges to themselves.
26081 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26084 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
26085 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
26086 Use gimple_call_builtin_p.
26089 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
26090 on temporarily removed DEBUG_INSNs.
26092 PR tree-optimization/79649
26093 * tree-loop-distribution.c (classify_partition): Give up on
26094 non-generic address space loads/stores.
26096 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
26098 * doc/loop.texi (Loop manipulation): Remove nonexistent
26099 tree_ssa_loop_version from the documentation.
26100 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
26102 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26105 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
26106 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
26107 * config/rs6000/rs6000.c: Include except.h.
26108 (rs6000_expand_split_stack_prologue): Call
26109 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
26111 2017-02-21 Martin Jambor <mjambor@suse.cz>
26114 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
26115 have been analyzed.
26117 2017-02-21 Martin Jambor <mjambor@suse.cz>
26119 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
26120 for backward compatibility only.
26121 * doc/invoke.texi (Option Summary): Remove all references to
26122 -fipa-cp-alignment.
26124 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
26128 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26130 * lra-constraints.c (curr_insn_transform): Handle
26131 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26133 2017-02-21 Martin Liska <mliska@suse.cz>
26135 * config/i386/i386.opt: Replace -masm-dialect with -masm.
26137 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
26139 PR translation/79638
26140 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
26142 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
26145 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
26146 (arm_function_ok_for_sibcall): Return false for an indirect call by
26147 descriptor if all the argument registers are used.
26148 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
26149 alignment of the function.
26151 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26153 PR tree-optimization/61441
26154 * simplify-rtx.c (simplify_const_unary_operation): For
26155 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
26156 the sNaN unmodified.
26158 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26160 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26161 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26162 instead of SYSTEM_HEADER_DIR.
26164 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
26165 Martin Liška <mliska@suse.cz>
26167 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
26168 Fix typos and grammar, use active voice, and clarify.
26170 2017-02-20 Marek Polacek <polacek@redhat.com>
26172 PR middle-end/79537
26173 * gimplify.c (gimplify_expr): Handle unused *&&L;.
26176 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
26178 2017-02-20 Jakub Jelinek <jakub@redhat.com>
26181 * config/i386/i386.c (ix86_expand_builtin): Handle
26182 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
26183 ix86_builtins_isa[fcode].isa as a requirement of those
26184 flags and any other flag in the bitmask.
26185 (ix86_init_mmx_sse_builtins): Use 0 instead of
26186 ~OPTION_MASK_ISA_64BIT as mask.
26187 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
26188 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
26189 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
26190 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
26192 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26195 * lra-constraints.c (split_reg): Check requested split mode
26196 is supported by the register.
26198 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26200 * lra-constraints.c (simplify_operand_subreg): Remove early
26203 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26206 * lra-constraints.c (curr_insn_transform): Tighten condition
26207 for converting SUBREG reloads from OP_OUT to OP_INOUT.
26209 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26212 * lra-constraints.c (curr_insn_transform): Handle
26213 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26215 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
26218 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
26220 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
26222 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
26225 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
26228 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26230 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
26233 2017-02-18 Jakub Jelinek <jakub@redhat.com>
26235 * final.c (last_columnnum, override_columnnum): New variables.
26236 (final_start_function): Set last_columnnum, pass it to begin_prologue
26237 hook and pass 0 to dwarf2out_begin_prologue.
26238 (final_scan_insn): Update override_columnnum. Pass last_columnnum
26239 to source_line debug hook.
26240 (notice_source_line): Compute last_columnnum and for debug_column_info
26241 return true on column changes.
26242 * debug.h (struct gcc_debug_hooks): Add column argument to
26243 source_line and begin_prologue hooks.
26244 (debug_nothing_int_charstar_int_bool): Remove prototype.
26245 (debug_nothing_int_int_charstar,
26246 debug_nothing_int_int_charstar_int_bool): New prototypes.
26247 (dwarf2out_begin_prologue): Add column argument.
26248 * debug.c (do_nothing_debug_hooks): Adjust source_line and
26249 begin_prologue hooks.
26250 (debug_nothing_int_charstar_int_bool): Remove.
26251 (debug_nothing_int_int_charstar,
26252 debug_nothing_int_int_charstar_int_bool): New functions.
26253 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
26254 through to dwarf2out_source_line.
26255 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
26256 (dwarf2out_source_line): Add column argument, emit it if requested.
26257 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
26259 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26260 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26261 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
26262 through to dwarf2out_begin_prologue.
26263 (vmsdbgout_source_line): Add column argument, pass it through to
26264 dwarf2out_source_line.
26265 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
26266 dbxout_source_line caller.
26267 (dbxout_source_line): Add column argument.
26269 * common.opt (gno-column-info, gcolumn-info): New options.
26270 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
26271 (check_die): Also test for multiple DW_AT_decl_column attributes.
26272 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
26273 DW_AT_decl_column if requested.
26274 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
26276 (gen_variable_die): Likewise.
26277 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
26278 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
26281 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
26282 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
26283 (ix86_handle_option): Handle OPT_m3dnowa.
26284 * doc/invoke.texi (-m3dnowa): Document.
26285 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
26286 -m3dnowa instead of -m3dnow -march=athlon.
26289 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
26290 instead of gcc_assert for K, r and R code checks. Formatting fixes.
26292 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26295 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
26296 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
26297 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
26298 generator for vsx_xxpermdi_<mode>_be.
26299 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
26300 force big-endian semantics.
26301 (vsx_xxpermdi_<mode>_be): New define_expand with same
26302 implementation as previous version of vsx_xxpermdi_<mode>.
26304 2017-02-17 Jakub Jelinek <jakub@redhat.com>
26306 PR tree-optimization/79327
26307 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
26308 variable, its initialization and use.
26310 2017-02-17 Julia Koval <julia.koval@intel.com>
26312 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
26313 (OPTION_MASK_ISA_PKU_UNSET): New.
26314 (ix86_handle_option): Handle -mrdpid.
26315 * config/i386/cpuid.h (bit_RDPID): New.
26316 * config/i386/driver-i386.c (host_detect_local_cpu):
26317 Detect RDPID feature.
26318 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
26319 * config/i386/i386-c.c (ix86_target_macros_internal):
26321 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
26322 (ix86_valid_target_attribute_inner_p): Add "rdpid".
26323 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26324 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
26325 * config/i386/i386.md (define_insn "rdpid"): New.
26326 * config/i386/i386.opt Add -mrdpid.
26327 * config/i386/immintrin.h (_rdpid_u32): New.
26329 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
26331 PR rtl-optimization/79541
26332 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
26333 instead of transforming it into USE.
26335 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
26337 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
26338 If HONOR_SNANS (SFmode) force the input to a register.
26339 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
26340 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
26341 an frsp or similar insn.
26343 2017-02-17 Martin Liska <mliska@suse.cz>
26345 PR rtl-optimization/79577
26346 * params.def (selsched-max-sched-times): Increase minimum to 1.
26348 2017-02-17 Martin Liska <mliska@suse.cz>
26350 PR rtl-optimization/79574
26351 * gcse.c (want_to_gcse_p): Prevent integer overflow.
26353 2017-02-17 Martin Liska <mliska@suse.cz>
26355 PR tree-optimization/79529
26356 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
26357 ssa_defined_default_def_p to handle cases which are implicitly
26359 * tree-ssa.c (ssa_defined_default_def_p): New function.
26360 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
26361 which are implicitly defined.
26362 * tree-ssa.h (ssa_defined_default_def_p): Declare.
26364 2017-02-17 Richard Biener <rguenther@suse.de>
26366 PR middle-end/79576
26367 * params.def (max-ssa-name-query-depth): Limit to 10.
26369 2017-02-17 Richard Biener <rguenther@suse.de>
26371 PR tree-optimization/79552
26372 * tree-ssa-structalias.c (visit_loadstore): Properly verify
26375 2017-02-17 Richard Biener <rguenther@suse.de>
26378 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
26380 2017-02-17 Marek Polacek <polacek@redhat.com>
26382 PR middle-end/79536
26383 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
26384 (fold_negate_expr): New wrapper.
26386 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
26388 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
26389 Correct terminology and de-emphasize pre-standard behavior.
26391 2017-02-16 Alan Modra <amodra@gmail.com>
26393 PR rtl-optimization/79286
26394 * ira.c (def_dominates_uses): New function.
26395 (update_equiv_regs): Don't create an equivalence for insns that
26396 may trap where the register def does not dominate the use.
26398 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
26400 PR rtl-optimization/78127
26401 * lra.c (lra): Call lra_eliminate before finish the loop after
26404 2017-02-16 Richard Biener <rguenther@suse.de>
26406 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
26408 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
26409 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
26410 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
26411 (isl_val_int_from_wi): New function.
26412 (extract_affine_gmp): Rename to ...
26413 (extract_affine_wi): ... this, take a widest_int.
26414 (extract_affine_int): Just wrap extract_affine_wi.
26415 (add_param_constraints): Use isl_val_int_from_wi.
26416 (add_loop_constraints): Likewise, and extract_affine_wi.
26418 2017-02-15 Jeff Law <law@redhat.com>
26420 PR middle-end/79521
26421 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
26422 ira_init_register_move_cost_if_necessary.
26424 2017-02-15 Martin Sebor <msebor@redhat.com>
26426 PR middle-end/32003
26427 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
26428 removed in a prior commit.
26430 2017-02-15 Bin Cheng <bin.cheng@arm.com>
26432 PR tree-optimization/79347
26433 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
26434 counters during peeling.
26436 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
26438 * Makefile.in (site.exp): Remove "set ISLVER".
26440 2017-02-15 Jakub Jelinek <jakub@redhat.com>
26443 * real.c (real_from_integer): Call real_convert even for decimal.
26445 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
26448 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
26450 2017-02-14 Andrew Pinski <apinski@cavium.com>
26452 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
26453 cores and change the partno/implementer to be correct.
26454 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
26455 the 'B" as the implementer.
26456 * config/aarch64/aarch64-tune.md: Regenerate.
26458 2017-02-14 Carl Love <cel@us.ibm.com>
26460 * config/rs6000/rs6000.c: Add case statement entry to make the
26461 xvcvuxdsp built-in argument unsigned.
26462 * config/rs6000/vsx.md: Fix the source and return operand types so they
26463 match the instruction definitions from the ISA document. Fix typo
26464 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
26467 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
26470 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
26471 member early_clobber_alts.
26472 * lra-lives.c (reg_early_clobber_p): New.
26473 (process_bb_lives): Use it.
26474 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
26475 (debug_operand_data): Initialize early_clobber_alts.
26476 (setup_operand_alternative): Set up early_clobber_alts.
26477 (collect_non_operand_hard_regs): Ditto. Pass early clobber
26478 alternatives to new_insn_reg.
26479 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
26481 (lra_update_insn_regno_info): Pass the new arg.
26483 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26485 PR middle-end/79505
26486 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
26487 (new_oacc_loop_raw): Don't clear already cleared fields.
26490 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
26491 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
26492 _mm512_prefetch_i64gather_ps): New inline functions and macros.
26494 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
26497 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
26499 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
26502 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
26503 the extra instruction to the right place to store 128-bit constant
26506 2017-02-14 Martin Sebor <msebor@redhat.com>
26508 PR middle-end/79448
26509 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
26510 warning for strings of unknown length.
26512 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26514 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
26516 2017-02-14 Jeff Law <law@redhat.com>
26519 * ira-costs.c (scan_one_insn): Initialize register move costs
26520 for pseudos seen in USE/CLOBBER insns.
26522 PR tree-optimization/79095
26523 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
26524 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
26525 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
26526 if the operands are known to be not equal, then the resulting range
26528 (intersect_ranges): If the new range is ~[0,0] and the old range is
26529 wide, then prefer ~[0,0].
26530 * tree-vrp.c (overflow_comparison_p_1): New function.
26531 (overflow_comparison_p): New function.
26532 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
26533 if NAME is used in an overflow test.
26534 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
26535 overflow check that can be expressed as an equality test, then adjust
26536 ops to be that equality test.
26538 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26540 * config/s390/s390-builtin-types.def: Remove flags argument.
26541 * config/s390/s390.c (s390_init_builtins): Likewise.
26543 2017-02-14 Martin Liska <mliska@suse.cz>
26545 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
26546 vector. Fix trailing white spaces.
26548 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
26550 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
26553 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26555 PR rtl-optimization/68664
26556 * config/arm/arm.c (arm_sched_can_speculate_insn):
26557 New function. Declare prototype.
26558 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26560 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26562 PR rtl-optimization/68664
26563 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
26565 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26567 2017-02-14 Amit Pawar <amit.pawar@amd.com>
26569 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
26570 max skip bytes for function, loop and jump.
26572 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26574 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
26575 ABS_EXPR for gimple dump.
26577 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26580 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
26582 PR tree-optimization/79408
26583 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
26584 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
26585 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
26586 also if rhs1 is INTEGER_CST.
26588 2017-02-14 Richard Biener <rguenther@suse.de>
26590 PR middle-end/79432
26591 * tree-into-ssa.c (insert_phi_nodes): When the function can
26592 have abnormal edges rewrite SSA names with broken use-def
26593 dominance out of SSA and register them for PHI insertion.
26595 2017-02-13 Martin Sebor <msebor@redhat.com>
26597 PR middle-end/79496
26598 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
26599 clearing info.nowrite flag when snprintf size argument is a range.
26601 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26603 * cprop.c (cprop_jump): Add missing space in string literal.
26604 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
26605 (get_constraint_for_component_ref): Likewise.
26606 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
26607 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
26608 * lra-constraints.c (process_alt_operands): Likewise.
26609 * ipa-inline.c (inline_small_functions): Likewise.
26610 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
26611 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26612 * trans-mem.c (diagnose_tm_1_op): Likewise.
26613 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
26614 (grid_parallel_clauses_gridifiable): Likewise.
26616 * config/nvptx/mkoffload.c (process): Add space in between
26619 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
26620 "MOD4_SSE_REGS" and "ALL_REGS".
26622 * spellcheck.c (test_data): Add , in between "foo" and "food".
26624 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26627 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
26628 boundary crossing check and subsequent code generation agree.
26630 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26632 * config/aarch64/aarch64.c (has_memory_op): Delete.
26633 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
26636 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26638 PR rtl-optimization/79388
26639 PR rtl-optimization/79450
26640 * combine.c (distribute_notes): When removing TEM_INSN for which
26641 corresponding dest has last value recorded, invalidate that last
26644 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26646 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
26647 of explicit '@'. Add missing assembly comment marker on branch costs
26650 2017-02-13 Nathan Sidwell <nathan@acm.org>
26652 * gengtype-lex.l (<in_struct>): Add '/'.
26654 2017-02-13 Martin Liska <mliska@suse.cz>
26657 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
26659 2017-02-13 Richard Biener <rguenther@suse.de>
26661 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
26663 * configure: Re-generate.
26664 * config.in: Likewise.
26665 * graphite-dependences.c: Simplify as if
26666 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
26667 * graphite-isl-ast-to-gimple.c: Likewise.
26668 * graphite-optimize-isl.c: Likewise.
26669 * graphite-poly.c: Likewise.
26670 * graphite-sese-to-poly.c: Likewise.
26671 * graphite.h: Likewise.
26672 * toplev.c: Include isl/version.h and use isl_version () for
26673 printing the ISL version.
26674 * doc/install.texi: Update ISL requirement.
26676 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
26678 * doc/standards.texi (Standards): Update reference to
26681 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
26683 * doc/extend.texi (Named Address Spaces): sourceware.org now
26685 * doc/install.texi (Binaries): Ditto.
26688 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26690 * doc/cpp.texi: Replace "stringify"/"stringification" with C
26691 standard terminology "stringize"/"stringizing" throughout.
26692 * doc/cppinternals.texi: Likewise.
26694 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26696 * doc/extend.texi: Fix some spelling mistakes and typos.
26697 * doc/invoke.texi: Likewise.
26699 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26702 * params.def (inline-min-speedup) Change from 10 to 8.
26704 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26706 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
26709 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26712 * ipa-inline-analysis.c (get_minimal_bb): New function.
26713 (record_modified): Use it.
26714 (remap_edge_change_prob): Handle also ancestor functions.
26716 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
26718 * doc/contrib.texi (Contributors): Remove broken link into
26719 the Mauve CVS repository.
26721 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26723 PR middle-end/79454
26724 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
26725 result computation whenever lhs doesn't have vector mode, not
26726 just when it has BLKmode.
26728 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26730 * doc/makefile.texi (profiledbootstrap): Refer to the
26731 installation instructions only in textual form.
26733 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26736 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
26738 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26740 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
26741 (Specific): Update mingw-w64 reference.
26743 (Specific): Remove broken link to Renesas RX processor.
26745 2017-02-10 Richard Biener <rguenther@suse.de>
26747 * toplev.c (process_options): Do not mention obsolete graphite
26748 options when printing sorry message about missing graphite support.
26749 Mention -floop-nest-optimize.
26751 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
26753 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
26754 (vtst_p16): Likewise.
26755 (vtstq_p8): Likewise.
26756 (vtstq_p16): Likewise.
26758 (vtstq_p64): Likewise.
26759 * config/arm/arm_neon.h (vgetq_lane_p64): New.
26760 (vset_lane_p64): New.
26761 (vsetq_lane_p64): New.
26763 2017-02-10 Jakub Jelinek <jakub@redhat.com>
26765 PR tree-optimization/79411
26766 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
26767 stmt operands are SSA_NAMEs used in abnormal phis.
26768 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
26771 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26774 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
26777 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26779 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
26781 2017-02-09 Jakub Jelinek <jakub@redhat.com>
26783 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
26787 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
26788 not arbitrary TREE_CONSTANT.
26791 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
26792 "omp declare target link" attribute unless is_global_var.
26793 * omp-offload.c (find_link_var_op): Likewise.
26795 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
26796 Chung-Lin Tang <cltang@codesourcery.com>
26798 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
26800 (gimplify_adjust_omp_clauses): Don't delete TILE.
26801 (gimplify_omp_for): Deal with TILE.
26802 * internal-fn.c (expand_GOACC_TILE): New function.
26803 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
26805 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
26806 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
26808 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
26809 avoid DIV for outermost collapse var.
26810 (expand_oacc_for): Insert tile element loop as needed. Adjust.
26811 Remove out of date comments, fix whitespace.
26812 * omp-general.c (omp_extract_for_data): Deal with tiling.
26813 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
26814 adjust OLF_DIM_BASE value.
26815 (struct omp_for_data): Add tiling field.
26816 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
26817 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
26818 for auto loops. Remove default auto determining, moved to
26819 oacc_loop_fixed_partitions.
26820 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
26821 stmts, add e_mask field.
26822 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
26823 (oacc_thread_numbers): Use oacc_dim_call.
26824 (oacc_xform_tile): New.
26825 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
26826 (finish_oacc_loop): Adjust for ifns vector.
26827 (oacc_loop_discover_walk): Append loop abstraction sites to list,
26828 add case for GOACC_TILE fns.
26829 (oacc_loop_xform_loop): Delete.
26830 (oacc_loop_process): Iterate over call list directly, and add
26831 handling for GOACC_TILE fns.
26832 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
26834 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
26835 vector partitioning to outer loops. Assign 2 partitions to loops
26836 when available. Add TILE handling.
26837 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
26838 (execite_oacc_device_lower): Process GOACC_TILE fns,
26839 ignore unknown specs.
26840 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
26841 * tree.c (omp_clause_num_ops): Adjust TILE ops.
26842 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
26844 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
26846 * configure.ac (ACX_BUGURL): Update.
26847 * configure: Regenerate.
26849 2017-02-09 Richard Biener <rguenther@suse.de>
26851 PR tree-optimization/69823
26852 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
26853 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
26855 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26857 * config/arc/arc-c.def: Add __NPS400__ definition.
26858 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
26859 (TARGET_NPS400): Define.
26861 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26863 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
26865 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
26866 pointer, arch_info.
26867 (arc_cpu_types): Fill the arch_info field with a pointer into the
26868 arc_arch_types table.
26869 (arc_selected_cpu): Declare.
26870 * config/arc/arc.c (arc_selected_cpu): Make global.
26871 (arc_selected_arch): Delete.
26872 (arc_base_cpu): Delete.
26873 (arc_override_options): Remove references to deleted variables,
26874 update access to arch information.
26875 (ARC_OPT): Update access to arch information.
26876 (ARC_OPTX): Likewise.
26877 * config/arc/arc.h (arc_base_cpu): Remove declaration.
26878 (TARGET_ARC600): Update access to arch information.
26879 (TARGET_ARC601): Likewise.
26880 (TARGET_ARC700): Likewise.
26881 (TARGET_EM): Likewise.
26882 (TARGET_HS): Likewise.
26883 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
26886 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
26889 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
26890 condition/operands for integer GE/LE/GEU/LEU operations.
26892 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
26894 PR translation/79397
26895 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
26898 2017-02-08 Martin Jambor <mjambor@suse.cz>
26901 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
26902 whether allocation happened.
26903 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
26904 nothing was allocated.
26906 2017-02-08 Jakub Jelinek <jakub@redhat.com>
26908 PR tree-optimization/79408
26909 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
26910 constant, but SSA_NAME with a known integer range, use the minimum
26911 of that range instead of op1 to determine if modulo can be replaced
26912 with its first operand.
26914 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26916 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
26918 2017-02-08 Richard Biener <rguenther@suse.de>
26920 PR tree-optimization/71824
26921 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26922 Check all loops contained in the merged region.
26924 2017-02-07 Andrew Pinski <apinski@cavium.com>
26926 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
26928 2017-02-07 Andrew Pinski <apinski@cavium.com>
26930 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
26931 (thunderxt88): Likewise.
26932 (thunderxt81): Disable LSE and change v8.1 to v8.
26933 (thunderxt83): Likewise.
26935 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26936 Richard Biener <rguenther@suse.de>
26938 PR middle-end/79399
26939 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
26940 type from int to size_t.
26941 * ira-costs.c (struct_costs_size): Change type from int to size_t.
26943 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26945 PR rtl-optimization/79386
26946 * cprop.c (bypass_conditional_jumps): Initialize
26947 bypass_last_basic_block already before splitting bbs after
26948 unconditional traps...
26949 (bypass_conditional_jumps): ... rather than here.
26952 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
26953 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
26954 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
26955 fix -masm=intel patterns.
26957 2017-02-07 Richard Biener <rguenther@suse.de>
26959 PR tree-optimization/79256
26960 PR middle-end/79278
26961 * builtins.c (get_object_alignment_2): Use min_align_of_type
26962 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
26963 and ADJUST_FIELD_ALIGN.
26965 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
26967 * doc/tm.texi: Regenerate.
26968 * stor-layout.c (layout_decl): Adjust.
26969 (update_alignment_for_field): Likewise.
26970 (place_field): Likewise.
26971 (min_align_of_type): Likewise.
26972 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
26973 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
26974 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
26975 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
26976 * config/frv/frv.c (frv_adjust_field_align): Likewise.
26977 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
26978 * config/i386/i386.c (x86_field_alignment): Likewise.
26979 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
26980 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
26981 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
26982 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
26983 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
26984 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
26988 2017-01-30 Richard Biener <rguenther@suse.de>
26990 PR tree-optimization/79256
26991 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26992 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26995 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
26997 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
26998 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
26999 builtins to SImode and emit a zero-extend, if necessary.
27001 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27003 * docs/invoke.texi (RISC-V Options): Alphabetize.
27005 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27007 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
27010 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27012 * config/riscv/riscv.c: New file.
27013 * common/config/riscv/riscv-common.c: Likewise.
27014 * config.gcc: Likewise.
27015 * config/riscv/constraints.md: Likewise.
27016 * config/riscv/elf.h: Likewise.
27017 * config/riscv/generic.md: Likewise.
27018 * config/riscv/linux.h: Likewise.
27019 * config/riscv/multilib-generator: Likewise.
27020 * config/riscv/peephole.md: Likewise.
27021 * config/riscv/pic.md: Likewise.
27022 * config/riscv/predicates.md: Likewise.
27023 * config/riscv/riscv-builtins.c: Likewise.
27024 * config/riscv/riscv-c.c: Likewise.
27025 * config/riscv/riscv-ftypes.def: Likewise.
27026 * config/riscv/riscv-modes.def: Likewise.
27027 * config/riscv/riscv-opts.h: Likewise.
27028 * config/riscv/riscv-protos.h: Likewise.
27029 * config/riscv/riscv.h: Likewise.
27030 * config/riscv/riscv.md: Likewise.
27031 * config/riscv/riscv.opt: Likewise.
27032 * config/riscv/sync.md: Likewise.
27033 * config/riscv/t-elf-multilib: Likewise.
27034 * config/riscv/t-linux: Likewise.
27035 * config/riscv/t-linux-multilib: Likewise.
27036 * config/riscv/t-riscv: Likewise.
27037 * configure.ac: Likewise.
27038 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
27039 Waterman as RISC-V maintainers.
27040 * doc/install.texi: Add RISC-V entries.
27041 * doc/invoke.texi: Add RISC-V options section.
27042 * doc/md.texi: Add RISC-V constraints section.
27043 * configure: Regenerated.
27045 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
27048 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
27049 false values to be constant vectors with all 0 or all 1 bits set.
27050 (vcondu<mode><mode>): Likewise.
27051 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
27053 (fpmask_comparison_operator): Update comment.
27054 (vecint_comparison_operator): New predicate.
27055 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
27056 vector conditionals when the true and false values are constant
27057 vectors with all 0 bits or all 1 bits set.
27059 2017-02-06 Martin Sebor <msebor@redhat.com>
27061 PR tree-optimization/79376
27062 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
27064 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
27066 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
27067 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
27068 to simplify split condition.
27070 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27072 * omp-expand.c (oxpand_omp_atomic_fetch_op,
27073 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
27076 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
27078 PR rtl-optimization/68664
27079 * target.def (can_speculate_insn): New hook.
27080 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
27081 * doc/tm.texi: Regenerate.
27082 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
27083 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
27084 (rs6000_sched_can_speculate_insn): New function.
27086 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27088 PR tree-optimization/79284
27089 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
27090 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
27091 vectorizable_mask_load_store, vectorizable_operation,
27092 vect_is_simple_cond, get_same_sized_vectype): Use it instead
27093 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
27094 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
27095 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
27096 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
27097 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27098 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
27099 is_gimple_assign (stmt). Replace another such test with
27100 is_gimple_assign (stmt).
27102 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27105 * config/avr/avr.c (rtl-iter.h): Include it.
27106 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
27107 (avr_legitimate_combined_insn): ...and implementation.
27109 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
27111 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
27112 * config/s390/s390.c (s390_const_operand_ok)
27113 (s390_canonicalize_comparison, s390_extract_part)
27114 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
27115 (s390_contiguous_bitmask_p, s390_rtx_costs)
27116 (legitimize_pic_address): Likewise.
27117 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
27118 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
27119 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
27120 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
27121 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
27123 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27125 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
27126 REGNO($0) == REGNO($1).
27128 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27130 * config/s390/linux.h(SIZE_TYPE): Add comment.
27132 2017-02-06 Julian Brown <julian@codesourcery.com>
27133 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27134 Virendra Pathak <virendra.pathak@broadcom.com>
27136 * config/aarch64/aarch64-cores.def: Change the scheduler
27138 * config/aarch64/aarch64.md: Include thunderx2t99.md.
27139 * config/aarch64/thunderx2t99.md: New file.
27141 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27143 * doc/standards.texi (Go Language): Update link to language
27146 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27148 * tree-eh.c (lower_resx): Sanitize profile.
27149 (cleanup_empty_eh_move_lp): Likewise.
27151 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27154 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
27156 * cfgloopmanip.h (loop_version): Update prototype.
27157 * modulo-sched.c (sms_schedule): Update call of loop_version.
27158 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
27159 * tree-parloops.c (gen_parallel_loop): Likewise.
27160 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
27161 * tree-ssa-loop-split.c (split_loop): Likewise.
27162 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
27163 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
27165 2017-02-05 Martin Liska <mliska@suse.cz>
27168 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
27170 (print_operand_address): Initialize a struct to zero.
27172 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27174 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
27175 garbage collector only in textual form.
27177 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27179 * doc/extend.texi (x86 specific memory model extensions for
27180 transactional memory): Simplify a phrase.
27182 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
27185 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
27186 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
27187 (atomic_storedi_1): Likewise.
27189 2017-02-04 Jakub Jelinek <jakub@redhat.com>
27191 PR tree-optimization/79338
27192 * tree-parloops.c (gather_scalar_reductions): Don't call
27193 vect_analyze_loop_form for loop->inner before destroying loop's
27196 2017-02-03 Martin Sebor <msebor@redhat.com>
27198 PR tree-optimization/79327
27199 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
27200 when precision has resulted in leading zeros.
27201 (format_integer): Adjust the likely counter to assume an unknown
27202 argument that may be zero is non-zero.
27204 2017-02-03 Jason Merrill <jason@redhat.com>
27207 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
27208 avoid copying non-taken branch.
27210 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27212 PR tree-optimization/79340
27213 * tree-vect-loop.c (vectorizable_reduction): Release
27214 vec_defs elements after safe_splicing them into other vectors.
27217 PR tree-optimization/79327
27218 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
27219 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
27221 (format_integer): Use wide_int_to_tree instead of build_int_cst
27222 + to_?hwi. If argmin is NULL, just set argmin and argmax to
27223 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
27224 of shortest and longest sequence.
27226 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
27228 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
27229 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
27231 2017-02-03 Walter Lee <walt@tilera.com>
27234 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
27235 after initial stackframe link reg save.
27236 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
27238 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27241 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
27242 wu for stxssp alternative.
27244 2017-02-03 Martin Sebor <msebor@redhat.com>
27246 PR tree-optimization/79352
27247 * gimple-fold.c (get_range_strlen): Add argument.
27248 (get_range_strlen): Change return type to bool.
27249 (get_maxval_strlen): Pass in a dummy argument.
27250 * gimple-fold.h (get_range_strlen): Change return type to bool.
27251 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
27252 * tree.h (array_at_struct_end_p): Add argument.
27253 * tree.c (array_at_struct_end_p): Handle it.
27255 2017-02-03 Martin Liska <mliska@suse.cz>
27258 * multiple_target.c (create_dispatcher_calls): Redirect edge
27259 from a caller of a dispatcher.
27260 (expand_target_clones): Make the clones local.
27261 (ipa_target_clone): Do both target clones and resolvers.
27262 (ipa_dispatcher_calls): Remove the pass.
27263 (pass_dispatcher_calls::gate): Likewise.
27264 (make_pass_dispatcher_calls): Likewise.
27265 * passes.def (pass_target_clone): Put as very first IPA early
27268 2017-02-03 Martin Liska <mliska@suse.cz>
27270 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
27271 in case of a function with ifunc attribute.
27273 2017-02-03 Martin Liska <mliska@suse.cz>
27275 * cgraph.c (cgraph_node::dump): Dump function version info.
27276 * symtab.c (symtab_node::dump_base): Add missing new line.
27278 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27280 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
27281 (ifcombine_ifandif): Use it.
27283 2017-02-03 Martin Liska <mliska@suse.cz>
27285 * doc/invoke.texi: Document default value for
27286 use-after-scope-direct-emission-threshold.
27288 2017-02-03 Martin Liska <mliska@suse.cz>
27290 PR tree-optimization/79339
27291 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
27292 (format_floating): Likewise.
27294 2017-02-03 Martin Liska <mliska@suse.cz>
27297 * ipa-prop.c (ipa_node_params_t::insert): Remove current
27299 (ipa_node_params_t::remove): Likewise.
27300 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
27301 initialization from removed ipa_node_params_t::insert.
27302 (ipa_node_params::~ipa_node_params): Move from removed
27303 ipa_node_params_t::release.
27304 * symbol-summary.h (symbol_summary::m_released): New member.
27305 Do not release a summary twice. Do not allow to call finalizer
27306 for types of a summary that live in GGC memory.
27308 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27310 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
27313 2017-02-02 Martin Sebor <msebor@redhat.com>
27315 PR middle-end/79275
27316 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
27317 (format_string): Tighten up the range of output for non-constant
27318 strings and correct the expected range for wide non-constant strings.
27320 2017-02-02 Martin Sebor <msebor@redhat.com>
27322 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
27324 PR middle-end/32003
27325 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
27327 (-fdump-tree-@var): Add to index and document how to come up
27328 with pass-specific option and dump file names.
27329 (-fdump-passes): Clarify where to look for output.
27331 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27333 PR middle-end/77445
27334 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
27335 statistics of the analyzed path; allow threading for speed when
27336 any of BBs along the path are optimized for speed.
27338 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
27340 PR middle-end/78468
27341 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
27342 settings of the virtual registers.
27345 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
27347 * explow.c (get_dynamic_stack_size): Take known alignment of stack
27348 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
27351 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27353 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
27354 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
27356 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27358 * config/s390/s390.md: Add missing comments with the expanded
27360 * config/s390/vector.md: Likewise.
27361 * config/s390/vx-builtins.md: Likewise.
27363 2017-02-02 Jakub Jelinek <jakub@redhat.com>
27366 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
27367 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
27368 conditions on a single line.
27370 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27372 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27373 __S390_VX__ to __VX__.
27375 2017-02-01 Andrew Pinski <apinski@cavium.com>
27377 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
27378 stmt_info to record_stmt_cost.
27379 (vect_get_known_peeling_cost): Pass stmt_info if known to
27381 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
27382 cpu_vector_cost field into
27383 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
27384 field into vec_int_stmt_cost and vec_fp_stmt_cost.
27385 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
27386 splitting of scalar_stmt_cost and vec_stmt_cost.
27387 (thunderx_vector_cost): Likewise.
27388 (cortexa57_vector_cost): LIkewise.
27389 (exynosm1_vector_cost): Likewise.
27390 (xgene1_vector_cost): Likewise.
27391 (thunderx2t99_vector_cost): Improve after the splitting of the two
27393 (aarch64_builtin_vectorization_cost): Update for the splitting of
27394 scalar_stmt_cost and vec_stmt_cost.
27396 2017-02-01 Torvald Riegel <triegel@redhat.com>
27397 Richard Henderson <rth@redhat.com>
27399 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
27400 conditional on existance of a fast atomic load.
27401 * optabs-query.c (can_atomic_load_p): New function.
27402 * optabs-query.h (can_atomic_load_p): Declare it.
27403 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
27404 no fast atomic load is available for the particular size of access.
27405 (expand_atomic_compare_and_swap): Likewise.
27406 (expand_atomic_load): Likewise.
27407 (expand_atomic_store): Likewise.
27408 (expand_atomic_fetch_op): Likewise.
27409 * testsuite/lib/target-supports.exp
27410 (check_effective_target_sync_int_128): Remove x86 because it provides
27411 no fast atomic load.
27412 (check_effective_target_sync_int_128_runtime): Likewise.
27414 2017-02-01 Richard Biener <rguenther@suse.de>
27416 * graphite.c: Include tree-vectorizer.h for find_loop_location.
27417 (graphite_transform_loops): Provide opt-info for optimized nests.
27418 * tree-parloop.c (parallelize_loops): Provide opt-info for
27419 parallelized loops.
27421 2017-02-01 Richard Biener <rguenther@suse.de>
27423 PR middle-end/79315
27424 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
27425 was not set before.
27427 2017-02-01 Richard Biener <rguenther@suse.de>
27429 PR tree-optimization/71824
27430 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27431 Verify the loops are valid in the merged SESE region.
27432 (scop_detection::can_represent_loop_1): Check analyzing the
27433 evolution of the number of iterations in the region succeeds.
27435 2017-01-31 Ian Lance Taylor <iant@golang.org>
27437 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
27438 REG_ARGS_SIZE note to 32-bit push insns and call insn.
27440 2017-01-31 David Malcolm <dmalcolm@redhat.com>
27442 PR preprocessor/79210
27443 * input.c (get_substring_ranges_for_loc): Replace line_width
27444 assertion with error-handling.
27446 2017-01-31 Richard Biener <rguenther@suse.de>
27448 PR tree-optimization/77318
27449 * graphite-sese-to-poly.c (extract_affine): Fix assert.
27450 (create_pw_aff_from_tree): Take loop parameter.
27451 (add_condition_to_pbb): Pass loop of the condition to
27452 create_pw_aff_from_tree.
27454 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27456 * config/s390/s390.c (s390_asan_shadow_offset): New function.
27457 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
27459 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
27463 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
27465 (convert_int_to_float128): Likewise.
27466 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
27467 (convert_int_to_float128): Likewise.
27468 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
27469 (UNSPEC_IEEE128_CONVERT): Likewise.
27470 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
27471 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
27472 Use local variables for IBM extended format.
27473 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
27474 (fix_trunc<mode>si2_fprs): Likewise.
27475 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
27476 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
27477 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
27478 to know that we can now have integers of all sizes in vector
27480 (fix<uns>_<mode>di2_hw): Likewise.
27481 (float<uns>_<mode>si2_hw): Likewise.
27482 (fix_<mode>si2_hw): Likewise.
27483 (fixuns_<mode>si2_hw): Likewise.
27484 (float<uns>_<mode>di2_hw): Likewise.
27485 (float_<mode>di2_hw): Likewise.
27486 (float_<mode>si2_hw): Likewise.
27487 (floatuns_<mode>di2_hw): Likewise.
27488 (floatuns_<mode>si2_hw): Likewise.
27489 (xscvqp<su>wz_<mode>): Delete, no longer used.
27490 (xscvqp<su>dz_<mode>): Likewise.
27491 (xscv<su>dqp_<mode>): Likewise.
27492 (ieee128_mfvsrd_64bit): Likewise.
27493 (ieee128_mfvsrd_32bit): Likewise.
27494 (ieee128_mfvsrwz): Likewise.
27495 (ieee128_mtvsrw): Likewise.
27496 (ieee128_mtvsrd_64bit): Likewise.
27497 (ieee128_mtvsrd_32bit): Likewise.
27499 2017-01-31 Martin Liska <mliska@suse.cz>
27502 * ipa-prop.c (ipa_free_all_node_params): Call release method
27503 instead of ~sumbol_summary to not to trigger double times
27506 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
27508 PR tree-optimization/71691
27509 * bitmap.h (class auto_bitmap): New.
27510 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
27511 is_maybe_undefined instead of ssa_undefined_value_p.
27513 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27515 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27516 __S390_ARCH_LEVEL__ to __ARCH__.
27518 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27520 PR tree-optimization/79267
27521 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
27522 if should_remove_lhs_p is true.
27524 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
27527 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
27528 (add_alignment_attribute): New.
27529 (base_type_die): Add alignment attribute.
27530 (subrange_type_die): Likewise.
27531 (modified_type_die): Likewise.
27532 (gen_array_type_die): Likewise.
27533 (gen_descr_array_type_die: Likewise.
27534 (gen_enumeration_type_die): Likewise.
27535 (gen_subprogram_die): Likewise.
27536 (gen_variable_die): Likewise.
27537 (gen_field_die): Likewise.
27538 (gen_ptr_to_mbr_type_die): Likewise.
27539 (gen_struct_or_union_type_die): Likewise.
27540 (gen_subroutine_type_die): Likewise.
27541 (gen_typedef_die): Likewise.
27542 (base_type_cmp): Compare alignment attribute.
27544 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27547 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
27548 (setb_unsigned) New pattern for setb with CCUNS.
27549 * config/rs6000/rs6000.c (expand_block_compare): Use a different
27550 subfc./subfe sequence to avoid overflow problems. Generate a
27551 shorter sequence with cmpld/setb for power9.
27552 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
27553 for generating subfc. instruction.
27554 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
27555 now uses this instruction.
27557 2017-01-30 Ian Lance Taylor <iant@google.com>
27560 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
27561 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
27563 2017-01-30 Martin Sebor <msebor@redhat.com>
27565 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
27566 Move constant to the right of a relational operator.
27567 (get_mpfr_format_length, format_character, format_string): Ditto.
27568 (should_warn_p, maybe_warn): Same.
27570 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
27572 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
27575 * asan.c (get_translation_unit_decl): Remove function.
27576 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
27578 2017-01-30 Martin Liska <mliska@suse.cz>
27580 PR gcov-profile/79259
27581 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
27582 -fprofile-generate.
27584 2017-01-30 Martin Liska <mliska@suse.cz>
27587 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
27588 Initialize variables with NULL value.
27590 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
27593 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
27595 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
27597 2017-01-30 Richard Biener <rguenther@suse.de>
27599 PR tree-optimization/79276
27600 * tree-vrp.c (process_assert_insertions): Properly adjust common
27601 when removing a duplicate.
27603 * gcc.dg/torture/pr79276.c: New testcase.
27605 2017-01-30 Richard Biener <rguenther@suse.de>
27607 PR tree-optimization/79256
27608 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27609 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27611 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
27613 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
27616 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
27617 ("*r<noxa>sbg_<mode>_sll_bitmask")
27618 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
27619 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
27620 Use contiguous_bitmask_nowrap_operand.
27622 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27625 * config/rs6000/altivec.h (vec_xl): Revise #define.
27626 (vec_xst): Likewise.
27628 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
27630 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
27632 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
27634 PR rtl-optimization/79194
27635 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
27636 traps before call to bypass_conditional_jumps.
27638 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
27640 PR tree-optimization/71374
27641 * lra-constraints.c (check_conflict_input_operands): New.
27642 (match_reload): Use it.
27644 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
27647 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
27648 account to calculate conflict_set.
27650 2017-01-27 Bin Cheng <bin.cheng@arm.com>
27652 PR rtl-optimization/78559
27653 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
27654 other_insn in combine.
27656 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27658 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
27659 uint16_type_node for BT_UINT16.
27661 2017-01-27 David Malcolm <dmalcolm@redhat.com>
27663 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
27664 "RTL Tests" to menu.
27665 (GIMPLE Tests): New node.
27666 (RTL Tests): New node.
27668 2017-01-27 Richard Biener <rguenther@suse.de>
27670 PR tree-optimization/79245
27671 * tree-loop-distribution.c (distribute_loop): Apply cost
27672 modeling also to detected patterns.
27674 2017-01-27 Richard Biener <rguenther@suse.de>
27676 PR tree-optimization/71433
27677 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
27678 (compare_assert_loc): New function.
27679 (process_assert_insertions): Sort and optimize assert locations
27680 to remove duplicates and push down identical assertions on
27681 edges to their destination block.
27683 2017-01-27 Richard Biener <rguenther@suse.de>
27685 PR tree-optimization/79244
27686 * tree-vrp.c (remove_range_assertions): Forcefully propagate
27687 out SSA names even if abnormal.
27689 2017-01-27 Jakub Jelinek <jakub@redhat.com>
27691 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
27692 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
27693 instead of MPFR_RNDN.
27695 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
27698 * arm.c (arm_option_override): Don't call build_target_option_node
27699 until after doing all option overrides.
27700 (arm_valid_target_attribute_tree): Likewise.
27702 2017-01-27 Martin Liska <mliska@suse.cz>
27704 * doc/invoke.texi (-fprofile-arcs): Document profiling support
27705 for {cd}tors and C++ {cd}tors.
27707 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
27709 * config/s390/s390.md ("*setmem_long_and")
27710 ("*setmem_long_and_31z"): Use zero_extend instead of and.
27712 2017-01-26 Martin Sebor <msebor@redhat.com>
27714 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
27717 2017-01-26 Martin Sebor <msebor@redhat.com>
27719 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
27720 HAVE_DFmode before using XFmode or DFmode.
27721 (parse_directive): Avoid using the z length modifier to avoid
27722 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
27724 PR middle-end/78703
27725 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
27726 to accept adjustment as an array.
27727 (get_int_range): New function.
27728 (struct directive): Make width and prec arrays.
27729 (directive::set_width, directive::set_precision): Call get_int_range.
27730 (format_integer, format_floating): Handle width and precision ranges.
27731 (format_string, parse_directive): Same.
27733 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27736 * dwarf2out.c (generate_skeleton_bottom_up): For children with
27737 comdat_type_p set, just clone them, but keep the children in the
27741 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
27742 which have direct callers with -fvar-tracking-assignments enabled
27744 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
27745 inside of type units.
27747 2017-01-26 Martin Sebor <msebor@redhat.com>
27749 PR middle-end/78703
27750 * gimple-ssa-sprintf.c (struct result_range): Add likely and
27752 (struct format_result): Replace number_chars, number_chars_min,
27753 and number_chars_max with a single member of struct result_range.
27755 (format_result::operator+=): Adjust.
27756 (struct fmtresult): Remove bounded. Handle likely and unlikely
27758 (fmtresult::adjust_for_width_or_precision): New function.
27759 (fmtresult:type_max_digits): New function.
27760 (bytes_remaining): Handle likely and unlikely counters.
27761 (min_bytes_remaining): Remove.
27762 (format_percent): Simplify.
27763 (format_integer, format_floating): Set likely and unlikely counters.
27764 (get_string_length, format_character, format_string): Same.
27765 (format_plain, should_warn_p): New function.
27766 (maybe_warn): Call should_warn_p. Update diagnostic messages
27767 and handle those for all directives, including plain strings.
27768 (format_directive): Handle likely and unlikely counters.
27769 Remove unnecessary quoting from diagnostics. Add an informational
27771 (add_bytes): Remove.
27772 (pass_sprintf_length::compute_format_length): Simplify.
27773 (try_substitute_return_value): Handle likely and unlikely counters.
27775 2017-01-26 Carl Love <cel@us.ibm.com>
27777 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
27778 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
27780 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
27783 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
27784 endianess for subregs into account.
27785 * lra-constraints.c (lra_constraints): Do risky transformations
27786 always on the first iteration.
27787 * lra-lives.c (check_pseudos_live_through_calls): Add arg
27788 last_call_used_reg_set.
27789 (process_bb_lives): Define and use last_call_used_reg_set.
27790 * lra.c (lra): Always continue after lra_constraints on the first
27793 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
27795 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
27797 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
27799 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27801 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
27802 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
27803 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
27804 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
27805 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
27806 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
27807 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
27808 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
27809 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
27811 2017-01-26 Marek Polacek <polacek@redhat.com>
27814 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
27815 for the third operand.
27817 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27819 PR middle-end/79236
27820 * omp-low.c (struct omp_context): Add simt_stmt field.
27821 (scan_omp_for): Return omp_context *.
27822 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
27823 context to the _simt_ SIMD stmt.
27824 (lower_omp_for): For combined SIMD with sibling _simt_
27825 SIMD, make sure to use the same decls in _looptemp_
27826 clauses as in the sibling.
27828 2017-01-26 David Sherwood <david.sherwood@arm.com>
27830 PR middle-end/79212
27831 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
27834 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27837 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
27838 emit fld b; fld a; if possible.
27840 * brig-builtins.def: Update copyright years.
27841 * config/arm/arm_acle_builtins.def: Update copyright years.
27843 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
27846 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
27847 constraint instead of o for the stxsd instruction.
27849 2017-01-25 Carl Love <cel@us.ibm.com>
27851 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
27852 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
27854 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
27856 * doc/invoke.texi (C++ Dialect Options): Fix typo.
27858 2017-01-25 Richard Biener <rguenther@suse.de>
27860 PR tree-optimization/69264
27861 * target.def (vector_alignment_reachable): Improve documentation.
27862 * doc/tm.texi: Regenerate.
27863 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
27865 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
27866 earlier changes with respect to TYPE_USER_ALIGN.
27867 (vector_alignment_reachable_p): Likewise. Improve dumping.
27869 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27872 * config/arm/arm.md (xordi3): Force constant operand into a register
27875 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27877 * doc/invoke.texi (-fstore-merging): Correct default optimization
27878 levels at which it is enabled.
27879 (-O): Move -fstore-merging from list to...
27882 2017-01-25 Richard Biener <rguenther@suse.de>
27885 * omp-expand.c: Include debug.h.
27886 (expand_omp_taskreg): Make sure to generate early debug before
27887 outlining anything from a function.
27888 (expand_omp_target): Likewise.
27889 (grid_expand_target_grid_body): Likewise.
27891 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
27894 * asan.c (get_translation_unit_decl): New function.
27895 (asan_add_global): Extract modules file name from globals
27896 TRANSLATION_UNIT_DECL name.
27898 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
27901 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
27902 for long calls with APCS frame and VFP.
27904 2017-01-24 David Malcolm <dmalcolm@redhat.com>
27906 * cfg.c (original_copy_tables_initialized_p): New function.
27907 * cfg.h (original_copy_tables_initialized_p): New decl.
27908 * cfgrtl.c (relink_block_chain): Guard the call to
27909 free_original_copy_tables with a call to
27910 original_copy_tables_initialized_p.
27911 * cgraph.h (symtab_node::native_rtl_p): New decl.
27912 * cgraphunit.c (symtab_node::native_rtl_p): New function.
27913 (symtab_node::needed_p): Don't assert for early assembly output
27914 for __RTL functions.
27915 (cgraph_node::finalize_function): Set "force_output" for __RTL
27917 (cgraph_node::analyze): Bail out early for __RTL functions.
27918 (analyze_functions): Update assertion to support __RTL functions.
27919 (cgraph_node::expand): Bail out early for __RTL functions.
27920 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
27922 * function.h (struct function): Update comment for field
27924 * gimple-expr.c: Include "tree-pass.h".
27925 (gimple_has_body_p): Return false for __RTL functions.
27926 * Makefile.in (OBJS): Add run-rtl-passes.o.
27927 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
27929 (gcc::pass_manager::get_clean_slate): New accessor.
27930 * passes.c: Include "insn-addr.h".
27931 (should_skip_pass_p): Add logging. Update logic for running
27932 "expand" to be compatible with both __GIMPLE and __RTL. Guard
27933 property-provider override so it is only done for gimple passes.
27935 (skip_pass): New function.
27936 (execute_one_pass): Call skip_pass when skipping passes.
27937 * read-md.c (md_reader::read_char): Support filtering
27938 the input to a subset of line numbers.
27939 (md_reader::md_reader): Initialize fields
27940 m_first_line and m_last_line.
27941 (md_reader::read_file_fragment): New function.
27942 * read-md.h (md_reader::read_file_fragment): New decl.
27943 (md_reader::m_first_line): New field.
27944 (md_reader::m_last_line): New field.
27945 * read-rtl-function.c (function_reader::create_function): Only
27946 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
27947 curr_properties. Set DECL_INITIAL to a dummy block.
27948 (read_rtl_function_body_from_file_range): New function.
27949 * read-rtl-function.h (read_rtl_function_body_from_file_range):
27951 * run-rtl-passes.c: New file.
27952 * run-rtl-passes.h: New file.
27954 2017-01-24 Jeff Law <law@redhat.com>
27956 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
27959 2017-01-24 Bin Cheng <bin.cheng@arm.com>
27961 PR tree-optimization/79159
27962 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
27963 (record_nonwrapping_iv): Improve boundary using above function if no
27964 value range information.
27966 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
27967 Martin Jambor <mjambor@suse.cz>
27969 * brig-builtins.def: New file.
27970 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
27971 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
27972 (DEF_HSAIL_SAT_BUILTIN): Likewise.
27973 (DEF_HSAIL_INTR_BUILTIN): Likewise.
27974 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
27975 * builtin-types.def (BT_INT8): New.
27976 (BT_INT16): Likewise.
27977 (BT_UINT8): Likewise.
27978 (BT_UINT16): Likewise.
27979 (BT_FN_ULONG): Likewise.
27980 (BT_FN_UINT_INT): Likewise.
27981 (BT_FN_UINT_ULONG): Likewise.
27982 (BT_FN_UINT_LONG): Likewise.
27983 (BT_FN_UINT_PTR): Likewise.
27984 (BT_FN_ULONG_PTR): Likewise.
27985 (BT_FN_INT8_FLOAT): Likewise.
27986 (BT_FN_INT16_FLOAT): Likewise.
27987 (BT_FN_UINT32_FLOAT): Likewise.
27988 (BT_FN_UINT16_FLOAT): Likewise.
27989 (BT_FN_UINT8_FLOAT): Likewise.
27990 (BT_FN_UINT64_FLOAT): Likewise.
27991 (BT_FN_UINT16_UINT32): Likewise.
27992 (BT_FN_UINT32_UINT16): Likewise.
27993 (BT_FN_UINT16_UINT16_UINT16): Likewise.
27994 (BT_FN_INT_PTR_INT): Likewise.
27995 (BT_FN_UINT_PTR_UINT): Likewise.
27996 (BT_FN_LONG_PTR_LONG): Likewise.
27997 (BT_FN_ULONG_PTR_ULONG): Likewise.
27998 (BT_FN_VOID_UINT64_UINT64): Likewise.
27999 (BT_FN_UINT8_UINT8_UINT8): Likewise.
28000 (BT_FN_INT8_INT8_INT8): Likewise.
28001 (BT_FN_INT16_INT16_INT16): Likewise.
28002 (BT_FN_INT_INT_INT): Likewise.
28003 (BT_FN_UINT_FLOAT_UINT): Likewise.
28004 (BT_FN_FLOAT_UINT_UINT): Likewise.
28005 (BT_FN_ULONG_UINT_UINT): Likewise.
28006 (BT_FN_ULONG_UINT_PTR): Likewise.
28007 (BT_FN_ULONG_ULONG_ULONG): Likewise.
28008 (BT_FN_UINT_UINT_UINT): Likewise.
28009 (BT_FN_VOID_UINT_PTR): Likewise.
28010 (BT_FN_UINT_UINT_PTR: Likewise.
28011 (BT_FN_UINT32_UINT64_PTR): Likewise.
28012 (BT_FN_INT_INT_UINT_UINT): Likewise.
28013 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
28014 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
28015 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
28016 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
28017 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
28018 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
28019 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
28020 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
28021 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
28022 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
28023 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
28024 * doc/frontends.texi: List BRIG FE.
28025 * doc/install.texi (Testing): Add BRIG tesring requirements.
28026 * doc/invoke.texi (Overall Options): Mention BRIG.
28027 * doc/standards.texi (Standards): Doucment BRIG HSA version.
28029 2017-01-24 Richard Biener <rguenther@suse.de>
28031 PR translation/79208
28032 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
28034 2017-01-24 Martin Jambor <mjambor@suse.cz>
28037 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
28038 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
28039 and known_contexts.
28041 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
28043 PR middle-end/79123
28044 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
28045 casts from signed to unsigned really don't have a range.
28047 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
28049 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
28050 GMP_RNDx for compatiblity.
28052 2017-01-24 Martin Liska <mliska@suse.cz>
28055 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
28056 that would prevent us to call alloca with -1 as argument.
28058 2017-01-24 Jakub Jelinek <jakub@redhat.com>
28060 * dwarf2out.c (output_compilation_unit_header, output_file_names):
28061 Avoid -Wformat-security warning.
28063 2017-01-23 Andrew Pinski <apinski@cavium.com>
28065 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
28068 2017-01-23 Martin Sebor <msebor@redhat.com>
28070 PR middle-end/78703
28071 * gimple-ssa-sprintf.c (warn_level): New global.
28072 (format_integer): Use it here and throughout the rest of the file.
28073 Use the same switch to compute sign as base.
28074 (maybe_warn): New function.
28075 (format_directive): Factor out warnings into maybe_warn.
28076 Add debugging output. Use warn_level.
28077 (add_bytes): Use warn_level.
28078 (pass_sprintf_length::compute_format_length): Add debugging output.
28079 (try_substitute_return_value): Same.
28080 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
28082 PR middle-end/78703
28083 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
28084 (struct fmtresult, format_integer, format_floating): Adjust.
28085 (fmtresult::fmtresult): Set max correctly in two argument ctor.
28086 (get_string_length, format_string,format_directive): Same.
28087 (pass_sprintf_length::compute_format_length): Same.
28088 (try_substitute_return_value): Simplify slightly.
28090 PR middle-end/78703
28091 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
28092 (fmtresult::operator+=): Outlined.
28093 (struct fmtresult): Add ctors.
28094 (struct conversion_spec): Rename...
28095 (struct directive): ...to this. Add and remove data members.
28096 (directive::set_width, directive::set_precision): New functions.
28097 (format_percent): Use fmtresult ctor.
28098 (get_width_and_precision): Remove.
28099 (format_integer): Make naming changes. Avoid computing width and
28101 (format_floating): Same. Adjust indentation.
28102 (format_character, format_none): New functions.
28103 (format_string): Moved character handling to format_character.
28104 (format_directive): Remove arguments, change return type.
28105 (parse_directive): New function.
28106 (pass_sprintf_length::compute_format_length): Move directive
28107 parsing to parse_directive.
28109 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28111 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
28112 (assign_assembler_name_if_needed): ... this.
28113 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
28114 (assign_assembler_name_if_needed): ... this.
28115 (free_lang_data_in_cgraph): Adjust callers.
28116 * cgraphunit.c (cgraph_node::analyze): Likewise.
28117 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
28120 2017-01-23 Richard Biener <rguenther@suse.de>
28122 PR tree-optimization/79088
28123 PR tree-optimization/79188
28124 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
28125 resetting loop bounds after last path deletion. Reset loop
28126 bounds of the target loop, make code match the comments.
28127 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
28128 Make sure loops need no fixups.
28130 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28132 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
28133 exponent support with double type for first argument.
28134 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
28135 type returned by __builtin_vec_extract_sig,
28136 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
28137 functions from "vector int" to "vector unsigned int" or from
28138 "vector long long int" to "vector unsigned long long int".
28139 Changed type returned by __builtin_vec_extract_exp,
28140 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
28141 functions from "vector int" to "vector unsigned int" or from
28142 "vector long long int" to "vector unsigned long long int".
28143 Changed return type of __builtin_vec_test_data_class,
28144 __builtin_vec_test_data_class_sp, and
28145 __builtin_vec_test_data_class_dp from "vector int" to
28146 "vector bool int" or from "vector long long int" to "vector bool
28147 long long int" and changed second argument type from "unsigned
28148 int" to "int". Added new overloaded function forms "vector float
28149 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
28150 "vector float __builtin_vec_insert_exp_sp (vector float, vector
28151 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
28152 double, vector unsigned long long int)" and "vector double
28153 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
28154 long int)". Changed return type of
28155 __builtin_scalar_test_data_class and
28156 __builtin_scalar_test_data_class_sp and
28157 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
28158 int" and changed second argument from "unsigned int" to "int".
28159 Changed type returned by __builtin_scalar_test_neg,
28160 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
28161 from "int" to "bool int". Added new overloaded function form
28162 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
28163 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
28164 exponent double-precision with floating point first argument.
28165 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
28166 documentation of scalar_test_data_class, scalar_test_neg,
28167 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
28168 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
28169 vec_test_data_class built-in functions to reflect refinements in
28170 their type signatures.
28172 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
28174 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
28176 (aarch64_elf_asm_destructor): Likewise.
28178 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
28180 PR rtl-optimization/78634
28181 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
28182 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
28183 * ifcvt.c (noce_try_cmove): Add missing cost check.
28185 PR rtl-optimization/71724
28186 * combine.c (if_then_else_cond): Look for situations where it is
28187 beneficial to undo the work of one of the recursive calls.
28189 2017-01-23 Bin Cheng <bin.cheng@arm.com>
28191 PR tree-optimization/70754
28192 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
28193 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
28194 combined stmt before it if not NULL.
28195 (combine_chains): Process refs reversely and compute dominance point
28198 2017-01-23 Martin Liska <mliska@suse.cz>
28200 PR tree-optimization/79196
28201 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
28202 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
28204 (strlen_optimize_stmt): Call the renamed function.
28206 2017-01-23 Michael Matz <matz@suse.de>
28208 PR tree-optimization/78384
28209 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
28211 2017-01-23 Richard Biener <rguenther@suse.de>
28213 PR tree-optimization/79186
28214 * tree-vrp.c (register_new_assert_for): Make sure we've seen
28215 both incoming edges before moving an assert.
28217 2017-01-23 Martin Jambor <mjambor@suse.cz>
28219 * ipa-prop.c (load_from_param_1): Removed.
28220 (load_from_unmodified_param): Bits from load_from_param_1 put back
28222 (load_from_param): Removed.
28223 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
28224 with stmt. Reverted back to use of load_from_unmodified_param.
28226 2017-01-23 Martin Jambor <mjambor@suse.cz>
28229 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
28230 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
28231 field a pointer to garbage collected vector, mark lattices and
28232 ipcp_orig_node with GTY((skip)).
28233 (ipa_get_param_count): Adjust to descriptors being a pointer.
28234 (ipa_get_param): Likewise.
28235 (ipa_get_type): Likewise.
28236 (ipa_get_param_move_cost): Likewise.
28237 (ipa_set_param_used): Likewise.
28238 (ipa_get_controlled_uses): Likewise.
28239 (ipa_set_controlled_uses): Likewise.
28240 (ipa_is_param_used): Likewise.
28241 (ipa_node_params_t): Move into garbage collector. New methods insert
28243 (ipa_node_params_sum): Annotate wth GTY(()).
28244 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
28246 (ipa_load_from_parm_agg): Adjust declaration.
28247 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
28248 * ipa-profile.c (ipa_profile): Likewise.
28249 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
28250 (ipa_populate_param_decls): Make descriptors parameter garbage
28252 (ipa_dump_param): Adjust to descriptors being a pointer.
28253 (ipa_alloc_node_params): Likewise.
28254 (ipa_initialize_node_params): Likewise.
28255 (load_from_param_1): Make descriptors parameter garbage collected.
28256 (load_from_unmodified_param): Likewise.
28257 (load_from_param): Likewise.
28258 (ipa_load_from_parm_agg): Likewise.
28259 (ipa_node_params::~ipa_node_params): Removed.
28260 (ipa_free_all_node_params): Remove call to delete operator.
28261 (ipa_node_params_t::insert): New.
28262 (ipa_node_params_t::remove): Likewise.
28263 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
28264 copy known_csts and known_contexts vectors.
28265 (ipa_read_node_info): Adjust to descriptors being a pointer.
28266 (ipcp_modif_dom_walker): Make m_descriptors field garbage
28268 (ipcp_transform_function): Make descriptors variable garbage
28271 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
28273 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
28274 * config/i386/avx512dqintrin.h: Ditto.
28275 * config/i386/avx512fintrin.h: Ditto.
28276 * config/i386/i386.c: Handle new builtins.
28277 * config/i386/i386-builtin.def: Add new builtins.
28278 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
28279 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
28281 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28282 Martin Liska <mliska@suse.cz>
28284 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
28285 * asan.c (asan_expand_poison_ifn): Support stores and use
28286 appropriate ASAN report function.
28287 * internal-fn.c (expand_ASAN_POISON_USE): New function.
28288 * internal-fn.def (ASAN_POISON_USE): Declare.
28289 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
28290 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
28291 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
28292 ASAN_POISON calls w/o LHS.
28293 * tree-ssa.c (execute_update_addresses_taken): Create clobber
28294 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
28295 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
28296 * gimplify.c (asan_poison_variables): Add attribute
28297 use_after_scope_memory to variables that really needs to live
28299 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
28300 having the attribute.
28302 2017-01-23 Martin Liska <mliska@suse.cz>
28304 * asan.c (create_asan_shadow_var): New function.
28305 (asan_expand_poison_ifn): Likewise.
28306 * asan.h (asan_expand_poison_ifn): New declaration.
28307 * internal-fn.c (expand_ASAN_POISON): Likewise.
28308 * internal-fn.def (ASAN_POISON): New builtin.
28309 * sanopt.c (pass_sanopt::execute): Expand
28310 asan_expand_poison_ifn.
28311 * tree-inline.c (copy_decl_for_dup_finish): Make function
28313 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
28314 * tree-ssa.c (is_asan_mark_p): New function.
28315 (execute_update_addresses_taken): Rewrite local variables
28316 (identified just by use-after-scope as addressable) into SSA.
28318 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28320 * doc/install.texi (Specific): opensource.apple.com uses https
28321 now. Remove trailing slash.
28323 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28325 * README.Portability: Remove note on an Irix compatibility issue.
28327 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
28329 * gcov.c (INCLUDE_ALGORITHM): Define.
28330 (INCLUDE_VECTOR): Define.
28331 No longer include <vector> and <algorithm> directly.
28333 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
28335 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
28337 * doc/invoke.texi (Code Gen Options): Ditto.
28339 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
28342 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
28344 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
28346 rtl-optimization/79125
28347 * cprop.c (local_cprop_pass): Handle cases where we make an
28348 unconditional trap.
28350 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
28354 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
28355 read from, for big endian.
28357 2017-01-20 Jiong Wang <jiong.wang@arm.com>
28359 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
28360 register pauth builtins for LP64 only.
28362 2017-01-20 Marek Polacek <polacek@redhat.com>
28365 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
28368 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28370 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
28372 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
28373 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
28374 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
28376 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28379 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
28380 in big-endian mode when they are not a single duplicated value.
28382 2017-01-20 Richard Biener <rguenther@suse.de>
28384 * BASE-VER: Bump to 7.0.1.
28386 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28388 * omp-low.c (omplow_simd_context): New struct. Use it...
28389 (lower_rec_simd_input_clauses): ...here and...
28390 (lower_rec_input_clauses): ...here to hold common data. Adjust all
28391 references to idx, lane, max_vf, is_simt.
28393 2017-01-20 Graham Markall <graham.markall@embecosm.com>
28395 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
28398 2017-01-20 Martin Jambor <mjambor@suse.cz>
28400 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
28401 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
28403 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
28404 (GTFILES): Rename hsa.c to hsa-common.c.
28405 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
28406 * hsa-dump.c: Likewise.
28407 * hsa-gen.c: Likewise.
28408 * hsa-regalloc.c: Likewise.
28409 * ipa-hsa.c: Likewise.
28410 * omp-expand.c: Likewise.
28411 * omp-low.c: Likewise.
28412 * toplev.c: Likewise.
28414 2017-01-20 Marek Polacek <polacek@redhat.com>
28417 * doc/invoke.texi: Document -Wduplicated-branches.
28418 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
28419 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
28420 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
28421 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
28422 return 0 only when not OEP_LEXICOGRAPHIC.
28423 (fold_build_cleanup_point_expr): Use the expression
28424 location when building CLEANUP_POINT_EXPR.
28425 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
28426 * tree.c (add_expr): Handle error_mark_node.
28428 2017-01-20 Martin Liska <mliska@suse.cz>
28431 * tree-profile.c (init_ic_make_global_vars): Do not call
28433 (gimple_init_gcov_profiler): Likewise.
28435 2017-01-20 Martin Liska <mliska@suse.cz>
28438 * cgraph.h (maybe_create_reference): Remove argument and
28440 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
28442 * ipa-cp.c (create_specialized_node): Likewise.
28443 * symtab.c (symtab_node::maybe_create_reference): Handle
28444 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
28446 2017-01-20 Martin Liska <mliska@suse.cz>
28448 * read-rtl-function.c (function_reader::create_function): Use
28449 build_decl instread of build_decl_stat.
28451 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
28453 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
28454 * config/i386/avx512dqintrin.h: Ditto.
28455 * config/i386/avx512fintrin.h: Ditto.
28456 * config/i386/i386-builtin-types.def: Add new types.
28457 * config/i386/i386.c: Handle new types.
28458 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
28459 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
28460 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
28461 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
28462 (__builtin_ia32_kshiftridi): New.
28463 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
28465 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
28469 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
28470 define to rs6000_init_stack_protect_guard.
28471 (rs6000_init_stack_protect_guard): New function.
28473 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28474 Yunqiang Su <yunqiang.su@imgtec.com>
28476 * config.gcc (supported_defaults): Add madd4.
28477 (with_madd4): Add validation.
28478 (all_defaults): Add madd4.
28479 * config/mips/mips.opt (mmadd4): New option.
28480 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28482 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28484 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
28485 (ISA_HAS_FUSED_MADD4): Likewise.
28486 * doc/invoke.texi (-mmadd4): Document the new option.
28487 * doc/install.texi (--with-madd4): Document the new option.
28489 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28491 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
28492 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
28493 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
28494 (aarch64_init_pauth_hint_builtins): New.
28495 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
28496 (aarch64_expand_builtin): Expand new builtins.
28498 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28500 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
28501 * combine-stack-adj.c (no_unhandled_cfa): Handle
28502 REG_CFA_TOGGLE_RA_MANGLE.
28503 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
28504 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
28505 info for return address signing.
28506 (aarch64_expand_epilogue): Likewise.
28508 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28510 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
28511 * config/aarch64/aarch64-protos.h
28512 (aarch64_return_address_signing_enabled): New declaration.
28513 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
28515 (aarch64_expand_prologue): Sign return address before it's pushed onto
28517 (aarch64_expand_epilogue): Authenticate return address fetched from
28519 (aarch64_override_options): Sanity check for ILP32 and ISA level.
28520 (aarch64_attributes): New function attributes for "sign-return-address".
28521 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
28522 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
28523 ("*do_return"): Generate combined instructions according to key index.
28524 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
28525 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
28527 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
28528 * config/aarch64/aarch64.opt (msign-return-address=): New.
28529 * doc/extend.texi (AArch64 Function Attributes): Documents
28530 "sign-return-address=".
28531 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
28533 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28535 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
28536 overall option summary.
28538 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28540 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
28541 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
28542 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
28543 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
28545 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
28547 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
28548 -mpower9-minmax by default for -mcpu=power9.
28549 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
28550 128-bit floating point.
28552 2017-01-20 Alan Modra <amodra@gmail.com>
28554 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
28555 optimizing for size.
28557 2017-01-20 Alan Modra <amodra@gmail.com>
28560 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
28561 for strcmp and strncmp from corresponding builtin decl.
28563 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28565 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
28566 instead of i386/rtems-64.h.
28567 * config/i386/rtems-64.h: Remove.
28569 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28573 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
28575 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
28577 2017-01-19 Tamar Christina <tamar.christina@arm.com>
28579 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
28580 Change int to HOST_WIDE_INT.
28581 * config/aarch64/aarch64-protos.h
28582 (aarch64_simd_gen_const_vector_dup): Likewise.
28583 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
28585 2017-01-19 David Malcolm <dmalcolm@redhat.com>
28587 * langhooks-def.h (lhd_type_for_size): New decl.
28588 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
28589 * langhooks.c (lhd_type_for_size): New function, taken from
28592 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
28594 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
28595 define_bypass for CR latency.
28596 (power9-cracked-alu): Update bypass latency and remove power9-branch.
28597 (power9-alu2): Add define_bypass for CR latency.
28599 (power9-mul): Update insn latency.
28600 (power9-mul-compare): Update insn latency, bypass latency and remove
28603 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28605 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
28607 * config/aarch64/aarch64.md
28608 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
28609 aarch64_nopcrelative_literal_loads.
28610 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
28612 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
28614 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
28615 TARGET_LOONGSON_3A.
28616 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
28618 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
28621 * config.gcc (supported_defaults): Add lxc1-sxc1.
28622 (with_lxc1_sxc1): Add validation.
28623 (all_defaults): Add lxc1-sxc1.
28624 * config/mips/mips.opt (mlxc1-sxc1): New option.
28625 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28627 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28628 __mips_no_lxc1_sxc1.
28629 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
28630 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
28631 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
28633 2017-01-19 Richard Biener <rguenther@suse.de>
28635 PR tree-optimization/72488
28636 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
28637 sure to restore SSA info.
28638 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
28640 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
28642 PR rtl-optimization/79121
28643 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
28644 of the inner type when shifting an extended value.
28646 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28649 * symtab.c (symtab_node::equal_address_to): Fix comparing of
28650 interposable aliases.
28652 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
28655 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
28656 Use the evmergelohi instruction.
28657 (mov_si<mode>_e500_subreg4_2_le): Likewise.
28658 (mov_sitf_e500_subreg8_2_be): Likewise.
28659 (mov_sitf_e500_subreg12_2_le): Likewise.
28660 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
28661 (mov_si<mode>_e500_subreg4_2_be): Likewise.
28662 (mov_sitf_e500_subreg8_2_le): Likewise.
28663 (mov_sitf_e500_subreg12_2_be): Likewise.
28665 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28667 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
28668 attribute from vecsimple to vecperm.
28669 (altivec_vbpermq2): Likewise.
28671 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28674 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
28676 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28677 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
28678 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
28679 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
28680 case where N arg is SIZE_MAX.
28681 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
28682 (cmpstrsi): Add pattern.
28684 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28686 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28687 __builtin_vec_revb builtins.
28688 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
28689 built-in functions to support generation of the ISA 3.0 XXBR<x>
28690 vector byte reverse instructions.
28691 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
28692 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
28693 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
28694 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
28695 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
28696 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
28697 (P9V_BUILTIN_VEC_REVB): Likewise.
28698 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
28699 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
28700 (p9_xxbrq_v16qi): Likewise.
28701 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
28702 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
28703 (p9_xxbrh_v8hi): Likewise.
28704 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
28705 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
28706 vec_revb built-in functions.
28708 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
28710 PR rtl-optimization/78952
28711 * config/i386/i386.md (any_extract): New code iterator.
28712 (*insvqi_2): Use any_extract for source operand.
28713 (*insvqi_3): Use any_shiftrt for source operand.
28715 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
28717 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
28719 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
28721 2017-01-18 Matthias Klose <doko@ubuntu.com>
28723 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
28725 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28727 * config/rs6000/altivec.h (vec_bperm): Change #define.
28728 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
28729 (altivec_vbpermq2): New define_insn.
28730 (altivec_vbpermd): Likewise.
28731 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
28732 function interface.
28733 (VBPERMD): Likewise.
28734 (VBPERM): New polymorphic function interface.
28735 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
28736 Add entries for P9V_BUILTIN_VEC_VBPERM.
28737 * doc/extend.texi: Add interfaces for vec_bperm.
28739 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28741 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
28742 first letter of error messages.
28743 (s390_resolve_overloaded_builtin): Likewise.
28744 * config/s390/s390.c (s390_expand_builtin): Likewise.
28745 (s390_invalid_arg_for_unprototyped_fn): Likewise.
28746 (s390_valid_target_attribute_inner_p): Likewise.
28747 * config/s390/s390.md ("tabort"): Likewise.
28749 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
28751 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
28752 (ISA_AVOID_DIV_HILO): New macro.
28753 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
28754 (ISA_HAS_DDIV): Likewise.
28756 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28758 * doc/invoke.texi (fabi-version): Correct number of occurrences.
28760 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28762 * doc/invoke.texi (fabi-version): Spelling fix.
28764 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28767 * doc/invoke.texi (fabi-version): Mention mangling fix for
28770 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28773 * doc/invoke.texi (fabi-version): Document discriminator mangling.
28775 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
28778 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
28779 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
28781 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
28783 (stack_protect_test): Ditto.
28784 * config/rs6000/rs6000.opt (mstack-protector-guard=,
28785 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
28787 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
28788 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
28789 -mstack-protector-guard-offset=.
28790 (RS/6000 and PowerPC Options): Ditto.
28792 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28794 * config/i386/i386.h (MASK_CLASS_P): New define.
28795 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
28796 there are no registers from different register sets also when
28797 mask registers are used. Update function comment.
28798 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
28799 to (*k/*r) and (*k/*km) alternatives.
28801 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
28803 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
28804 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
28805 (EH_RETURN_HANDLER_RTX): New define.
28806 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
28807 Force frame pointer in EH return functions.
28808 (aarch64_expand_epilogue): Add barrier for eh_return.
28809 (aarch64_final_eh_return_addr): Remove.
28810 (aarch64_eh_return_handler_rtx): New function.
28811 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
28813 (aarch64_eh_return_handler_rtx): New prototype.
28815 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28817 * config/rs6000/altivec.h (vec_rlmi): New #define.
28818 (vec_vrlnm): Likewise.
28819 (vec_rlnm): Likewise.
28820 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
28821 (UNSPEC_VRLNM): Likewise.
28822 (VIlong): New mode iterator.
28823 (altivec_vrl<VI_char>mi): New define_insn.
28824 (altivec_vrl<VI_char>nm): Likewise.
28825 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
28827 (VRLDNM): Likewise.
28828 (RLNM): New polymorphic function entry.
28829 (VRLWMI): New monomorphic function entry.
28830 (VRLDMI): Likewise.
28831 (RLMI): New polymorphic function entry.
28832 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
28833 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
28834 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
28837 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28840 * dwarf2out.c (field_byte_offset): Restore the
28841 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
28842 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
28843 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
28846 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28849 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
28851 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28854 * dwarf2out.c (add_data_member_location_attribute): For constant
28855 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
28856 instead of DW_AT_data_member_location, DW_AT_bit_offset and
28857 DW_AT_byte_size attributes.
28859 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28861 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
28862 after forcing to constant memory when the code model is medium.
28864 2017-01-17 Julia Koval <julia.koval@intel.com>
28867 * config/i386/avx512fintrin.h
28868 (_mm512_i32gather_ps): Change __addr type to void const*.
28869 (_mm512_mask_i32gather_ps): Ditto.
28870 (_mm512_i32gather_pd): Ditto.
28871 (_mm512_mask_i32gather_pd): Ditto.
28872 (_mm512_i64gather_ps): Ditto.
28873 (_mm512_mask_i64gather_ps): Ditto.
28874 (_mm512_i64gather_pd): Ditto.
28875 (_mm512_mask_i64gather_pd): Ditto.
28876 (_mm512_i32gather_epi32): Ditto.
28877 (_mm512_mask_i32gather_epi32): Ditto.
28878 (_mm512_i32gather_epi64): Ditto.
28879 (_mm512_mask_i32gather_epi64): Ditto.
28880 (_mm512_i64gather_epi32): Ditto.
28881 (_mm512_mask_i64gather_epi32): Ditto.
28882 (_mm512_i64gather_epi64): Ditto.
28883 (_mm512_mask_i64gather_epi64): Ditto.
28884 (_mm512_i32scatter_ps): Change __addr type to void*.
28885 (_mm512_mask_i32scatter_ps): Ditto.
28886 (_mm512_i32scatter_pd): Ditto.
28887 (_mm512_mask_i32scatter_pd): Ditto.
28888 (_mm512_i64scatter_ps): Ditto.
28889 (_mm512_mask_i64scatter_ps): Ditto.
28890 (_mm512_i64scatter_pd): Ditto.
28891 (_mm512_mask_i64scatter_pd): Ditto.
28892 (_mm512_i32scatter_epi32): Ditto.
28893 (_mm512_mask_i32scatter_epi32): Ditto.
28894 (_mm512_i32scatter_epi64): Ditto.
28895 (_mm512_mask_i32scatter_epi64): Ditto.
28896 (_mm512_i64scatter_epi32): Ditto.
28897 (_mm512_mask_i64scatter_epi32): Ditto.
28898 (_mm512_i64scatter_epi64): Ditto.
28899 (_mm512_mask_i64scatter_epi64): Ditto.
28900 * config/i386/avx512pfintrin.h
28901 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
28902 (_mm512_mask_prefetch_i32gather_ps): Ditto.
28903 (_mm512_mask_prefetch_i64gather_pd): Ditto.
28904 (_mm512_mask_prefetch_i64gather_ps): Ditto.
28905 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
28906 (_mm512_prefetch_i32scatter_ps): Ditto.
28907 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
28908 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
28909 (_mm512_prefetch_i64scatter_pd): Ditto.
28910 (_mm512_prefetch_i64scatter_ps): Ditto.
28911 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
28912 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
28913 * config/i386/avx512vlintrin.h
28914 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
28915 (_mm_mmask_i32gather_ps): Ditto.
28916 (_mm256_mmask_i32gather_pd): Ditto.
28917 (_mm_mmask_i32gather_pd): Ditto.
28918 (_mm256_mmask_i64gather_ps): Ditto.
28919 (_mm_mmask_i64gather_ps): Ditto.
28920 (_mm256_mmask_i64gather_pd): Ditto.
28921 (_mm_mmask_i64gather_pd): Ditto.
28922 (_mm256_mmask_i32gather_epi32): Ditto.
28923 (_mm_mmask_i32gather_epi32): Ditto.
28924 (_mm256_mmask_i32gather_epi64): Ditto.
28925 (_mm_mmask_i32gather_epi64): Ditto.
28926 (_mm256_mmask_i64gather_epi32): Ditto.
28927 (_mm_mmask_i64gather_epi32): Ditto.
28928 (_mm256_mmask_i64gather_epi64): Ditto.
28929 (_mm_mmask_i64gather_epi64): Ditto.
28930 (_mm256_i32scatter_ps): Change __addr type to void*.
28931 (_mm256_mask_i32scatter_ps): Ditto.
28932 (_mm_i32scatter_ps): Ditto.
28933 (_mm_mask_i32scatter_ps): Ditto.
28934 (_mm256_i32scatter_pd): Ditto.
28935 (_mm256_mask_i32scatter_pd): Ditto.
28936 (_mm_i32scatter_pd): Ditto.
28937 (_mm_mask_i32scatter_pd): Ditto.
28938 (_mm256_i64scatter_ps): Ditto.
28939 (_mm256_mask_i64scatter_ps): Ditto.
28940 (_mm_i64scatter_ps): Ditto.
28941 (_mm_mask_i64scatter_ps): Ditto.
28942 (_mm256_i64scatter_pd): Ditto.
28943 (_mm256_mask_i64scatter_pd): Ditto.
28944 (_mm_i64scatter_pd): Ditto.
28945 (_mm_mask_i64scatter_pd): Ditto.
28946 (_mm256_i32scatter_epi32): Ditto.
28947 (_mm256_mask_i32scatter_epi32): Ditto.
28948 (_mm_i32scatter_epi32): Ditto.
28949 (_mm_mask_i32scatter_epi32): Ditto.
28950 (_mm256_i32scatter_epi64): Ditto.
28951 (_mm256_mask_i32scatter_epi64): Ditto.
28952 (_mm_i32scatter_epi64): Ditto.
28953 (_mm_mask_i32scatter_epi64): Ditto.
28954 (_mm256_i64scatter_epi32): Ditto.
28955 (_mm256_mask_i64scatter_epi32): Ditto.
28956 (_mm_i64scatter_epi32): Ditto.
28957 (_mm_mask_i64scatter_epi32): Ditto.
28958 (_mm256_i64scatter_epi64): Ditto.
28959 (_mm256_mask_i64scatter_epi64): Ditto.
28960 (_mm_i64scatter_epi64): Ditto.
28961 (_mm_mask_i64scatter_epi64): Ditto.
28962 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
28963 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
28964 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
28965 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
28966 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
28967 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
28968 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
28969 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
28970 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
28971 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
28972 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
28973 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
28974 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
28975 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
28976 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
28977 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
28978 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
28979 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
28980 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
28981 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
28982 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
28983 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
28984 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
28985 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
28986 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
28987 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
28988 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
28989 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
28990 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
28991 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
28992 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
28993 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
28994 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
28995 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
28996 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
28997 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
28998 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
28999 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
29000 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
29001 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
29002 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
29003 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
29004 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
29005 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
29006 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
29007 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
29008 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
29009 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
29010 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
29011 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
29012 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
29013 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
29014 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
29015 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
29016 definitions accordingly.
29018 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
29019 Kuan-Lin Chen <kuanlinchentw@gmail.com>
29022 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
29025 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
29028 * ira-conflicts.c (ira_build_conflicts): Update total conflict
29029 hard regs for inner regno.
29031 2017-01-17 Martin Liska <mliska@suse.cz>
29034 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
29035 assumption and add comment.
29037 2017-01-17 Nathan Sidwell <nathan@acm.org>
29039 * ipa-visibility.c (localize_node): New function, broken out of ...
29040 (function_and_variable_visibility): ... here. Call it.
29042 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29044 PR middle-end/77445
29045 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
29046 correctly set frequency of oudgoing edge.
29047 (duplicate_thread_path): Fix profile updating.
29049 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29052 * configure.ac: Add GCC_BASE_VER.
29053 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
29054 version from BASE-VER file.
29055 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
29056 (gcc.o): Depend on $(BASEVER).
29057 * common.opt (dumpfullversion): New option.
29058 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
29059 * doc/invoke.texi: Document -dumpfullversion.
29060 * doc/install.texi: Document --with-gcc-major-version-only.
29061 * configure: Regenerated.
29063 2017-01-17 Richard Biener <rguenther@suse.de>
29065 PR tree-optimization/71433
29066 * tree-vrp.c (register_new_assert_for): Merge same asserts
29067 on all incoming edges.
29068 (process_assert_insertions_for): Handle insertions at the
29071 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
29073 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
29074 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
29076 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
29079 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
29082 2017-01-17 Alan Modra <amodra@gmail.com>
29085 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
29086 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
29087 symbolic stack limit when pic.
29089 2017-01-16 Martin Sebor <msebor@redhat.com>
29091 PR tree-optimization/78608
29092 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
29094 2017-01-16 Jeff Law <law@redhat.com>
29097 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
29098 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
29099 for several include directories that may be relative to sysroot.
29100 * config/i386/x-mingw32 (gplus_includedir): Define.
29101 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
29102 (native_system_includedir): Likewise.
29103 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
29104 override if TARGET_SYSTEM_ROOT is defined.
29105 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
29107 PR tree-optimization/79090
29108 PR tree-optimization/33562
29109 PR tree-optimization/61912
29110 PR tree-optimization/77485
29111 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
29112 and computed trims into the dump file.
29114 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29116 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
29118 2017-01-16 Jakub Jelinek <jakub@redhat.com>
29121 * gimplify.c (gimplify_init_constructor): If want_value and
29122 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
29126 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
29127 sequence. Formatting fixes.
29128 (doloop_optimize): Formatting fixes.
29131 * gcc.c (debug_level_greater_than_spec_func): New function.
29132 (static_spec_functions): Add debug-level-gt spec function.
29133 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
29135 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29136 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29137 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
29138 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
29139 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
29140 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
29142 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
29144 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
29145 QImode fixups to general and mask registers only.
29147 2017-01-16 Carl Love <cel@us.ibm.com>
29149 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
29150 for built-in functions
29151 vector signed char vec_nabs (vector signed char)
29152 vector signed short vec_nabs (vector signed short)
29153 vector signed int vec_nabs (vector signed int)
29154 vector signed long long vec_nabs (vector signed long long)
29155 vector float vec_nabs (vector float)
29156 vector double vec_nabs (vector double)
29157 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
29159 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
29160 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
29161 * doc/extend.texi: Update the documentation file for the new built-in
29164 2017-01-16 Martin Sebor <msebor@redhat.com>
29166 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
29169 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29171 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
29172 UNSPEC_VSX__XXSPLTD to require special splat handling.
29174 2017-01-16 David Malcolm <dmalcolm@redhat.com>
29177 * system.h: Poison strndup.
29179 2017-01-16 Alan Modra <amodra@gmail.com>
29182 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
29185 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
29187 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
29189 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
29191 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
29192 call recog here. Assert that INSN_CODE (insn) is non-negative.
29194 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29197 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
29199 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
29200 in the currently scheduled RTL fragment.
29202 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29204 PR rtl-optimization/78751
29205 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
29208 2017-01-14 Jeff Law <law@redhat.com>
29210 PR tree-optimization/79090
29211 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
29212 variable length stores.
29213 (compute_trims): Delete dead assignment to *trim_tail.
29214 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
29217 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
29219 PR rtl-optimization/78626
29220 PR rtl-optimization/78727
29221 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
29222 of a block, and split such blocks after everything else is finished.
29224 2017-01-14 Alan Modra <amodra@gmail.com>
29227 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
29228 target legitimate_combined_insn.
29229 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
29230 (rs6000_legitimate_combined_insn): New function.
29231 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
29233 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
29234 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
29235 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
29237 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
29239 * doc/frontends.texi (G++ and GCC): Remove references to Java.
29241 2017-01-13 Jeff Law <law@redhat.com>
29243 PR tree-optimization/33562
29244 PR tree-optimization/61912
29245 PR tree-optimization/77485
29246 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
29248 (delete_dead_assignment): Likewise.
29249 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
29250 statement to delete_dead_call and delete_dead_assignment.
29252 2017-01-13 David Malcolm <dmalcolm@redhat.com>
29255 * substring-locations.c (format_warning_va): Strengthen case 1 so
29256 that both endpoints of the substring must be within the format
29257 range for just the substring to be printed.
29259 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
29261 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
29262 * config/i386/i386.c (ix86_target_string): Add missing options
29263 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
29264 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
29265 flags_other and ix86_target_other to flags2_other. Display unknown
29267 (ix86_valid_target_attribute_inner_p): Add missing options and
29268 reorder options by implied ISAs, as in ix86_target_string.
29270 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29272 * hash-table.h (hash_table::too_empty_p): New function.
29273 (hash_table::expand): Use it.
29274 (hash_table::traverse): Likewise.
29275 (hash_table::empty_slot): Use sizeof (value_type) instead of
29276 sizeof (PTR) to convert bytes to elements. Shrink the table
29277 if the current size is excessive for the current number of
29280 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29282 * ira-costs.c (record_reg_classes): Break from the inner loop
29283 early once alt_fail is known to be true. Update outer loop
29284 handling accordingly.
29286 2017-01-13 Jeff Law <law@redhat.com>
29288 * tree-ssa-dse.c (decrement_count): New function.
29289 (increment_start_addr, maybe_trim_memstar_call): Likewise.
29290 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
29291 when we know the partially dead statement is a mem* function.
29293 PR tree-optimization/61912
29294 PR tree-optimization/77485
29295 * tree-ssa-dse.c: Include expr.h.
29296 (maybe_trim_constructor_store): New function.
29297 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
29299 PR tree-optimization/33562
29300 PR tree-optimization/61912
29301 PR tree-optimization/77485
29302 * doc/invoke.texi: Document new dse-max-object-size param.
29303 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
29304 * tree-ssa-dse.c: Include params.h.
29305 (dse_store_status): New enum.
29306 (initialize_ao_ref_for_dse): New, partially extracted from
29308 (valid_ao_ref_for_dse, normalize_ref): New.
29309 (setup_live_bytes_from_ref, compute_trims): Likewise.
29310 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
29311 (maybe_trim_partially_dead_store): Likewise.
29312 (maybe_trim_complex_store): Likewise.
29313 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
29314 Track what bytes live from the original store. Return tri-state
29315 for dead, partially dead or live.
29316 (dse_dom_walker): Add constructor, destructor and new private members.
29317 (delete_dead_call, delete_dead_assignment): New extracted from
29319 (dse_optimize_stmt): Make a member of dse_dom_walker.
29320 Use initialize_ao_ref_for_dse.
29322 PR tree-optimization/33562
29323 PR tree-optimization/61912
29324 PR tree-optimization/77485
29325 * sbitmap.h (bitmap_count_bits): Prototype.
29326 (bitmap_clear_range, bitmap_set_range): Likewise.
29327 * sbitmap.c (bitmap_clear_range): New function.
29328 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
29330 2017-01-13 Martin Liska <mliska@suse.cz>
29333 * function.c (set_cfun): Add new argument force.
29334 * function.h (set_cfun): Likewise.
29335 * ipa-inline-transform.c (inline_call): Use the function when
29336 strict alising from is dropped for function we inline to.
29338 2017-01-13 Richard Biener <rguenther@suse.de>
29340 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
29341 for dumping GIMPLE INTEGER_CSTs.
29343 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29345 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
29346 to 201112L since C++17.
29348 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
29351 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
29352 if -fsanitize=kernel-address is present.
29354 2017-01-13 Richard Biener <rguenther@suse.de>
29356 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
29357 as _Literal ( type ) number in case usual suffixes do not
29358 preserve all information.
29360 2017-01-13 Richard Biener <rguenther@suse.de>
29362 PR tree-optimization/77283
29363 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
29364 and ssa-iterators.h.
29365 (is_feasible_trace): Implement a cost model based on joiner
29368 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
29371 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
29372 char or short to __float128/_Float128 directly.
29374 2017-01-12 Martin Sebor <msebor@redhat.com>
29376 to -Wformat-overflow.
29377 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
29378 (min_bytes_remaining): Same.
29379 (get_string_length): Same.
29380 (format_string): Same.
29381 (format_directive): Same.
29383 (pass_sprintf_length::handle_gimple_call): Same.
29385 2017-01-12 Jakub Jelinek <jakub@redhat.com>
29387 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
29388 info.nowrite calls with no lhs that can't throw. Return bool
29389 whether gsi_remove has been called or not.
29390 (pass_sprintf_length::handle_gimple_call): Return bool whether
29391 try_substitute_return_value called gsi_remove. Formatting fix.
29392 (pass_sprintf_length::execute): Don't use gsi_remove if
29393 handle_gimple_call returned true.
29396 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
29397 be removed due to side-effects, don't remove following barrier nor
29398 turn the successor edge into fallthru edge.
29400 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29403 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
29404 element-reversing loads and stores as not swappable.
29406 2017-01-12 Nathan Sidwell <nathan@acm.org>
29407 Nicolai Stange <nicstange@gmail.com>
29409 * combine.c (try_combine): Don't ignore result of overlap checking
29410 loop. Combine overlap & asm check into single loop.
29412 2017-01-12 Richard Biener <rguenther@suse.de>
29414 * tree-pretty-print.c (dump_generic_node): Provide -gimple
29415 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
29417 2017-01-12 Richard Biener <rguenther@suse.de>
29419 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
29420 and TS_TARGET_OPTION directly derive from TS_BASE.
29421 * tree-core.h (tree_optimization_option): Derive from tree_base.
29422 (tree_target_option): Likewise.
29424 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
29426 * config/i386/i386.c (memory_address_length): Increase len
29427 only when rip_relative_addr_p returns false.
29429 2017-01-11 Julia Koval <julia.koval@intel.com>
29431 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
29432 (OPTION_MASK_ISA_SGX_SET): New.
29433 (ix86_handle_option): Handle OPT_msgx.
29434 * config.gcc: Added sgxintrin.h.
29435 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
29436 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
29437 * config/i386/i386.c (ix86_target_string): Add -msgx.
29439 (ix86_option_override_internal): Handle new options.
29440 (ix86_valid_target_attribute_inner_p): Add sgx.
29441 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
29442 * config/i386/i386.opt: Add msgx.
29443 * config/i386/sgxintrin.h: New file.
29444 * config/i386/x86intrin.h: Add sgxintrin.h.
29446 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29449 * fold-const.c (maybe_nonzero_address): Return 1 for function
29451 (tree_single_nonzero_warnv_p): Don't handle function local objects
29455 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
29458 2017-01-11 David Malcolm <dmalcolm@redhat.com>
29461 * opts.c: Include "spellcheck.h"
29462 (struct string_fragment): New struct.
29463 (struct edit_distance_traits<const string_fragment &>): New
29465 (get_closest_sanitizer_option): New function.
29466 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
29468 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29470 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
29472 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
29473 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
29474 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
29475 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
29476 for initial die_offset if dwarf_split_debug_info.
29477 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
29478 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
29480 (output_skeleton_debug_sections): Formatting fix. Use
29481 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
29482 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
29484 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
29486 * config/arm/cortex-a53.md: Add bypasses for
29487 cortex_a53_r2f_cvt.
29488 (cortex_a53_r2f): Only use for transfers.
29489 (cortex_a53_f2r): Likewise.
29490 (cortex_a53_r2f_cvt): Add reservation for conversions.
29491 (cortex_a53_f2r_cvt): Likewise.
29493 2017-01-11 Tamar Christina <tamar.christina@arm.com>
29495 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
29496 to all inlined functions, change static to extern.
29498 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
29501 * config/arm/arm.c (legitimize_pic_address): Handle reference to
29503 (arm_assemble_integer): Likewise.
29505 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
29507 * config.gcc: Use new awk script to check CPU, FPU and architecture
29508 parameters for --with-... options.
29509 * config/arm/parsecpu.awk: New file
29510 * config/arm/arm-cpus.in: New file.
29511 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
29513 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
29515 * config/arm/t-arm: Update dependency rules.
29516 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
29517 of processing .def files.
29518 * config/arm/genopt.sh: Deleted.
29519 * config/arm/gentune.sh: Deleted.
29520 * config/arm/arm-cores.def: Deleted.
29521 * config/arm/arm-arches.def: Deleted.
29522 * config/arm/arm-fpus.def: Deleted.
29523 * config/arm/arm-tune.md: Regenerated.
29524 * config/arm/arm-tables.opt: Regenerated.
29525 * config/arm/arm-cpu.h: New generated file.
29526 * config/arm/arm-cpu-data.h: New generated file.
29527 * config/arm/arm-cpu-cdata.h: New generated file.
29529 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
29532 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
29534 (input_varpool_node): Unpack dynamically_initialized bit.
29536 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
29538 PR rtl-optimization/79032
29539 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
29540 the alignment of the adjusted memory reference against that of MODE,
29541 instead of the alignment of the original memory reference.
29543 2017-01-11 Martin Jambor <mjambor@suse.cz>
29545 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
29547 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
29548 decorated functions.
29550 2017-01-11 Richard Biener <rguenther@suse.de>
29552 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
29553 set range/nonnull info for PHI results. Do not set it on
29554 stmts marked for removal.
29556 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
29558 * expr.c (store_field): In the bitfield case, fetch the return value
29559 from the registers before applying a single big-endian adjustment.
29560 Always do a final load for a BLKmode value not larger than a word.
29562 2017-01-10 David Malcolm <dmalcolm@redhat.com>
29565 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29566 that we correctly handle column numbers greater than
29567 LINE_MAP_MAX_COLUMN_NUMBER.
29569 2017-01-10 Martin Sebor <msebor@redhat.com>
29571 PR middle-end/78245
29572 * gimple-ssa-sprintf.c (get_destination_size): Call
29573 {init,fini}object_sizes.
29574 * tree-object-size.c (addr_object_size): Adjust.
29575 (pass_through_call): Adjust.
29576 (pass_object_sizes::execute): Adjust.
29577 * tree-object-size.h (fini_object_sizes): Declare.
29579 2017-01-10 Martin Sebor <msebor@redhat.com>
29581 PR tree-optimization/78775
29582 * builtins.c (get_size_range): Move...
29583 * calls.c: ...to here.
29584 (alloc_max_size): Accept zero argument.
29585 (operand_signed_p): Remove.
29586 (maybe_warn_alloc_args_overflow): Call get_size_range.
29587 * calls.h (get_size_range): Declare.
29589 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
29591 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
29592 from TI's devices.csv file as of September 2016.
29593 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
29595 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
29597 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
29598 * doc/invoke.texi: Likewise.
29599 * doc/md.texi: Likewise.
29600 * doc/objc.texi: Likewise.
29602 2017-01-10 Joshua Conner <joshconner@google.com>
29604 * config/arm/fuchsia-elf.h: New file.
29605 * config/fuchsia.h: New file.
29606 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
29607 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
29609 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
29611 2016-01-10 Richard Biener <rguenther@suse.de>
29613 PR tree-optimization/79034
29614 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29615 Propagate out degenerate PHIs in the joiner.
29617 2017-01-10 Martin Liska <mliska@suse.cz>
29619 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
29620 (sort_congruence_classes_by_decl_uid): Likewise.
29621 (sort_congruence_class_groups_by_decl_uid): Likewise.
29622 (sem_item_optimizer::merge_classes): Sort class, groups in these
29623 classes and members in the groups by DECL_UID of declarations.
29624 This would make merge operations stable.
29626 2017-01-10 Martin Liska <mliska@suse.cz>
29628 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
29629 usage of m_classes_vec.
29630 (sem_item_optimizer::~sem_item_optimizer): Likewise.
29631 (sem_item_optimizer::get_group_by_hash): Likewise.
29632 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
29633 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
29634 (sem_item_optimizer::verify_classes): Likewise.
29635 (sem_item_optimizer::process_cong_reduction): Likewise.
29636 (sem_item_optimizer::dump_cong_classes): Likewise.
29637 (sem_item_optimizer::merge_classes): Likewise.
29638 * ipa-icf.h (congruence_class_hash): Rename from
29639 congruence_class_group_hash. Remove declaration of m_classes_vec.
29641 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
29643 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
29644 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
29645 * config.gcc: Add avx512vpopcntdqintrin.h.
29646 * config/i386/avx512vpopcntdqintrin.h: New.
29647 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
29648 * config/i386/i386-builtin-types.def: Add new types.
29649 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
29650 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
29651 __builtin_ia32_vpopcountq_v8di_mask): New.
29652 * config/i386/i386-c.c (ix86_target_macros_internal): Define
29653 __AVX512VPOPCNTDQ__.
29654 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
29655 (PTA_AVX512VPOPCNTDQ): Define.
29656 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
29657 TARGET_AVX512VPOPCNTDQ_P): Define.
29658 * config/i386/i386.opt: Add mavx512vpopcntdq.
29659 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
29660 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
29662 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29664 PR middle-end/77484
29665 * predict.def (PRED_CALL): Set to 67.
29667 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
29669 * expr.c (store_field): In the bitfield case, if the value comes from
29670 a function call and is of an aggregate type returned in registers, do
29671 not modify the field mode; extract the value in all cases if the mode
29672 is BLKmode and the size is not larger than a word.
29674 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
29677 * config/i386/cpuid.h: Fix undefined behavior.
29679 2017-01-04 Jeff Law <law@redhat.com>
29681 PR tree-optimization/79007
29682 PR tree-optimization/67955
29683 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
29684 conservative for pt.null when flag_non_call_exceptions is on.
29686 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29688 PR translation/79019
29689 PR translation/79020
29690 * params.def (PARAM_INLINE_MIN_SPEEDUP,
29691 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
29692 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
29694 * config/avr/avr.opt (maccumulate-args): Likewise.
29695 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
29696 * common.opt (freport-bug): Likewise.
29697 * cif-code.def (CIF_FINAL_ERROR): Likewise.
29698 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
29699 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
29700 translatable string.
29701 * config/i386/i386.c (function_value_32): Likewise.
29702 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
29703 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
29705 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
29706 * common/config/msp430/msp430-common.c (msp430_handle_option):
29708 * symtab.c (symtab_node::verify_base): Likewise.
29709 * opts.c (set_debug_level): Likewise.
29710 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
29711 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
29712 missing whitespace to translatable strings.
29713 * config/avr/avr.md (bswapsi2): Fix typo in comment.
29714 * config/sh/superh.h: Likewise.
29715 * config/i386/xopintrin.h: Likewise.
29716 * config/i386/znver1.md: Likewise.
29717 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
29718 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
29719 * double-int.h (struct double_int): Likewise.
29720 * double-int.c (div_and_round_double): Likewise.
29721 * wide-int.cc: Likewise.
29722 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
29723 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
29724 * cfgcleanup.c (crossjumps_occured): Renamed to ...
29725 (crossjumps_occurred): ... this.
29726 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
29729 PR tree-optimization/78899
29730 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
29731 returning bool return struct loop *, NULL for failure and the new
29733 (versionable_outer_loop_p): Don't version outer loop if it has
29734 dont_vectorized bit set.
29735 (tree_if_conversion): When versioning outer loop, ensure
29736 tree_if_conversion is performed also on the inner loop of the
29737 non-vectorizable outer loop copy.
29738 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
29739 LOOP_VECTORIZED in inner loop of the scalar outer loop and
29740 prevent vectorization of it.
29741 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
29742 the outer loop vectorization of the non-scalar version is attempted
29743 before vectorization of the inner loop in scalar version. If
29744 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
29745 vectorization of its inner loop.
29746 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
29747 has 2 inner loops, rename also on edges from bb whose single pred
29748 is outer_loop->header. Fix typo in function comment.
29750 2017-01-09 Martin Sebor <msebor@redhat.com>
29753 * asan.c (asan_emit_stack_protection): Increase local buffer size
29754 to avoid snprintf truncation warning.
29756 2017-01-09 Andrew Pinski <apinski@cavium.com>
29758 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
29759 to reference thunderx2t99 for the tuning structure
29760 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
29762 (thunderx2t99_extra_costs): This.
29763 * config/aarch64/aarch64-tune.md: Regenerate.
29764 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
29765 (vulcan_addrcost_table): This.
29766 (vulcan_regmove_cost): Rename to ...
29767 (thunderx2t99_regmove_cost): This.
29768 (vulcan_vector_cost): Rename to ...
29769 (thunderx2t99_vector_cost): this.
29770 (vulcan_branch_cost): Rename to ...
29771 (thunderx2t99_branch_cost): This.
29772 (vulcan_tunings): Rename to ...
29773 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
29774 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
29776 2017-01-09 Martin Jambor <mjambor@suse.cz>
29780 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
29781 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
29782 (propagate_vr_accross_jump_function): Use the above function for all
29783 value range computations for pass-through jump functions and type
29784 converasion from explicit value range values.
29785 (ipcp_propagate_stage): Do not attempt to deduce types of formal
29786 parameters from TYPE_ARG_TYPES.
29787 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
29788 (ipa_write_node_info): Stream type of the actual argument.
29789 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
29791 2017-01-09 Martin Liska <mliska@suse.cz>
29794 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
29795 (lookup_compiler): Do not show error message with have_E.
29797 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29799 PR tree-optimization/78938
29800 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
29801 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
29802 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
29803 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
29806 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29808 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
29811 2017-01-09 Richard Biener <rguenther@suse.de>
29813 PR tree-optimization/78997
29814 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
29815 name condition properly.
29817 2017-01-09 Richard Biener <rguenther@suse.de>
29820 * dwarf2out.c (is_cxx): New overload with context.
29821 (is_naming_typedef_decl): Use it.
29823 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29825 * invoke.texi (Option Summary): Correct spacing in option lists
29826 and add line breaks to fix over-long lines.
29828 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29830 PR middle-end/17660
29832 * extend.texi (Common Variable Attributes): Add xref to GCC
29833 Internals manual to explain mode attribute keywords.
29835 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29838 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
29839 and Preprocessor Options.
29840 (Options for Linking): Document -pthread here....
29841 (RS/6000 and PowerPC Options): ...not here.
29842 (Solaris 2 Options): ...or here.
29843 * doc/cppopts.texi: Document -pthread.
29845 2017-01-08 Martin Sebor <msebor@redhat.com>
29847 PR middle-end/77708
29848 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
29849 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
29850 New member functions.
29851 (format_directive): Used them.
29853 (pass_sprintf_length::handle_gimple_call): Same.
29854 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
29855 to avoid truncation for any argument.
29856 (extract_affine_mul): Same.
29857 * tree.c (get_file_function_name): Same.
29859 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29861 PR middle-end/77484
29862 * predict.def (PRED_INDIR_CALL): Set to 86.
29864 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29866 PR preprocessor/54124
29867 * doc/cppopts.texi: Reformat -d subtable to list the full name
29868 of the options. Add cross-reference to the docs for the general
29869 compiler -d options.
29870 * doc/invoke.texi (Developer Options): Add cross-reference to the
29871 preprocessor-specific -d option documentation.
29873 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29875 PR preprocessor/13498
29876 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
29877 redudant material, and reflect new command-line options.
29878 (System Headers): Likewise.
29880 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29882 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
29883 -isystem, and -idirafter. Copy-edit.
29884 * doc/cppopts.texi: Copy-edit. Remove contradiction about
29885 default for -ftrack-macro-expansion. Delete obsolete and
29886 badly-formatted implementation details about -fdebug-cpp output.
29887 * doc/cppwarnopts.texi: Copy-edit.
29889 2017-01-07 David Malcolm <dmalcolm@redhat.com>
29892 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29893 that the transition from a max line width >= 1<<10 to narrower
29894 lines works correctly.
29896 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
29898 * doc/options.texi (PerFunction): New.
29899 * opt-functions.awk (switch_flags): Map both Optimization and
29900 PerFunction to CL_OPTIMIZATION.
29901 * opth-gen.awk: Test for PerFunction flag along with
29903 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
29904 it only when the latter is present. Skip those that don't in
29905 the hash function generator.
29906 * common.opt (fvar-tracking): Mark as PerFunction instead of
29908 (fvar-tracking-assignments): Likewise.
29909 (fvar-tracking-assignments-toggle): Likewise.
29910 (fvar-tracking-uninit): Likewise.
29912 2017-01-07 Jakub Jelinek <jakub@redhat.com>
29914 PR translation/79018
29915 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
29918 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
29921 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
29922 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
29923 TARGET_LONG_JUMP_TABLE_OFFSETS.
29924 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
29925 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
29926 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
29927 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
29928 * config/m68k/m68k.md (tablejump expander): Likewise.
29929 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
29930 TARGET_LONG_JUMP_TABLE_OFFSETS.
29931 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
29932 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
29934 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
29935 David Holsgrove <david.holsgrove@xilinx.com>
29937 * common/config/microblaze/microblaze-common.c
29938 (TARGET_EXCEPT_UNWIND_INFO): Remove.
29939 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
29941 * config/microblaze/microblaze.c (microblaze_must_save_register)
29942 (microblaze_expand_epilogue, microblaze_return_addr): Handle
29944 (microblaze_eh_return): New function.
29945 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
29946 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
29947 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
29948 * config/microblaze/microblaze.md (eh_return): New pattern.
29950 2017-01-06 Jakub Jelinek <jakub@redhat.com>
29952 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
29953 GCC_DIAGNOSTIC_STRINGIFY): Define.
29955 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
29957 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29959 * config/arm/arm.md (<mcrr>): New.
29961 * config/arm/arm.c (arm_arch5te): New.
29962 (arm_option_override): Set arm_arch5te.
29963 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
29965 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
29966 (arm_mcrr_qualifiers): ... this. New.
29967 (MRRC_QUALIFIERS): Define to...
29968 (arm_mrrc_qualifiers): ... this. New.
29969 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
29971 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
29972 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
29973 (MRRCI, mrrc, MRRC): New.
29974 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
29975 VUNSPEC_MRRC2): New.
29977 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29979 * config/arm/arm.md (<mcr>): New.
29981 * config/arm/arm.c (arm_coproc_builtin_available): Add
29982 support for mcr, mrc, mcr2 and mrc2.
29983 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
29984 (arm_mcr_qualifiers): ... this. New.
29985 (MRC_QUALIFIERS): Define to ...
29986 (arm_mrc_qualifiers): ... this. New.
29987 (MCR_QUALIFIERS): Define to ...
29988 (arm_mcr_qualifiers): ... this. New.
29989 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
29991 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
29992 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
29993 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
29994 VUNSPEC_MRC2): New.
29996 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29998 * config/arm/arm.md (*ldc): New.
30002 * config/arm/arm.c (arm_coproc_builtin_available): Add
30003 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
30004 (arm_coproc_ldc_stc_legitimate_address): New.
30005 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
30006 'qualifier_const_pointer'.
30007 (LDC_QUALIFIERS): Define to...
30008 (arm_ldc_qualifiers): ... this. New.
30009 (STC_QUALIFIERS): Define to...
30010 (arm_stc_qualifiers): ... this. New.
30011 * config/arm/arm-protos.h
30012 (arm_coproc_ldc_stc_legitimate_address): New.
30013 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
30014 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
30015 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
30016 stc2, stcl, stc2l): New.
30017 * config/arm/constraints.md (Uz): New.
30018 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
30019 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
30020 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
30021 VUNSPEC_STC2L): New.
30023 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30025 * config/arm/arm.md (<cdp>): New.
30026 * config/arm/arm.c (neon_const_bounds): Rename this ...
30027 (arm_const_bounds): ... this.
30028 (arm_coproc_builtin_available): New.
30029 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
30030 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
30031 (CDP_QUALIFIERS): Define to...
30032 (arm_cdp_qualifiers): ... this. New.
30034 (arm_expand_builtin_args): Add case for 6 arguments.
30035 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
30036 (arm_const_bounds): ... this.
30037 (arm_coproc_builtin_available): New.
30038 * config/arm/arm_acle.h (__arm_cdp): New.
30040 * config/arm/arm_acle_builtins.def (cdp): New.
30042 * config/arm/iterators.md (CDPI,CDP,cdp): New.
30043 * config/arm/neon.md: Rename all 'neon_const_bounds' to
30044 'arm_const_bounds'.
30045 * config/arm/types.md (coproc): New.
30046 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
30047 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
30048 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
30049 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
30051 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30053 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
30054 (UBINOP_QUALIFIERS): New.
30056 (acle_builtin_data): New. Change comment.
30057 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
30058 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
30059 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
30060 arm_acle_builtins.def.
30061 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
30062 (arm_init_acle_builtins): New.
30063 (CRC32_BUILTIN): Remove.
30064 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
30065 crc32cb, crc32ch and crc32cw.
30066 (arm_init_crc32_builtins): Remove.
30067 (arm_init_builtins): Use arm_init_acle_builtins rather
30068 than arm_init_crc32_builtins.
30069 (arm_expand_acle_builtin): New.
30070 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
30071 * config/arm/arm_acle_builtins.def: New.
30073 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30075 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
30076 (arm_builtin_datum): ... this.
30077 (arm_init_neon_builtin): Rename to ...
30078 (arm_init_builtin): ... this. Add a new parameters PREFIX
30079 and USE_SIG_IN_NAME.
30080 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
30081 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
30082 'arm_builtin_datum'.
30083 (arm_init_vfp_builtins): Likewise.
30084 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
30085 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
30086 (arm_expand_neon_args): Rename to ...
30087 (arm_expand_builtin_args): ... this. Rename builtin_arg
30088 enum values and differentiate between ARG_BUILTIN_MEMORY
30089 and ARG_BUILTIN_NEON_MEMORY.
30090 (arm_expand_neon_builtin_1): Rename to ...
30091 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
30092 values, arm_expand_builtin_args and add bool parameter NEON.
30093 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
30094 (arm_expand_vfp_builtin): Likewise.
30095 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
30097 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30099 PR middle-end/77484
30100 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
30101 * predict.c (tree_estimate_probability_bb): Reverse direction of
30102 polymorphic call predictor.
30104 2017-01-06 David Malcolm <dmalcolm@redhat.com>
30106 * passes.c (execute_one_pass): Split out pass-skipping logic into...
30107 (determine_pass_name_match): ...this new function and...
30108 (should_skip_pass_p): ...this new function.
30110 2017-01-06 Nathan Sidwell <nathan@acm.org>
30112 * ipa-visibility.c (function_and_variable_visibility): Reformat
30113 comments and long lines. Remove extrneous if.
30114 * symtab.c (symtab_node::make_decl_local): Fix code format.
30115 (symtab_node::set_section_for_node): Fix comment typo.
30117 2017-01-06 Martin Liska <mliska@suse.cz>
30120 * lra-constraints.c: Rename invariant to lra_invariant.
30121 * predict.c (set_even_probabilities): Initialize e to NULL.
30123 2017-01-05 Martin Sebor <msebor@redhat.com>
30125 PR tree-optimization/78910
30126 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
30127 (format_integer): Correct off-by-one error in the handling
30128 of precision with negative numbers in signed conversions..
30130 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
30132 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
30134 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30136 PR tree-optimization/71016
30137 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
30138 factor_out_conditional_conversion. Formatting fix.
30139 (factor_out_conditional_conversion): Add cond_stmt argument.
30140 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
30141 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
30144 2017-01-05 David Malcolm <dmalcolm@redhat.com>
30146 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
30147 read-rtl-function.o, and selftest-rtl.o.
30148 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
30149 (selftest::aarch64_test_loading_full_dump): New function.
30150 (selftest::aarch64_run_selftests): New function.
30151 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
30152 selftest::aarch64_run_selftests.
30153 * config/i386/i386.c
30154 (selftest::ix86_test_loading_dump_fragment_1): New function.
30155 (selftest::ix86_test_loading_call_insn): New function.
30156 (selftest::ix86_test_loading_full_dump): New function.
30157 (selftest::ix86_test_loading_unspec): New function.
30158 (selftest::ix86_run_selftests): Call the new functions.
30159 * emit-rtl.c (maybe_set_max_label_num): New function.
30160 * emit-rtl.h (maybe_set_max_label_num): New decl.
30161 * function.c (instantiate_decls): Guard call to
30162 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
30163 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
30165 * gensupport.c (gen_reader::gen_reader): Pass "false"
30166 for new "compact" param of rtx_reader.
30167 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
30168 rather than an empty string for NULL strings.
30169 * read-md.c: Potentially include config.h rather than bconfig.h.
30170 Wrap include of errors.h with #ifdef GENERATOR_FILE.
30171 (have_error): New global, copied from errors.c.
30172 (md_reader::read_name): Rename to...
30173 (md_reader::read_name_1): ...this, adding "out_loc" param,
30174 and converting "missing name or number" to returning false, rather
30176 (md_reader::read_name): Reimplement in terms of read_name_1.
30177 (md_reader::read_name_or_nil): New function.
30178 (md_reader::read_string): Handle "(nil)" by returning NULL.
30179 (md_reader::md_reader): Add new param "compact".
30180 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
30181 (md_reader::read_file): New method.
30182 * read-md.h (md_reader::md_reader): Add new param "compact".
30183 (md_reader::read_file): New method.
30184 (md_reader::is_compact): New accessor.
30185 (md_reader::read_name): Convert return type from void to file_location.
30186 (md_reader::read_name_or_nil): New decl.
30187 (md_reader::read_name_1): New decl.
30188 (md_reader::m_compact): New field.
30189 (noop_reader::noop_reader): Pass "false" for new "compact" param
30191 (rtx_reader::rtx_reader): Add new "compact" param.
30192 (rtx_reader::read_rtx_operand): Make virtual and convert return
30193 type from void to rtx.
30194 (rtx_reader::read_until): New decl.
30195 (rtx_reader::handle_any_trailing_information): New virtual function.
30196 (rtx_reader::postprocess): New virtual function.
30197 (rtx_reader::finalize_string): New virtual function.
30198 (rtx_reader::m_in_call_function_usage): New field.
30199 (rtx_reader::m_reuse_rtx_by_id): New field.
30200 * read-rtl-function.c: New file.
30201 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
30202 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
30203 (selftest::verify_three_block_rtl_cfg): New decl.
30204 * read-rtl-function.h: New file.
30205 * read-rtl.c: Potentially include config.h rather than bconfig.h.
30206 For host, include function.h, memmodel.h, and emit-rtl.h.
30207 (one_time_initialization): New function.
30208 (struct compact_insn_name): New struct.
30209 (compact_insn_names): New array.
30210 (find_code): Handle insn codes in compact dumps.
30211 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
30212 (bind_subst_iter_and_attr): Likewise.
30213 (add_condition_to_string): Likewise.
30214 (add_condition_to_rtx): Likewise.
30215 (apply_attribute_uses): Likewise.
30216 (add_current_iterators): Likewise.
30217 (apply_iterators): Likewise.
30218 (initialize_iterators): Guard usage of apply_subst_iterator with
30219 #ifdef GENERATOR_FILE.
30220 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
30221 (md_reader::read_mapping): Likewise.
30222 (add_define_attr_for_define_subst): Likewise.
30223 (add_define_subst_attr): Likewise.
30224 (read_subst_mapping): Likewise.
30225 (check_code_iterator): Likewise.
30226 (rtx_reader::read_rtx): Likewise. Move one-time initialization
30228 (one_time_initialization): New function.
30229 (rtx_reader::read_until): New method.
30230 (read_flags): New function.
30231 (parse_reg_note_name): New function.
30232 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
30233 Handle reuse_rtx ids.
30234 Wrap iterator lookup within #ifdef GENERATOR_FILE.
30235 Add parsing support for RTL dumps, mirroring the special-cases in
30236 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
30237 values, and calling handle_any_trailing_information.
30238 (rtx_reader::read_rtx_operand): Convert return type from void
30239 to rtx, returning return_rtx. Handle case 'e'. Call
30240 finalize_string on XSTR and XTMPL fields.
30241 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
30242 "(nil)" values were omitted. Call the postprocess vfunc on the
30244 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
30245 class ctor. Initialize m_in_call_function_usage. Call
30246 one_time_initialization.
30247 * rtl-tests.c (selftest::test_uncond_jump): Call
30248 set_new_first_and_last_insn.
30249 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
30250 * selftest-rtl.c: New file.
30251 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
30252 (selftest::get_insn_by_uid): New decl.
30253 * selftest-run-tests.c (selftest::run_tests): Call
30254 read_rtl_function_c_tests.
30255 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
30256 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
30259 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
30261 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
30262 operands in a special way. Assert that pos+len <= mode precision.
30264 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30266 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
30267 3 argument Alias with unlimited for the negative form.
30268 (fno-vect-cost-model): Removed.
30270 2017-01-05 Martin Liska <mliska@suse.cz>
30272 * hsa-gen.c (gen_hsa_divmod): New function.
30273 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
30275 2017-01-05 Martin Liska <mliska@suse.cz>
30278 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
30281 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30283 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
30284 small constant length operands.
30286 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30288 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
30289 between loop iterations.
30291 2017-01-05 Martin Liska <mliska@suse.cz>
30294 * gimplify.c (gimplify_decl_expr): Compare to
30295 asan_poisoned_variables instread of checking flags.
30296 (gimplify_target_expr): Likewise.
30297 (gimplify_expr): Likewise.
30298 (gimplify_function_tree): Conditionally initialize
30299 asan_poisoned_variables.
30301 2017-01-04 Jeff Law <law@redhat.com>
30303 PR tree-optimizatin/78812
30304 * rtl.h (contains_mem_rtx_p): Prototype.
30305 * ifcvt.c (containts_mem_rtx_p): Move from here to...
30306 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
30307 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
30308 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
30309 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
30311 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30313 * input.c (assert_char_at_range): Default-initialize actual_range.
30315 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30317 * df-scan.c (df_ref_create_structure): Make regno unsigned,
30318 to match the caller.
30320 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30322 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
30323 insns after final jump in test to emit dummy move.
30325 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30327 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
30328 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
30330 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30332 * multiple_target.c (create_dispatcher_calls): Init e_next.
30333 * tree-ssa-loop-split.c (split_loop): Init border.
30334 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
30337 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
30342 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
30343 (altivec_register_operand): Do not return true if the operand
30344 contains a SUBREG mixing SImode and SFmode.
30345 (vsx_register_operand): Likewise.
30346 (vsx_reg_sfsubreg_ok): New predicate.
30347 (vfloat_operand): Do not return true if the operand contains a
30348 SUBREG mixing SImode and SFmode.
30349 (vint_operand): Likewise.
30350 (vlogical_operand): Likewise.
30351 (gpc_reg_operand): Likewise.
30352 (int_reg_operand): Likewise.
30353 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
30354 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
30355 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
30357 (rs6000_emit_move_si_sf_subreg): New helper function.
30358 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
30359 fixup SUBREGs involving SImode and SFmode.
30360 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
30361 numbers for the new peephole2 optimization.
30362 (peephole2 for SFmode unions): New peephole2 to optimize cases in
30363 the GLIBC math library that do AND/IOR/XOR operations on single
30364 precision floating point.
30365 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
30366 target macros to say whether we need to avoid SUBREGs mixing
30368 (TARGET_ALLOW_SF_SUBREG): Likewise.
30369 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
30370 (UNSPEC_SI_FROM_SF): Likewise.
30371 (iorxor): Change spacing.
30372 (and_ior_xor): New iterator for AND, IOR, and XOR.
30373 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
30374 (movdi_from_sf_zero_ext): Likewise.
30375 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
30376 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
30377 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
30378 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
30379 (fms<mode>4): Likewise.
30380 (fnma<mode>4): Likewise.
30381 (fnms<mode>4): Likewise.
30382 (nfma<mode>4): Likewise.
30383 (nfms<mode>4): Likewise.
30385 2017-01-04 Marek Polacek <polacek@redhat.com>
30388 * doc/invoke.texi: Document -Wpointer-compare.
30390 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30392 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
30395 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
30396 descriptions for -gdwarf-5 and emit them as uleb128 instead of
30399 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
30402 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
30403 documentation of the powerpc_popcntb_ok attribute.
30404 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30405 code to issue warning messages if a requested CPU configuration is
30406 not supported by the binary (assembler and loader) toolchain.
30407 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
30408 made to define a built-in function that has been disabled.
30409 (paired_init_builtins): Add assertion to prevent ICE if attempt is
30410 made to define a built-in function that has been disabled.
30411 (altivec_init_builtins): Add comment explaining why definition
30412 of the DST built-in functions is not preceded by an assertion
30413 check. Add assertions to prevent ICE if attempts are made to
30414 define an altivec predicate or an abs* built-in function that has
30416 (htm_init_builtins): Add comment explaining why definition of the
30417 htm built-in functions is not preceded by an assertion check.
30419 2017-01-04 Jeff Law <law@redhat.com>
30421 PR tree-optimizatin/67955
30422 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
30423 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
30424 the points-to solution does not include pt_null. Use DECL_PT_UID
30427 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
30429 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
30430 Use gen_int_mode instead of gen_lopwart for const_int operands.
30432 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30434 PR tree-optimization/71563
30435 * match.pd: Simplify X << Y into X if Y is known to be 0 or
30436 out of range value - has low bits known to be zero.
30438 2017-01-04 Alan Modra <amodra@gmail.com>
30440 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
30441 * configure: Regenerate.
30442 * config.in: Regenerate.
30444 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30447 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
30448 a substring of the message, but strcmp with the whole message. Ifdef
30449 ENABLE_NLS, translate the message first using dgettext.
30451 2017-01-03 Jeff Law <law@redhat.com>
30453 PR tree-optimizatin/78856
30454 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
30455 (mark_threaded_blocks): Remove code to truncate thread paths that
30456 cross multiple loop headers. Instead invalidate the cached loop
30457 iteration information and handle case of a thread path walking
30458 into an irreducible region.
30460 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
30463 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
30464 assertions. Add support for doing the signbit if the IEEE 128-bit
30465 floating point value is in a GPR.
30466 * config/rs6000/rs6000.md (Fsignbit): Delete.
30467 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
30468 Update the length attribute if the value is in a GPR.
30469 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
30470 the sign or zero extension instruction, since the value is always 0/1.
30471 (signbit<mode>2_dm2): Delete using <Fsignbit>.
30474 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
30475 extracting SImode to a GPR register so that we can generate a
30476 store, limit the vector to be in a traditional Altivec register
30477 for the vextuwrx instruction.
30479 2017-01-03 Ian Lance Taylor <iant@google.com>
30481 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
30483 2017-01-03 Martin Sebor <msebor@redhat.com>
30485 PR tree-optimization/78696
30486 * gimple-ssa-sprintf.c (format_floating): Correct handling of
30487 precision. Use MPFR for %f for greater fidelity. Correct handling
30489 (pass_sprintf_length::compute_format_length): Set width and precision
30490 specified by asrerisk to void_node for vararg functions.
30491 (try_substitute_return_value): Adjust dump output.
30493 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
30495 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
30497 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
30499 * doc/invoke.texi (SPARC options): Document -mlra as the default.
30500 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
30501 -mlra/-mno-lra was passed to the compiler.
30503 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
30505 PR rtl-optimization/65618
30506 * emit-rtl.c (try_split): Move initialization of "before" and
30507 "after" to just before the call to emit_insn_after_setloc.
30509 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
30511 * doc/md.texi (Standard Names): Remove reference to Java frontend.
30513 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
30515 * dwarf2out.c (gen_enumeration_type_die): When
30516 -gno-strict-dwarf, add a DW_AT_encoding attribute.
30518 2017-01-03 Jakub Jelinek <jakub@redhat.com>
30520 PR tree-optimization/78965
30521 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
30522 Change first argument from const call_info & to call_info &. For %n
30523 set info.nowrite to false.
30525 PR middle-end/78901
30526 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
30527 possibly throwing calls.
30529 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
30530 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
30531 and fns handling, rather than in a separate case SSA_NAME.
30533 2017-01-02 Jeff Law <law@redhat.com>
30535 * config/darwin-driver.c (darwin_driver_init): Const-correctness
30536 fixes for first_period and second_period variables.
30538 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
30541 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
30542 (*insvqi_1): New insn pattern.
30543 (*insvqi_1_mem_rex64): Ditto.
30544 (*insvqi_2): Ditto.
30545 (*insvqi_3): Rename from *insvqi.
30547 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
30549 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
30551 * doc/cfg.texi (Edges): Remove reference to Java.
30552 (Maintaining the CFG): Ditto.
30554 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30556 PR middle-end/77674
30557 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
30558 transparent aliases.
30560 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30562 PR middle-end/77484
30563 * predict.def (PRED_CALL): Update hitrate.
30564 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
30565 * predict.c (tree_estimate_probability_bb): Split CALL predictor
30566 into direct/indirect/polymorphic variants.
30568 2017-01-01 Jakub Jelinek <jakub@redhat.com>
30570 Update copyright years.
30572 * gcc.c (process_command): Update copyright notice dates.
30573 * gcov-dump.c (print_version): Ditto.
30574 * gcov.c (print_version): Ditto.
30575 * gcov-tool.c (print_version): Ditto.
30576 * gengtype.c (create_file): Ditto.
30577 * doc/cpp.texi: Bump @copying's copyright year.
30578 * doc/cppinternals.texi: Ditto.
30579 * doc/gcc.texi: Ditto.
30580 * doc/gccint.texi: Ditto.
30581 * doc/gcov.texi: Ditto.
30582 * doc/install.texi: Ditto.
30583 * doc/invoke.texi: Ditto.
30585 Copyright (C) 2017 Free Software Foundation, Inc.
30587 Copying and distribution of this file, with or without modification,
30588 are permitted in any medium without royalty provided the copyright
30589 notice and this notice are preserved.